2021
Euriell Le Corronc, Yannick Pencolé, Alexandre Sahuguède, Claire Paya
Failure detection and localization for timed event graphs in (max,+)-algebra
Journal of Discrete Event Dynamic Systems: Theory and Applications 26 2021
In this paper, we address the problem of failure detection and
localization in a Timed Discrete Event System (TDES) such (max, +)-linear
system graphically modeled by a Timed Event Graph (TEG). The considered
failures are changes on holding times or tokens of the TEG places that can
provoke shifts between an observed outcoming timed flow and an expected
outcoming timed flow (for a given incoming timed flow). Indicators are built
to first detect such shifts relying on the (max, +) algebraic framework and
the residuation theory. An analysis of the indicators’ values provides
information about time or event failure that could have happen. Then, thanks to the
knowledge of the behavior of the system through its corresponding TEG, sets
of failures that could explain the detected shifts are obtained. It comes from
matrices of signatures for each indicator built on each observable output of
the system. An example of application is proposed to experiment exhaustively
failures of type time and event on each place of the TEG.
Yannick Pencolé, Audine Subias
Diagnosability of Event Patterns in Safe Labeled Time Petri Nets: A Model-Checking Approach
IEEE Transactions on Automation Science and Engineering January 2021
Checking the diagnosability of a time discrete event
system usually consists in determining whether a single fault
event can always be identified with certainty after a finite amount
of time. The aim of this paper is to extend this type of analysis to
more complex behaviors, called event patterns, and to propose an
effective method to check diagnosability with the use of model-
checking techniques. To do so, we propose to convert the pattern
diagnosability problem into checking a linear-time property over
a specific time Petri net.
Yannick Pencolé, Audine Subias, Camille Coquand
A Model Checking method to solve the event pattern diagnosis problem in safe labeled time Petri nets
32nd International Workshop on Principle of Diagnosis Hamburg Germany September 2021
This paper addresses the problem of identifying whether a given pattern of
unobservable events has occurred in a partially observable timed discrete event system.
The systems that are considered here are modelled with safe labeled time Petri nets.
To solve the problem, we propose to use a model checking approach. Given the model of
the system, the pattern to analyse, and the sequence of observations produced by the
system, the method firstly builds automatically a labeled time Petri net and a pair
of queries that represent the pattern diagnosis problem to solve. The second step then
consists in using a model-checker for time Petri net that provides the answer to the
queries. The solution provided by the model-checker is then analysed to determine whether
the pattern under investigation has definitely occurred or not.
Claire Paya, Euriell Le Corronc, Yannick Pencolé, Philippe Vialletelle
Observer-Based Detection and Localization of Time Shift Failures in (max, +)-Linear Systems
17th IEEE International Conference on Automation Science and Engineering Lyon France August 2021
This paper addresses a problem of failure de-
tection and localization in production lines modeled, through
Timed Event Graphs (TEG), as (max,+) linear systems with
disturbances, over which observers can be developed. The state
of the observed system is estimated and an indicator that
returns true if a time shift failure is detected is defined. The
localization step is proposed for elementary structures of TEG
through the results of the detection.
Claire Paya, Euriell Le Corronc, Yannick Pencolé, Philippe Vialletelle
Model-Based Diagnosis of Time Shift Failures in Discrete Event Systems: A (Max,+) Observer-Based Approach
Intelligent Decision Technologies, Proceedings of the 13th KES-IDT 2021 Conference 545-555 virtual conference Italy June 2021
This paper addresses the problem of diagnosing the occurrence
of time shift failures in systems like automated production lines.
The model of the system is represented as a Timed Event Graph (TEG)
that is characterized as a (max,+)-linear system. The proposed method
aims at detecting and localizing the source of time shift failures by the
design of a set of indicators. These indicators rely on the residuation
theory on (max,+)-linear systems and a (max,+) observer that estimates
the internal state of the observed system.
2020
Claire Paya, Euriell Le Corronc, Yannick Pencolé, Philippe Vialletelle
Observer-based detection of time shift failures in (max,+)-linear systems
31st International Workshop on Principles of Diagnosis Virtual conference United States September 2020
In this paper, we address the problem of failure
detection in production lines modeled as Timed Event Graphs (TEG).
The proposed method represents TEGs as (max,+)-linear systems with
disturbance and aims at detecting time shift failures
in the underlying production lines. To do so, we
will reconstruct the state of the observed system
and define an indicator relying on the residuation
theory on (max,+)-linear systems.
Yannick Pencolé, Audine Subias
Formal Chronicle Analyses and Comparisons: How to Deal with Negative Behaviors
Journal of Universal Computer Science 26(2) 268-292 2020
The overall context of this paper is the event-based behavior analysis and focuses on modeling and analyzing behaviors of interest involving time information. Any behavior of interest from any time event system is concisely defined as a set of time constrained events that must occur (positive behavior) and a set of time constrained events that must not occur (negative behavior). This article proposes a formal extension of the chronicle formalism that allows for the concise description of positive and negative behaviors. Based on this new formalism, several criteria are introduced, they formally characterize and compare a set of chronicles. A fully proved implementation of the proposed criteria is then described; it relies on the use of polyhedron techniques to solve systems of linear inequalities.
Marie-Odile Cordier, Philippe Dague, Yannick Pencolé, Louise Travé-Massuyès
Diagnosis and Supervision: Model-Based Approaches
A Guided Tour of Artificial Intelligence Research. Volume I: Knowledge Representation, Reasoning and Learning 673-706 2020
This chapter is devoted to diagnosis and supervision. It is organized as follows: after a section dedicated to the logical formalization of model-based diagnosis, the focus is made on diagnosis of discrete event systems modeled by automata. In the last section, one presents more succinctly the works that allowed to make the bridge between the approaches proposed by the Artificial Intelligence community and those proposed by the Automatic Control community.
Johanne Bakalara, Yannick Pencolé, Audine Subias
How to use Model Checking for diagnosing fault patterns in Petri nets
15th IFAC Workshop on Discrete Event Systems Rio de Janeiro Brazil 2020
This paper deals with the problem of fault pattern diagnosis in discrete event systems
modelled by Petri nets. A general framework for implementing a pattern diagnosis function by
model-checking is proposed. Two different approaches with experimental results are presented
and compared to a third approach from the literature.
Éric Lubat, Silvano Dal Zilio, Didier Le Botlan, Yannick Pencolé, Audine Subias
A New Product Construction for the Diagnosability of Patterns in Time Petri Net
59th IEEE Conference on Decision and Control Jeju Island Korea, Republic Of 2020
We propose a method to decide the diagnosability
of patterns in labeled Time Petri nets (TPN) that gracefully
extends a classic approach for the diagnosability of single faults.
Our approach is based on a new technique for computing the
language intersection of TPN and on an associated extension
of the State Class Graph construction. Our approach has been
implemented and we report on some experimental results.
2019
Elodie Chanthery, Louise Travé-Massuyès, Yannick Pencolé, Régis de Ferluc, Brice Dellandrea
Applying Active Diagnosis to Space Systems by On-Board Control Procedures
IEEE Transactions on Aerospace and Electronic Systems 55(5) 2568-2580 2019
The instrumentation of real systems is often designed for control purposes and control inputs are designed to achieve nominal control objectives. Hence, the available measurements may not be sufficient to isolate faults with certainty and diagnoses are ambiguous. Active diagnosis formulates a planning problem to generate a sequence of actions that, applied to the system, enforces diagnosability and allows to iteratively refine ambiguous diagnoses. This paper analyzesthe requirements for applying active diagnosis to space systems and proposes ActHyDiag as an effective framework to solve this problem. It presents the results of applying ActHyDiag to a real-space case study and of implementing the generated plans in the form of on-board control procedures. The case study is a redundant SpaceWire network where up to six instruments, monitored and controlled by the on-board software hosted in the satellite management unit, are transferring science data to a mass memory unit through SpaceWire routers. Experiments have been conducted on a real physical benchmark developed by Thales Alenia Space and demonstrate the effectiveness of the plans proposed by ActHyDiag.
Claire Paya, Euriell Le Corronc, Yannick Pencolé
Detection of time shift failures in (max, +)-linear systems with time intervals for assembly line
IEEE International Conference on Systems Man, and Cybernetics 1801-1806 Bari Italy October 2019
In this paper, we address the problem of failure detection in assembly lines modeled as Timed Event Graphs (TEG). The proposed method represents TEGs as (max,+)-linear systems with time intervals and aims at detecting time shift failures in the underlying assembly lines. To do so, we propose the definition of a set of indicators relying on the residuation theory on (max,+) linear systems that handle certain and uncertain observable outputs.
Éric Lubat, Silvano Dal Zilio, Didier Le Botlan, Yannick Pencolé, Audine Subias
A State Class Construction for Computing the Intersection of Time Petri Nets Languages
International Conference on Formal Modeling and Analysis of Timed Systems 79-95 Amsterdam Netherlands August 2019
We propose a new method for computing the language intersection of two Time Petri nets (TPN); that is the sequence of labels in timed traces common to the execution of two TPN. Our approach is based on a new product construction between nets and relies on the State Class construction, a widely used method for checking the behaviour of TPN. We prove that this new construct does not add additional expressive power, and yet that it can leads to very concise representation of the result. We have implemented our approach in a new tool, called Twina. We report on some experimental results obtained with this tool and show how to apply our approach on two interesting problems: first, to define an equivalent of the twin-plant diagnosability methods for TPN; then as a way to check timed properties without interfering with a system.
Gregor Gössler, Thomas Mari, Yannick Pencolé, Louise Travé-Massuyès
Towards Causal Explanations of Property Violations in Discrete Event Systems
30th International Workshop on Principles of Diagnosis Klagenfurt Austria November 2019
Model-Based Diagnosis of discrete event systems (DES) usually aims at detecting failures and isolating faulty event occurrences based on a behavioural model of the system and an observable execution log. The strength of a diagnostic process is to determine what happened that is consistent with the observations. In order to go a step further and explain why the observed outcome occurred, we borrow techniques from causal analysis. As opposed to the classical fault diagnosis problem, we consider that a system is failing as soon as a specific behavioural property is violated by the current run of the system. We then formally define different notions of explanation for DES in order to extract the relevant part of a property violation that can be understood by a human operator.
2018
Yannick Pencolé, Audine Subias
Timed pattern diagnosis in timed workflows: a model checking approach.
14th Workshop on Discrete Event Systems 94-99 Sorrento Coast Italy May 2018
In this article we introduce the problem of timed pattern diagnosis in timed system and propose a method to solve a sub-class of theses problems: the problem of timed pattern diagnosis of workflows. It consists in searching for all the possible evolutions of the workflow that are consistent with a timed sequence of observations and determining whether these evolutions match a pattern of timed events instead of a single fault event. The formal characterization of the diagnosis problem is based on the notion of pattern matching. Defined as a reachability problem, the diagnosis problem is then solved by model checking.
Christian Artigues, Emmanuel Hébrard, Yannick Pencolé, Andreas Schutt
A Study of Evacuation Planning for Wildfires
7th International Workshop on Constraint Modelling and Reformulation Lille France August 2018
The GEO-SAFE project gathers researchers and fire emergency practitioners from EU and Australia with the aim to design innovative models and efficient response tools based on optimization methodsfor fighting wildfires. In this paper, we consider an evacuation planning problem issued from discussions with practitioners, where a wildfire is threatening a region with intermediate populated centers. As in earlier approaches in case of a flood, we use a constraint optimization model involving malleable tasks to represent the evacuation of a population and a cumulative constraint per route segments. Indeed, in order to mitigate congestion risks, the authorities may delay the start of the evacuation but they may also affect the rate of evacuation by modulating the method used to raise the alarm. However, we consider a different objective: we maximize the minimum “safety margin”, weighted by the population, over every road segment. We introduce a new heuristic and a global flow constraint propagator. Moreover, we also propose an instance generator based on a random generation of road networks and basic fire propagation models. This generator produces challenging benchmarks even with very few evacuation tasks. Finally, we report the results of extensive computational experiments done using CP Optimizer.
Christian Artigues, Emmanuel Hébrard, Yannick Pencolé, Andreas Schutt, Peter Stuckey
Data Instance generator and optimization models for evacuation planning in the event of wildfire
GEOSAFE Workshop on Robust Solutions for Fire Fighting 75-86 L'Aquila Italy July 2018
One critical part of decision support during the response phase to a wildfire is the ability to perform large-scale evacuation planning. While in practice most evacuation planning is principally designed by experts using simple heuristic approaches or scenario simulations, more recently optimization approaches to evacuation planning have been carried out, notably in the context of floodings. Evacuation planning in case of wildfires is much harder as wildfire propagations are inherently less predictable than floods. This paper present a new optimization model for evacuation planning in the event of wildfire aiming at maximizing the temporal safety margin between the evac-uees and the actual or potential wildfire front. As a first contribution, an open-source data instance generator based on road network generation via quadtrees and a basic fire propagation model is proposed to the community. As a second contribution we propose 0–1 integer programming and constraint programming formulations enhanced with a 1 simple compression heuristic that are compared on 240 problem instances build by the generator. The results show that the generated instances are computationally challenging and that the contraint programming framework obtains the best performance.
Euriell Le Corronc, Alexandre Sahuguède, Yannick Pencolé, Claire Paya
Localization of time shift failures in (max,+)-linear systems
14th Workshop on Discrete Event Systems 186-191 Sorrento Coast Italy May 2018
The goal of this paper is to propose a localization method of time shift failures in timed discrete event systems (TDES) called (max, +)-linear systems and graphically represented by Timed Event Graph (TEG). First, a detection process produces indicators that determine whether such failures have happened by the observation of incoming and outcoming timed flows. Then, thanks to the knowledge of the behavior of the system through its corresponding TEG, set of failures that could explain the detected timed shift are obtained. It comes from matrices of signatures for each indicator built on each observable output of the system.
Yannick Pencolé, Audine Subias
Diagnosis of supervision patterns on bounded labeled Petri nets by Model Checking
28th International Workshop on Principles of Diagnosis 184-199 Brescia Italy January 2018
This paper investigates the problem of pattern diagnosis of systems modeled as bounded labeled Petri nets that extends the diagnosis problem on single fault events to more complex behaviors. An effective method to solve the diagnosis problem is proposed. It relies on a matching relation between the system and the pattern that turns the pattern diagnosis problem into a model-checking problem.
Yannick Pencolé, Gerald Steinbauer, Clemens Mühlbacher, Louise Travé-Massuyès
Diagnosing Discrete Event Systems Using Nominal Models Only
28th International Workshop on Principles of Diagnosis 169-183 Brescia Italy January 2018
Complex technical systems usually show a dynamic behavior that is often conveniently represented with a discrete event model. Such a behavior is the result of dynamic components which interact with each other. Due to the complexity of technical systems faults are not totally avoidable. In order to deal with such faults diagnosing the system at run-time is of great interest. To perform such a diagnosis it is common to use fault models. Such models are in practice often hard to obtain. To address this problem we show a diagnosis approach for discrete event systems which uses the model of the nominal behavior only. In order to perform this diagnosis we adopt the well known idea of consistency based diagnosis.
Cody Christopher, Yannick Pencolé, Alban Grastien
Inference of fault signatures of discrete-event systems from event logs
28th International Workshop on Principles of Diagnosis 219-233 Brescia Italy January 2018
In this paper, we propose a method to diagnose faults in a discrete event system that only relies on past observed logs and not on any behavioural model of the system. Given a set of tagged logs produced by the system, the first objective is to extract from them a set of fault signatures. These fault signatures are represented with a set of critical observations that are the support of the diagnosis method. We first propose a method to compute the fault signatures from an initial log journal and follow with detail on how the signatures can then be updated when new logs are available.
2017
Euriell Le Corronc, Alexandre Sahuguède, Yannick Pencolé
Détection et localisation de fautes temporelles dans les systèmes (max,+)-linéaires
11ème Colloque sur la Modélisation des Systèmes Réactifs Marseille France November 2017
L'automatisation du diagnostic de pannes dans les systèmes industriels tels que les systèmes manufacturiers ou de transport souvent modélisés par des Systèmes à Evénements Discrets (SED) est un enjeu majeur. Il est nécessaire de déterminer rapidement les problèmes sur les équipements afin de les réparer au plus tôt et ainsi maximiser le temps opérationnel et productif du système. Parmi les différents types de pannes, les pertes de performances temporelles sont régulièrement rencontrées comme par exemple une baisse du taux de production de pièces dans un système d'assemblage ou encore un ralentissement de convoyeurs dans un réseau de transport de bagages dans un aéroport. Ces problèmes ne sont pas facilement modélisable avec les méthodes de diagnostic à base de modèles à événements discrets connues à ce jour si bien que les outils classiques de diagnostic ne sont a priori pas bien adaptés pour en déterminer la ou les causes.
L'objectif de cet article est de proposer une méthode de détection et de localisation de fautes temporelles pour une classe de SED temporisés appelés systèmes (max,+)-linéaires. Ce type de système permettant la représentation de phénomènes de synchronisation entre ressources, de durées d'opération et de temps de transmission est particulièrement bien adapté pour identifier des décalages temporels non souhaités typiques des problèmes dans les procédés industriels.
La méthode proposée se déroule en plusieurs étapes. A partir d'un flux d'observations temporisées, le problème consiste tout d'abord à déterminer si ce flux résulte d'un comportement en présence d'une faute temporelle ou non. Des indicateurs de fautes sont utilisés pour répondre à cette question et reposent sur l'utilisation de la théorie de la résiduation appliquée aux systèmes (max,+)-linéaires. Si la faute est détectée, il est alors nécessaire de la localiser au mieux dans le système. Les fautes temporelles potentielles sont donc répertoriées et ajoutées au modèle (max,+) du système en comportement normal. Ensuite, de manière analogue à la théorie classique du diagnostic dans les systèmes à temps continu, des matrices de signatures sont proposées et permettent de déterminer les fautes candidates expliquant les observations.
Yannick Pencolé, Audine Subias
Diagnostic de motifs de comportements dans les systèmes temporels
11ème Colloque sur la Modélisation des Systèmes Réactifs Marseille France November 2017
Dans cet article nous proposons une formulation du problème de diagnostic de motifs d'un système temporel. Il s'agit de rechercher à partir d'une séquence d’observations du système, toutes les évolutions de celui-ci qui non seulement produisent la séquence observée mais aussi qui sont concordantes avec les motifs d'intérêt considérés. La caractérisation formelle du problème de diagnostic s'appuie sur la notion de pattern matching (concordance de motifs). Posé sous forme d’un problème d'atteignabilité, le problème de diagnostic est résolu par model checking.
Houssam-Eddine Gougam, Yannick Pencolé, Audine Subias
Diagnosability analysis of patterns on bounded labeled
prioritized Petri nets
Journal of Discrete Event Dynamic Systems: Theory and Applications 27(1) 143-180 March 2017
Checking the diagnosability of a discrete event system aims at determining whether a fault can always be identified with certainty after the observation of a bounded number of events. This paper investigates the problem of pattern diagnosability of systems modeled as bounded labeled prioritized Petri nets that extends the diagnosability problem on single fault events to more complex behaviors. An effective method to automatically analyze the diagnosability of a pattern is proposed. It relies on a specific Petri net product that turns the pattern diagnosability problem into a model-checking problem.
Alexandre Sahuguède, Euriell Le Corronc, Yannick Pencolé
Design of indicators for the detection of time shift failures in (max, +)-linear systems
20th World Congress of the International Federation of Automatic Control Toulouse France July 2017
In this paper, we address the problem of failure detection in a timed discrete event system (TDES). We first introduce the problem of detecting time shift failures in a TDES modeled as a (max, +)-linear system. Then we propose the definition of an indicator that relies on the (max, +) algebraic framework and show how it can detect time shift failures in the case of a single output system. Finally, an extension is proposed to deal with multiple outputs.
2016
Yannick Pencolé, Elodie Chanthery, Thierry Peynot
Definition of Model-based diagnosis problems with Altarica
27th International Workshop on Principles of Diagnosis Denver, Colorado United States October 2016
This paper presents a framework for modeling diagnosis problems based on a formal language called Altarica. The initial purpose of the language Altarica was to define a modeling language
for safety analysis. This language has been developed as a collaboration between academics and industrial partners and is used in some industrial companies. The paper shows that the expres-
sivity of this language, mixing event-based and state-based models, is sufficient to model classical model-based diagnosis problems (logic-based and event-based) and problems that combine state-based and event-based behaviors. This modeling framework, whose semantics is fully specified, is a promising framework to design richer diagnosis problems. As example, we introduce a robotic
diagnosis problem that benefits from the full expressivity of the language.
2015
Ghyslain Maitre, Yannick Pencolé, Audine Subias, Houssam-Eddine Gougam
Modélisation et Analyse de chroniques pour le diagnostic
Modélisation des Systèmes Réactifs Nancy France November 2015
Dans cet article nous proposons un modèle de chronique qui peut être vu comme une extension d'un Problème Temporel Simple, permettant d'intégrer dans les chroniques les contraintes d'interdiction représentant l'absence d'un événement dans l'ensemble du comportement modélisé ou sur une partie prédéfinie. Sur la base de cette formalisation nous proposons des critères permettant de caractériser les chroniques et de les comparer afin d'évaluer la pertinence de la base de chroniques utilisée pour le diagnostic. En particulier nous proposons une méthode d'analyse de diagnosticabilité reposant sur un test d'exclusivité des chroniques. Les chroniques sont modélisées par réseaux de Petri temporels et le test d'exclusivité est posé sous forme d'un problème d'atteignabilité et résolu par une méthode de vérification automatique de modèle.
Yannick Pencolé
Random generator of k diagnosable discrete event systems
26th international workshop on Principles of Diagnosis 277-280 Paris France August 2015
This paper presents a random generator of discrete event systems that are by construction k-diagnosable. The aim of this generator is to provide an almost infinite set of diagnosable systems for
creating benchmarks. The goal of such benchmarks is to provide a solid set of examples to test and compare algorithms that solve many problems around diagnosable discrete event systems.
Elodie Chanthery, Yannick Pencolé, Pauline Ribot, Louise Travé-Massuyès
HyDiag: extended diagnosis and prognosis for hybrid systems
26th international workshop on Principles of Diagnosis 281-284 Paris France August 2015
HYDIAG is a software developed in Matlab by the DISCO team at LAAS-CNRS. It is currently a software designed to simulate, diagnose and prognose hybrid systems using model-based techniques. An extension to active diagnosis is also provided. This paper aims at presenting the native HY D IAG tool, and its different extensions to prognosis and active diagnosis. Some results on an academic example are given.
Louise Travé-Massuyès, Renaud Pons, Pauline Ribot, Yannick Pencolé, Carine Jauberthie
Condition-based Monitoring and Prognosis in an Error-Bounded Framework
26th international workshop on Principles of Diagnosis 83-90 Paris France August 2015
Condition-based maintenance is recognized as a better health management strategy than regularly planned inspections as used nowadays by most companies. In practice, it is however difficult to implement because it means being able to predict the time to go before a failure occurs. This prediction relies on knowing the current health status of the system’s components and on predicting how components age. This paper demonstrates the applicability of interval-based tools in integrated health management architectures, hence proposing an alternative to the standard statistical approach.
Yannick Pencolé, Romain Pichard, Pierre Fernbach
Modular fault diagnosis in discrete-event systems with a CPN diagnoser
9th IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes 470-475 Paris France September 2015
This paper addresses the problem of fault diagnosis in discrete-event system. The system under investigation is modelled as a labelled Petri net. We first propose an equivalent encoding of the classical diagnoser with the help of a CPN diagnoser. We then apply this encoding to define a modular CPN diagnoser.
2014
Houssam-Eddine Gougam, Audine Subias, Yannick Pencolé
Discriminability Analysis of Supervision Patterns by Net Unfoldings
12th IFAC - IEEE International Workshop on Discrete Event Systems 459-464 Cachan France May 2014
In this paper, we are interested in the discriminability of supervision patterns, in discrete event systems (DES). Discriminability — as opposed to diagnosability — is the possibility to detect the exclusive occurrence of a particular behavior of interest — called the supervision pattern. To this end, we propose to adapt the classical twin-plant approach to Petri nets unfolding. The usage of unfoldings permits us to avoid the combinatorial explosion associated with marking graphs. The method can also be used to solve the classical problem of discrete event systems’ diagnosability.
Yannick Pencolé
DITO: a CSP-based diagnostic engine
21st European Conference on Artificial Intelligence 699-704 Prague Czech Republic August 2014
We present a new generic diagnostic engine to diagnose systems that are modelled with first order logic. The originality of this engine is that it takes benefit of recent advances in constraint programming to perform satisfiability tests by the use of an off-the-shelf CSP solver. The second contribution of this paper is the definition of an incremental search strategy that deals with the intrinsic complexity of the problem to look for minimal diagnoses in complex digital systems. The use of the DITO engine and its strategy is fully illustrated on the c6288 circuit that is part of the classical ISCAS85 benchmark.
Xingyu Su, Alban Grastien, Yannick Pencolé
Window-based Diagnostic Algorithms for Discrete Event Systems: What Information to Remember
25th International Workshop on Principles of Diagnosis Graz Austria November 2014
Window-based diagnostic algorithms only use the most recent observations to diagnose a discrete event system. This approach can cause precision loss if less recent observations are necessary to understand the system behavior. This paper formally presents two extensions of window-based algorithms that solve this issue by carrying over some information about the current state of the system from one window to the next. We then show how precision for the two algorithms can be decided using our precision test depending on the amount of information that is retained. Finally, we
propose a procedure that minimizes this amount of information so that the proposed algorithms remain precise
2013
Marie-Odile Cordier, Philippe Dague, Yannick Pencolé, Louise Travé-Massuyès
Diagnostic et supervision : approches à base de modèles
Représentation des connaissances et formalisation des raisonnements - chapitre 18 2013
Pauline Ribot, Yannick Pencolé, Michel Combacau
Generic characterization of diagnosis and prognosis for complex heterogeneous systems
International Journal of Prognostics and Health Management 4(2) 2013
Maintenance efficiency of complex industrial systems is an important economical and business issue. Main difficulties come from the choice of maintenance actions. A wrong choice can lead to maintenance costs that are not acceptable. In this paper, we propose a generic health monitoring system that integrates some diagnostic and prognostic capabilities to determine the current and future state of a large and complex system such as an aircraft. The diagnostic function aims at identifying faulty components that may cause global system failures. The prognostic function estimates the remaining time until the next global system failure. A formal and generic modeling framework for a complex system encapsulating the knowledge required to get the consistent coordination of the diagnostic and prognostic functions is presented. We propose in this framework to take into account component redundancies which is common in systems like aircrafts. Moreover, an original coupling of diagnosis and prognosis is established based on the characterization of the system operational modes and on a decentralized architecture of the monitoring system.
Houssam-Eddine Gougam, Audine Subias, Yannick Pencolé
Diagnosticabilité de motifs de supervision par dépliage de réseaux de Petri
Journées Doctorales Journées Nationales MACS Strasbourg France July 2013
Ce papier s’intéresse au problème de la diagnosticabilité des motifs de supervision dans les systèmes à événements discrets. Ces motifs permettent de prendre en compte
des comportements de fautes complexes. La méthode d’analyse de diagnosticabilité s’appuie sur l’utilisation des réseaux de Petri et sur les méthodes de dépliage de manière à appréhender les problèmes d’explosion combinatoire induits par des approches classiques d’analyse par graphe des marquages. L’approche proposée consiste à vérifier la diagnosticabilité de chaque motif en adaptant la méthode classique du produit jumelé (twin-plant). La non-diagnosticabilité du motif de faute est vérifiée par la recherche de séquences infinies ambiguës dans le dépliage du produit.
Houssam-Eddine Gougam, Audine Subias, Yannick Pencolé
Supervision Patterns: Formal Diagnosability Checking by Petri Net Unfolding
4th IFAC Workshop on Dependable Control of Discrete Systems 73-78 York United Kingdom September 2013
This paper addresses the problem of checking diagnosability of supervision patterns in discrete-event systems. With a supervision pattern, it is possible to represent a complex behavior of the system, and especially a faulty behavior. As opposed to classical diagnosability analyzers that check by exploring the marking graph of the underlying net, the proposed method relies on Petri net unfoldings and thus avoids the combinatorial explosion induced by the use of marking graphs. The method is an adaptation of the twin-plant method to net unfolding: a pattern is diagnosable if the unfolding representing the twin-plant does not implicitly contain infinite sequences of events that are ambiguous.
2012
Houssam-Eddine Gougam, Audine Subias, Yannick Pencolé
Timed diagnosability analysis based on chronicles
8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes 1256-1261 Mexico Mexico August 2012
Automated chronicle recognition is an efficient and robust method for fault diagnosis in timed discrete-event systems (TDES). This paper addresses the problem of diagnosability of TDES with regards to such a diagnosis method. We propose a fully automated chain to a priori check whether faults can be identified with certainty based on a given set of chronicles. To deal with the time aspects inherent to the chronicles, we first propose an automated translation of
chronicles into a set of Labeled Time Petri Nets with Priorities. The diagnosability analysis is then performed on the state class graph of these nets and consists in determining whether the recognition of a chronicle is exclusive or not.
Nuno Belard, Michel Combacau, Yannick Pencolé
Meta-Diagnosis in FDI: Reasoning About False Analytical Redundancy Relations
8th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes 379-384 Mexico Mexico August 2012
Fault Detection and Isolation (FDI) analytical-redundancy-based systems rely on a model of a real-world system and on some observations obtained from sensor readings to determine what faults are present in that same system at a given time. In this framework, it is sometimes assumed that the models used are a true representation of the artefact under study. Unfortunately, in real-world applications this is not always the case and errors in models may entail false diagnoses with huge economic consequences. Call the problem of detecting and identifying faults in models a problem of meta-diagnosis; an unsolved issue in the FDI community and a very difficult problem to address especially in the case of complex systems. In this paper, we contribute by providing this community with a method of meta-diagnosis making use of the link between the FDI analytical redundancy approach and the DX consistency-based logical approach; and illustrate such contribution with a DC motor example. Finally, the meta-diagnosis is generalised for detecting and identifying errors in observations and algorithms.
Yannick Pencolé
Fault Diagnosis in Discrete-Event Systems: How to Analyse Algorithm Performance?
Diagnostic reasoning: Model Analysis and Performance 19-25 Montpellier France August 2012
This paper addresses the problem of analyzing the performance of algorithms that solve the fault diagnosis problem in discrete-event systems in an experimental way. To achieve this purpose, we define a set of metrics for algorithm comparison on one hand. On the second hand, we propose an experimental platform based on a tool called DIADES (Diagnosis of Discrete-Event Systems) to run experiments and measure different algorithms.
Carine Jauberthie, Yannick Pencolé, Renaud Pons, Pauline Ribot, Louise Travé-Massuyès
Diagnosis and prognosis in health monitoring systems. State of the art.
12680 December 2012
2011
Nuno Belard, Yannick Pencolé, Michel Combacau
Théorie de méta-diagnostic: raisonnement sur les systèmes de diagnostic
Journées de l'Intelligence Artificielle Fondamentale Lyon France June 2011
Le principe du diagnostic à base de modèle est mis en œuvre à l’aide d’un algorithme qui, s’appuyant sur un modèle du système sous-jacent et de son observation, génère des hypothèses de diagnostic. Dans les applications réelles et contrairement aux hypothèses classiques, il se peut que ni le modèle, ni les observations, ni même l’algorithme implanté ne remplissent les pré-requis nécessaires à la bonne marche du système de diagnostic ; ce qui peut aboutir, le cas échéant, à de graves conséquences économiques. Le problème de méta-diagnostic consiste à déterminer les anomalies du système de diagnostic utilisé. Dans cet article, nous proposons dans un premier temps, une théorie générale du méta-diagnostic avec une sémantique clairement définie pour résoudre ce problème. Puis, nous établissons une liste des pré-requis au bon fonctionnement d’un système de diagnostic et leurs relations. Enfin, à l’aide du cadre théorique proposé, nous proposons et résolvons des exemples de problèmes de méta-diagnostic s’appuyant sur les prérequis définis précédemment
Nuno Belard, Yannick Pencolé, Michel Combacau
MEDITO: a logic-based meta diagnosis tool
IEEE International Conference on Tools with Artificial Intelligence 709-716 Boca Raton (FL) United States November 2011
In every Model-Based Diagnosis (MBD) approach, a model of a real-world system and some observations of such a system are used by a diagnostic algorithm to compute diagnoses. Contrary to MBD classical hypotheses, real-world applications provide us with empirical data suggesting that diagnostic systems, i.e. a model, observations and a diagnostic algorithm, are sometimes abnormal with respect to some required properties. This is where Meta-Diagnosis comes into play with a theory to determine abnormalities in diagnostic systems. Unfortunately, Artificial Intelligence lacks of a tool putting meta-diagnosis theory to practice. Our first contribution in this paper is such a tool, called MEDITO. Moreover, we provide a real-world example of MEDITO's application at meta-diagnosing an Airbus landing gear extraction and retraction system with successful results.
Nuno Belard, Yannick Pencolé, Michel Combacau
A theory of meta-diagnosis: reasoning about diagnostic systems
22nd International Joint Conference on Artificial Intelligence 731-737 Barcelona Spain 2011
In Model-Based Diagnosis, a diagnostic algorithm is typically used to compute diagnoses using a model of a real-world system and some observations. Contrary to classical hypothesis, in real-world applications it is sometimes the case that either the model, the observations or the diagnostic algorithm are abnormal with respect to some required properties; with possibly huge economical consequences. Determining which abnormalities exist constitutes a meta-diagnostic problem. We contribute, first, with a general theory of meta-diagnosis with clear semantics to handle this problem. Second, we propose a series of typically required properties and relate them between themselves. Finally, using our meta-diagnostic framework and the studied properties and relations, we model and solve some common meta-diagnostic problems.
2010
Nuno Belard, Yannick Pencolé, Michel Combacau
Defining and exploring properties in diagnostic systems
21st International Workshop on Principles of Diagnosis 161-168 Portland (OR) United States October 2010
Every model-based diagnostic approach relies on a representation of a real-world system, in this paper called believed system. The believed system
is used along with the observations about the real-world system to generate a diagnostic problem to be solved. In this paper it is firstly argued that
believed systems can differ from real-world systems in many different manners. As so, properties of believed systems, diagnostic problems and
diagnostic results are introduced. Then, a series of relations between these properties are proved. The importance of such relations, sometimes seen
as intuitive, is that they are necessary to formally prove the accordance between the real-world system and the believed system; to formally prove
that a believed system and a diagnostic problem will produce high-quality diagnostic results; or even to ease diagnostic algorithms, since for systems and problems with certain properties, different model-based diagnostic approaches produce the same diagnostic results. In order to introduce the referred properties and reasoning about them a framework of diagnosis based on the difference between the believed and the real systems is proposed.
Priscilla Kan John, Alban Grastien, Yannick Pencolé
Synthesis of a distributed and accurate diagnoser
21st International Workshop on Principles of Diagnosis 209-216 Portland (OR) United States October 2010
The complex behaviour of large discrete event systems makes such systems difficult to diagnose. Using decentralised techniques helps limit combinatorial explosion but is not sufficient. Often, the complexity of the diagnosis is dependent on
how components in the system are connected and the number of connections between them. We propose to augment a decentralised junction tree-based approach by ignoring some connections on the system. This helps reduce the complexity, and hence the cost, of the diagnostic reasoning required. However accuracy of the diagnosis is also reduced. We get around this problem by performing an off-line analysis to determine which connections can be safely ignored.
Elodie Chanthery, Yannick Pencolé, Nicolas Bussac
An AO*-like algorithm implementation for active diagnosis
10th International Symposium on Artificial Intelligence, Robotics and Automation in Space 378-385 Sapporo Japan August 2010
This paper details an algorithm that solves the active diagnosis problem. This algorithm relies on an AO*-search and computes a conditional plan to apply in order to refine the diagnosis. The input of the AO* is a set of ambiguous diagnosis candidates (represented as belief states) processed by the diagnoser at the time when the active diagnosis session starts. The search-space of AO* is defined by the behavioral model of the system that is common to the diagnoser and the diagnosis planner. Several problems have to be addressed. The first one is to adapt AO* algorithm to an optimization problem with various costs and rewards. The challenge is to find a relevant criterion that may be used by all kinds of autonomous systems (like satellites or rovers). The second challenge is to define an efficient heuristic in order to prune the search-space and to guide the search. This heuristic has to take into account the goal of active diagnosis and the mission goal of the autonomous system (mission model). Finally, the computational time required by this planner must be compatible with the time constraints inherent to autonomy in spatial systems.
Anika Schumann, Yannick Pencolé, Sylvie Thiébaux
A decentralised symbolic diagnosis approach
19th European Conference on Artificial Intelligence 99-104 Lisbon Portugal August 2010
This paper considers the diagnosis of large discrete-event systems consisting of many components. The problem is to determine, on-line, all failures and states that explain a given sequence of observations. Several model-based diagnosis approaches deal with this problem but they usually have either poor time performance or result in space explosion. Recent work has shown that both problems can be tackled when encoding diagnosis approaches symbolically by means of binary decision diagrams. This paper further improves upon these results and presents a decentralised symbolic diagnosis method that computes the diagnosis information for each component off-line and then combines them on-line. Experimental results show that our method provides significant improvements over existing approaches.
Priscilla Kan John, Alban Grastien, Yannick Pencolé, Pauline Ribot
Synthèse d'un diagnostiqueur distribué et précis
17ème congrès francophone AFRIF-AFIA Reconnaissance des Formes et Intelligence Artificielle 646-653 Caen France January 2010
Le diagnostic de grands systèmes à événements discrets distribués est confronté au problème d’explosion combinatoire. Une mise en œuvre de l’algorithme de diagnostic (diagnostiqueur) utilise un arbre de jonction sur la topologie. L’efficacité de cette méthode dépendant directement du degré d’intrication de la topologie, nous proposons d’adapter l’approche en ignorant certaines connexions. Cependant, ce faisant, on met en péril la précision du diagnostic. Aussi, nous effectuons au préalable une analyse pour déterminer quelles connexions peuvent être ignorées sans compromettre la précision du diagnostiqueur.
2009
Xavier Pucel, Louise Travé-Massuyès, Yannick Pencolé
Another Point of View on Diagnosability
7th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes 95-100 Barcelona Spain June 2009
This paper provides a new definition of diagnosability, that allows one to check the diagnosability of any set of system states, and by extension of properties that depend on the system state. The existing definitions and approaches for checking diagnosability apply to faults or sets of faults, and comparison shows that the new definition generalizes the existing ones. This new definition is applied to repair preconditions, and an example shows how this brings complementary information compared to classical fault diagnosability.
Luca Console, Danilo Ardagna, Liliana Ardissono, Stefano Bocconi, Cinzia Cappiello, Marie-Odile Cordier, Khalil Drira, Johann Eder, Gerhard Friedrich, Mariagrazia Figini, Roberto Furnari, Anna Goy, Karim Guennoun, Andreas Hess, Vladimir Ivanchenko, Xavier Le Guillou, Marek Lehmann, Jürgen Mangler, Yingmin Li, Tarek Melliti, Stefano Modafferi, Enrico Mussi, Yannick Pencolé, Giovanna Petrone, Barbara Pernici, Claudia Picardi, Xavier Pucel, Sophie Robin, Laurence Rozé, Marino Segnan, Nick Tahamtan, Annette ten Teije, Daniele Theiseder-Dupré, Louise Travé-Massuyès, Frank van Harmelen, Thierry Vidal, Audine Subias
WS-DIAMOND web services - DIAgnosability, MONitoring and Diagnosis
At Your Service. Service-Oriented Computing from an EU Perspective 213-240 June 2009
Yannick Pencolé, Audine Subias
A chronicle-based diagnosability approach for discrete timed-event systems: application to Web-Services
Journal of Universal Computer Science 15(17) 3246-3272 November 2009
This paper addresses the problem of diagnosability analysis in Web Services. In particular, it focuses on the analysis of the impact of time to the diagnostic capabilities in Web Service workflows. The diagnosability analysis that is proposed in this paper aims at determining the diagnostic capabilities of a previously developped algorithm for the diagnosis of Web Services. This diagnostic algorithm is based on chronicle recognitions. Faults that can occur during the execution of service workflows are described by means of chronicles.
To perform this diagnosability analysis, the problem is firstly defined as a languagebased analysis which leads to the definition of exclusiveness tests between the languages represented by the chronicles. To deal with the time aspects inherent to the chronicles, we then propose to perform the automatic analysis by the use of time Petri nets. Exclusiveness tests are then defined on reachability graphs of time Petri nets which implicitly represent chronicle languages.
Yuhong Yan, Philippe Dague, Yannick Pencolé, Marie-Odile Cordier
A Model-Based Approach for Diagnosing Fault in Web Service Processes
International Journal of Web Services Research 6(1) 87-110 2009
Web services based on a service-oriented architecture framework provide a suitable technical foundation for business process management and integration. A business process can be composed of a set of Web services that belong to different companies and interact with each other by sending messages. Web service orchestration languages are defined by standard organizations to describe business processes composed of Web services. A business process can fail for many reasons, such as faulty Web services or mismatching messages. It is important to find out which Web services are responsible for a failed business process because we could penalize these Web services and exclude them from the business process in the future. In this paper, we propose a model-based approach to diagnose the faults in a Web service-composed business process. We convert a Web service orchestration language, more specifically BPEL4WS, into synchronized automata, so that we have a formal description of the topology and variable dependency of the business process. After an exception is thrown, the diagnoser can calculate the business process execution trajectory based on the formal model and the observed evolution of the business process. The faulty Web services are deduced from the variable dependency on the execution trajectory. We demonstrate our diagnosis technique with an example.
Pauline Ribot, Yannick Pencolé, Michel Combacau
Functional Prognostic Architecture for the Maintenance of Complex Systems
7th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes Barcelona Spain June 2009
This paper addresses the problem of the maintenance of a complex system of heterogeneous components. With the ecreasing costs of sensors, it now becomes possible to really benefit of on-line adaptive prognostic methods in order to support the optimization of the maintenance actions and scheduling. In this paper, we propose a generic health monitoring architecture that encompasses the available prognostic methods and provides a common support for the maintenance decision on a complex system. The output of this architecture takes into account the composed nature of the system by not only providing component prognoses but also function prognosis. In order to provide a prognosis for the whole system, component prognoses are combined by taking into account functional dependencies in the system.
Pauline Ribot, Yannick Pencolé, Michel Combacau
Diagnosis and prognosis for the maintenance of complex systems
IEEE International Conference on Systems, Man and Cybernetics, 2009. 4146-4151 San Antonio, Texas United States October 2009
This paper adresses the problem of maintenance of a complex and heterogeneous system like an aircraft. To optimise maintenance, it is required to embed in the system a health monitoring system that implements diagnostic and prognostic capabilities. This paper thus presents a formal characterisation of the diagnostic and prognostic problems in order to support the maintenance of a complex system.
Elodie Chanthery, Yannick Pencolé
Modélisation et intégration du diagnostic actif dans une architecture embarquée
Journal européen des systèmes automatisés 43 789-803 2009
Cet article présente la formalisation du principe de diagnostic actif, et son intégration comme un module fonctionnel dans une architecture embarquée. L’objectif du diagnostic actif est de trouver une séquence d’actions permettant de raffiner le diagnostic sans changer radicalement le plan de mission. On définit le problème comme un problème de planification. Un critère pondéré est explicité. Le but est de trouver le meilleur plan conditionnel permettant au diagnostiqueur d’atteindre une région non ambiguë. Les conflits possibles entre le plan de la mission et le plan pour le diagnostic sont gérés par un contrôleur d’exécution au sein d’une architecture embarquée. Le contrôle d’exécution est spécifié à l’aide de réseaux de Petri.
Elodie Chanthery, Yannick Pencolé
Monitoring and Active Diagnosis for Discrete-Event Systems
7th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes Barcelona Spain July 2009
This article presents an original way to enrich the monitoring of discrete-event systems named active diagnosis. The objective of on-line active diagnosis is to find an admissible sequence of actions (or plan) that refines the diagnosis without radically changing the mission plan. This paper has two major contributions. First, active diagnosis is formally defined in the finite-state automata theory framework. This leads to the definition of a complete active diagnoser which on line monitors the system behavior. Secondly, from the complete active diagnoser is defined a planning problem. The goal is to find a conditional plan that defines an admissible sequence of actions. These actions are applied on the physical system and may conduct the active diagnoser into a diagnosable region.
Elodie Chanthery, Yannick Pencolé
Principles of self-maintenance in an on-board architecture including active diagnosis
The IJCAI-09 Workshop on Self-* and Autonomous Systems: reasoning and integration challenges 43-50 Pasadena, California United States July 2009
This article presents some principles for self-maintenance in an on-board architecture that embeds on-line active diagnosis. The objective of active diagnosis is to find an action plan that refines the diagnosis without radically changing the mission plan. This leads to the definition of a planning problem that relies on an active diagnoser. According to a decision criterion, the aim is then to find the best conditional plan to reach a diagnosable region of the active diagnoser. The interactions problems between the result of the planning for diagnosis and the mission planning are then detailed and a solution that uses the execution controller of the on-board architecture is proposed.
2008
Marie-Odile Cordier, Yannick Pencolé, Louise Travé-Massuyès, Thierry Vidal
Characterizing and checking self-healability
18th European Conference on Artificial Intelligence 789-790 Patras Greece July 2008
Real-life complex systems are often required to offer high reliability and quality of service and must be provided with self-management abilities, even in faulty situations. They are expected to be self-aware of their current state and survive autonomously the occurrence of faults, still managing to provide the desired functionality. In other words, such systems must be self-healing. Designing self-healing systems requires to be able to evaluate the joint degree of self-awareness and reactiveness. In the artificial intelligence community, these two properties are better known as diagnosability, i.e. the capability of a system to exhibit different observables for different anticipated faulty situations, and repairability, i.e. the ability of a system and its repair actions to cope with any unexpected situation.
Checking separately diagnosability and repairability leads to a conservative assessement of self-healability. In this paper, we show that neither standard diagnosability nor repairability of every anticipated fault are necessary to achieve self-healability. Our main contribution consists of defining self-healability as a joint property bridging diagnosability and repairability, which requires a new definition
of diagnosability that allows diagnosable subsets of faults to overlap, as opposed to the standard definitions which rely on a partition.
Marie-Odile Cordier, Yannick Pencolé, Louise Travé-Massuyès, Thierry Vidal
Caractérisation des systèmes autoguérissants : diagnostiquer ce que l'on va réparer
16e congrès francophone AFRIF-AFIA Reconnaissance des Formes et Intelligence Artificielle Amiens France January 2008
Les systèmes complexes basés sur des architectures informatiques se doivent d’être capables “d’autoguérison”, c’est-à-dire capables, sans intervention extérieure, de détecter, diagnostiquer et réparer les effets de l’occurrence de fautes, pour continuer à assurer leurs fonctionnalités. Les concepteurs de tels systèmes ont besoin d’outils permettant de vérifier avant leur mise en œuvre opérationnelle qu’ils
sont bien “autoguérissants”. Mais auparavant, il est nécessaire de définir précisément et formellement ce qu’autoguérison signifie. La diagnosticabilité, qui est la capacité d’un système à déterminer l’état fautif dans lequel il se trouve à partir des observations dont il dispose, est une propriété bien connue, mais qui néglige les capacités de réparation de ces fautes. La réparabilité au contraire, vue comme la capacité d’un système à disposer de plans de réparation adaptés aux fautes, ne prend pas en compte les capacités de diagnostic. Notre article établit le lien entre ces deux propriétés, définissant la capacité d’autoguérison comme une combinaison de la diagnosticabilité, dont il nous a fallu étendre la définition classique, et de la réparabilité, dont nous proposons une première définition formelle. Nous suggérons à partir de là un algorithme de vérification pratique et des stratégies pour aider les concepteurs à rendre leur système autoguérissant.
Pauline Ribot, Yannick Pencolé, Michel Combacau
Diagnosis and prognosis for the maintenance of complex systems
International Conference on Prognostics and Health Management 2008 Denver, Colorado United States October 2008
This paper adresses the problem of maintenance of a complex and heterogeneous system like an aircraft. To optimise maintenance, it is required to embed in the system a health monitoring system that implements diagnostic and prognostic capabilities. This paper thus presents a formal characterisation of the diagnostic and prognostic problems in order to support the maintenance of a complex system.
Pauline Ribot, Yannick Pencolé, Michel Combacau
Design requirements for the diagnosability of distributed discrete event systems
19th International Workshop on Principles of Diagnosis 347-354 Blue Mountains, New South Wales Australia September 2008
We address the problem of fault diagnosabilty in distributed discrete-event systems. Previous works mainly propose different ways to check whether a fault is diagnosable or not. Nowadays, due to the complexity of the engineered systems, this checking is not enough and a better feedback is required in order to redesign and guarantee the diagnosability of a fault. This paper defines the problem of the automatic computation of design requirements for the diagnosability of distributed discrete event system as a cost optimization problem.
Xavier Pucel, Louise Travé-Massuyès, Yannick Pencolé
Another Point of View on Diagnosability
19th International Workshop on Principles of Diagnosis 331-338 Blue Mountains, New South Wales Australia September 2008
This paper provides a new definition of diagnosability, that allows one to check the diagnosability of any set of system states, and by extension of properties that depend on the system state. The existing definitions and approaches for checking diagnosability apply to faults or sets of faults, and comparison shows that the new definition generalizes the existing ones. This new definition is applied to repair preconditions, and an example shows how this brings complementary information compared to classical fault diagnosability.
Marie-Odile Cordier, Yannick Pencolé, Louise Travé-Massuyès, Thierry Vidal
Caractérisation des systèmes autoguérissants : diagnostiquer ce que l'on peut réparer
Information-Interaction-Intelligence 8(2) 123-151 2008
Les architectures informatiques nécessitent aujourd'hui des capacités "d'autoguérison". c'est-à-dire de diagnostic de l'occurrence de fautes et de réparation de leurs effets, de manière autonome, pour continuer à assurer leurs fonctionnalités. Cela est particulièrement vrai dans le domainge des Services Web, auxquels nous appliquons actuellement ces travaux. Les concepteurs de tels systèmes ont besoin d'outils permettant de vérifier avant leur mise en œuvre opérationnelle qu'ils sont bien "autoguénssants". Pour cela, une première étape consiste à définir formellement ce qu'autoguérison signifie. La diagnosticabiliré est la capacité d'un système à déterminer l'état fautif dans lequel il se trouve à partir des observations dont il dispose. La réparabilité est la capacité d'un système à disposer de plans de réparation adaptés aux fautes. Nous proposons une nouvelle définition de diagnosticabilité, qui ne s'appuie plus sur une partition des fautes comme le fait la définition classique, mais sur un ensemble couvrant de macrofautes et permet d'associer à un système son niveau de diagnosticailité. Nous proposons ensuite une première définition formelle de la réparabilité. Ces deux définitions sont celles qui conviennent pour caractériser la capacité d'auloffuérison d'un système. Nous proposons ainsi une définition d'autoguérison en combinant pour la première fois explicitement la diugnosticabilité et la réparabilité. Un théorème peut alors être démontré dont découle assez directement un algorithme de vérification de l'autoguérison. Nous terminons en montrant comment ce travail peut servir à élaborer des stratégies et aider les concepteurs à analyser leur système et le rendre autoguérissant.
Michel Combacau, Yannick Pencolé, Pauline Ribot
Deliverable WP4 Projet ARCHISTIC - Logical characterisation of the diagnostic problem
September 2008
Pauline Ribot, Yannick Pencolé, Michel Combacau
Deliverable WP5 Projet ARCHISTIC - Solutions for failure prognosis - Prognostic function specification
July 2008
Pauline Ribot, Yannick Pencolé, Michel Combacau
Deliverable WP9.1 Projet ARCHISTIC - Fault diagnosability assurance - Diagnosis performance criteria
September 2008
Cinzia Cappiello, Marie-Odile Cordier, Johann Eder, Enrico Mussi, Yannick Pencolé, Barbara Pernici, Xavier Pucel, Audine Subias, Gaston Tagni, Nick Tahamtan, Annette ten Teije, Louise Travé-Massuyès, Thierry Vidal
A new design methodology. Deliverable D5.2
8799 February 2008
Mickael Dievart, Xavier Desforges, Philippe Charbonnaud, Bernard Archimède, Michel Combacau, Yannick Pencolé, Pauline Ribot
Archistic Project. WP3. Preliminary process and software architecture.
8394 September 2008
Xavier Pucel, Louise Travé-Massuyès, Yannick Pencolé
Another Point of View on Diagnosability
4th Starting Artificial Intelligence Researchers' Symposium 151-162 Patras Greece July 2008
This paper provides a new definition of diagnosability, that allows one to check the diagnosability of any set of system states, and by extension of properties that depend on the system state. The existing definitions and approaches for checking diagnosability apply to faults or sets of faults, and comparison shows that the new definition generalizes the existing ones. This new definition is applied to repair preconditions, and an example shows how this brings complementary information compared to classical fault diagnosability.
2007
Alban Grastien, Yannick Pencolé
Robust diagnosis in composite discrete-event systems
7633 November 2007
Yannick Pencolé
Assistance for the design of a diagnosable component-based system
7322 May 2007
Pauline Ribot, Yannick Pencolé, Michel Combacau
Archistic project, WP2 deliverable. Diagnosis and prognosis - State of the art
7184 April 2007
Michel Combacau, Yannick Pencolé
Agreed definitions for Archistic project. WP1. Fundamentals and requirements
7188 April 2007
Anika Schumann, Yannick Pencolé, Sylvie Thiébaux
A spectrum of symbolic on-line diagnosis approaches
17th International Workshop on Principles of Diagnosis 194-201 Burgos Spain May 2007
This paper deals with the monitoring and diagnosis of large discrete-event systems. The problem is to determine, on-line, all faults and states that explain the flow of observ ations. Model-based diagnosis approaches that first compile the diagnosis information off-line
suffer from space explosion, and those that operate on-line without any prior compilation have poor time performance. Our contribution is a broader spectrum of approaches that suits applications with diverse time and space requirements. Approaches on this spectrum differ in the amount of reasoning and compilation performed off-line and therefore in the way they resolve the tradeoff between the space occupied by the compiled information and the time taken to produce a diagnosis. We tackle the space and time complexity of diagnosis by encoding all approaches in a symbolic framework based on binary decision diagrams. This allows for the compact representation of the compiled diagnosis information, and for its handling across many states at once rather than for each state individually. Our experiments demonstrate the diversity and scalability of our symbolic methods spectrum, as well as its superiority over the corresponding enumerative implementations.
Yannick Pencolé, Xavier Pucel, Audine Subias, Louise Travé-Massuyès
Characterization of diagnosability and repairability for self-healing web services. Deliverable D5.1
7221 April 2007
Liliana Ardissono, Stefano Bocconi, Cinzia Cappiello, Luca Console, Marie-Odile Cordier, Johann Eder, Gerhard Friedrich, Mariagrazia Figini, Roberto Furnari, Anna Goy, Karim Guennoun, Vladimir Ivanchenko, Xavier Le Guillou, Stefano Modafferi, Enrico Mussi, Yannick Pencolé, Giovanna Petrone, Barbara Pernici, Claudia Picardi, Filippo Ramoni, Xavier Pucel, Marino Segnan, Audine Subias, Daniele Theiseder-Dupré, Louise Travé-Massuyès, Thierry Vidal
WS-DIAMOND: an approach to web services - DIAgnosability, MONitoring and Diagnosis
European Research Consortium for Informatics and Mathematics 70 25-26 July 2007
Liliana Ardissono, Stefano Bocconi, Cinzia Cappiello, Luca Console, Marie-Odile Cordier, Johann Eder, Gerhard Friedrich, Mariagrazia Figini, Anna Goy, Karim Guennoun, Vladimir Ivanchenko, Xavier Le Guillou, Stefano Modafferi, Enrico Mussi, Yannick Pencolé, Giovanna Petrone, Barbara Pernici, Claudia Picardi, Filippo Ramoni, Xavier Pucel, Marino Segnan, Audine Subias, Daniele Theiseder-Dupré, Louise Travé-Massuyès, Thierry Vidal
WS-DIAMOND: an approach to web services - DIAgnosability, MONitoring and Diagnosis
Expanding the Knowledge Economy: Issues, Applications, Case Studies 105-112 October 2007
Liliana Ardissono, Stefano Bocconi, Cinzia Cappiello, Luca Console, Marie-Odile Cordier, Johann Eder, Gerhard Friedrich, Mariagrazia Figini, Roberto Furnari, Anna Goy, Karim Guennoun, Vladimir Ivanchenko, Xavier Le Guillou, Stefano Modafferi, Enrico Mussi, Yannick Pencolé, Giovanna Petrone, Barbara Pernici, Claudia Picardi, Xavier Pucel, Filippo Ramoni, Marino Segnan, Audine Subias, Daniele Theiseder-Dupré, Louise Travé-Massuyès, Thierry Vidal
WS-DIAMOND: an approach to web services - DIAgnosability, MONitoring and Diagnosis
eChallenges e-2007 The Hague Netherlands October 2007
The goal of this chapter is to present the guidelines and achievements of the WS-DIAMOND Step Project, funded by the EU Commission under the FET-Open Framework, grant IST-516933. It started in September 2005 and will run until February 2008.
Luca Console, Danilo Ardagna, Liliana Ardissono, Stefano Bocconi, Cinzia Cappiello, Marie-Odile Cordier, Philippe Dague, Khalil Drira, Johann Eder, Gerhard Friedrich, Mariagrazia Figini, Roberto Furnari, Anna Goy, Karim Guennoun, Andreas Hess, Vladimir Ivanchenko, Xavier Le Guillou, Marek Lehmann, Yingmin Li, Jürgen Mangler, Tarek Melliti, Stefano Modafferi, Enrico Mussi, Yannick Pencolé, Giovanna Petrone, Barbara Pernici, Claudia Picardi, Xavier Pucel, Sophie Robin, Laurence Rozé, Marino Segnan, Nick Tahamtan, Annette ten Teije, Daniele Theiseder-Dupré, Louise Travé-Massuyès, Frank van Harmelen, Thierry Vidal
WS-DIAMOND: Web Services - DIAgnosability, MONitoring and Diagnosis
18th International Workshop on Principles of Diagnosis 243-250 Nashville, Tennessee United States May 2007
Self-healing software is one of the challenges for IST research. The WS-DIAMOND project aims at making a step in this direction by developing a framework for self-healing Web Services. In particular, the project aims at: - Defining an framework for self-healing service execution of conversationally complex Web Services, where
monitoring, detection and diagnosis of anomalous situations, due to functional or non-functional errors, are carried on and repair/ reconfiguration is performed, thus
guaranteeing reliability and availability of Web Services; - Defining a methodology and tools for service design that guarantee effective and efficient diagnosability/
repairability during execution. The research builds upon results in different areas such as model-based diagnosis, semantic Web Services, cooperative information systems and Web Service composition. It goes beyond a number of current projects in the area of Service Oriented Computing, which do not consider the monitoring, diagnosing and repairing of Web Services. This paper describes the achievements in the first phase of the project.
Pauline Ribot, Yannick Pencolé, Michel Combacau
Characterization of requirements and costs for the diagnosability of distributed discrete event systems
5th Workshop on Advanced Control and Diagnosis Grenoble France November 2007
We address the problem of fault diagnosabilty in a distributed discrete-event systems. Previous works mainly propose different ways to check whether a fault is diagnosable or not. Nowadays, due to the complexity of the engineered systems, this checking is not enough and a better feedback is required in order to redesign and guarantee the diagnosability of a fault. This paper defines the problem
of the automatic computation of design requirements for the diagnosability of distributed discrete event system as a cost optimization problem.
Anika Schumann, Yannick Pencolé
Scalable Diagnosability Checking of Event-Driven Systems
20th International Joint Conference on Artificial Intelligence 575-580 Hyderabad India January 2007
Diagnosability of systems is an essential property that determines how accurate any diagnostic reasoning can be on a system given any sequence of observations. Generally, in the literature of dynamic event-driven systems, diagnosability analysis is performed by algorithms that consider a system as a whole and their response is either a positive answer or a counter example. In this paper, we present an original framework for diagnosability checking. The diagnosability problem is solved in a distributed way in order to take into account the distributed nature of realistic problems. As opposed to all other approaches, our algorithm also provides an exhaustive and synthetic view of the reasons why the system is not diagnosable. Finally, the presented algorithm is scalable in practice: it provides an approximate and useful solution if the computational resources are not sufficient.
Marie-Odile Cordier, Yannick Pencolé, Louise Travé-Massuyès, Thierry Vidal
Self-Healability = Diagnosability + Repairability
18th International Workshop on Principles of Diagnosis 251-258 Nashville, Tennessee United States May 2007
Real-life complex systems are often required to have a high level of autonomy, even in faulty situations. The diagnosability analysis is the a priori study of the capability of a system to be self-aware about its current state by analysing the observations received by the sensors. The repairability analysis is the a priori study of the capability of a system to react to faults by applying repair actions. Even though they are strongly related, these properties are generally analysed independently. This paper builds upon the state of the art in both domains and proposes a joint property, called self-healability, which achieves a bridge between diagnosability and repairability. We first revisit and extend the classical definitions of diagnosability and repairability. Then, we give our definition of self-healability illustrated with several examples.
Anika Schumann, Yannick Pencolé, Sylvie Thiébaux
A Spectrum of Symbolic On-line Diagnosis Approaches
Twenty-Second Conference on Artificial Intelligence 335-340 Vancouver Canada 2007
This paper deals with the monitoring and diagnosis of large discrete-event systems. The problem is to determine, on-line, all faults and states that explain the flow of observations. Model-based diagnosis approaches that first compile the diagnosis information off-line suffer from space explosion, and those that operate on-line without any prior compilation have poor time performance. Our contribution is a broader spectrum of approaches that suits applications with diverse time and space requirements. Approaches on this spectrum differ in the amount of reasoning and compilation performed off-line and therefore in the way they resolve the tradeoff be-
tween the space occupied by the compiled information and the time taken to produce a diagnosis. We tackle the space and time complexity of diagnosis by encoding all approaches in a symbolic framework based on binary decision diagrams.
This allows for the compact representation of the compiled diagnosis information, and for its handling across many states at once rather than for each state individually. Our experiments demonstrate the diversity and scalability of our symbolic methods spectrum, as well as its superiority over the corresponding enumerative implementations.
2006
Yannick Pencolé, Anika Schumann, Dmitry Kamenetsky
Towards low-cost fault diagnosis in large component-based systems
6th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes 1473-1478 Beijing China August 2006
We address the problem of fault diagnosis in discrete-event systems. Our contribution is the development of a set of specialised diagnosers whose computation is much more realistic than that of the classical diagnoser. A specialised diagnoser is devoted to the diagnosis of one particular type of fault and is based on the observation of only a subpart of the system.
Alban Grastien, Yannick Pencolé
Model of reconfigurable discrete event system
6881 November 2006
Khalil Drira, Karim Guennoun, Francisco Moo-Mena, Yannick Pencolé, Xavier Pucel, Audine Subias, Louise Travé-Massuyès
Requirements, application scenarios, overall architecture, and test/validation specification, common working environment and standards at Milestone M1
6879 March 2006
Louise Travé-Massuyès, Audine Subias, Yannick Pencolé, Xavier Pucel
Characterization of diagnosis and repair for web services
6871 June 2006
Louise Travé-Massuyès, Audine Subias, Yannick Pencolé, Xavier Pucel
Specification of diagnosis algorithms for web services. Phase 1
6872 September 2006
Anika Schumann, Yannick Pencolé
Efficient on-line failure identification for discrete-event systems
6th IFAC Symposium on Fault Detection, Supervision and Safety of Technical Processes 1294-1299 Beijing China August 2006
The paper addresses the problem of diagnosing complex embedded systems. Given a flow of observations from the system, the goal is to explain these observations by identifying possible failures. Approaches that
efficiently compute the possible failures, like the classical diagnoser approach, suffer from high space complexity. This paper presents a method that generalises the observable behaviour of a system resulting in a smaller finite state machine, that efficiently maps observations to possible failures.
2005
Yannick Pencolé
Assistance for the design of a diagnosable component-based system
17th IEEE International Conference on Tools with Artificial Intelligence 549-556 Hong-Kong Hong Kong November 2005
Diagnosability of component-based systems is a property that characterises the ability to diagnose fault events given a flow of observations. In this paper, we use model-based reasoning techniques and we propose a theoretical framework to analyse diagnosability in a decentralised way. We then introduce an algorithm that performs diagnosability analyses and provides useful information for the design of a diagnosable component-based system.
Yuhong Yan, Marie-Odile Cordier, Yannick Pencolé, Alban Grastien
Monitoring Web Service Networks in a Model-based Approach
Third European Conference on Web Services 192-203 Växjö Sweden November 2005
The goal of Web service effort is to achieve universal interoperability between applications by using Web standards: this emergent technology is a promising way to integrate business applications. A business process can then be seen as a set of Web services that could belong to different companies and interact with each other by sending messages. In that context, neither a global model nor a global mechanism are available to monitor and trace faults when the business process fails. In this paper, we address this issue and propose to use model-based reasoning approaches on Discrete-Event Systems (DES). This paper presents an automatic method to model Web service behaviors and their interactions as a set of synchronized discrete-event systems. This modeling is the first step before tracing the evolution of the business process and diagnosing business process faults.
Yannick Pencolé, Marie-Odile Cordier
A formal framework for the decentralised diagnosis of large scale discrete event systems and its application to telecommunication networks
Artificial Intelligence 164(2) 121-170 May 2005
We address the problem of diagnosing large discrete event systems. Given a flow of observations from the system, the goal is to explain these observations on-line by identifying and localising possible failures and their consequences across the system. Model-based diagnosis approaches deal with this problem but, apart very recent proposals, either they require the computation of a global model
of the system which is not possible with large discrete event systems, or they cannot perform online diagnosis. The contribution of this paper is the description and the implementation of a formal
framework for the on-line decentralised diagnosis of such systems, framework which is based on the “divide and conquer” principle and does not require the global model computation. This paper finally
describes the use of this framework in the monitoring of a real telecommunication network.
2004
Yannick Pencolé
Diagnosability analysis of distributed discrete event systems
16th European Conference on Artificial Intelligence 43-47 Valencia Spain August 2004
This paper addresses the diagnosability problem of distributed discrete event systems. Until now, the problem of diagnosability has always been solved by considering centralised approaches, monolithic systems. These approaches do not use the fact that real monitored systems are generally modelled in a distributed manner. In this paper, we propose a framework for the diagnosability analysis of such systems: we study the diagnosability of the system using the fact that it is based on a set of communicating components. In the case where the system is not diagnosable, we also want to provide more accurate information in order to better understand the causes.
Anika Schumann, Yannick Pencolé, Sylvie Thiébaux
Diagnosis of discrete-event systems using bdds
15th International Workshop on Principles of Diagnosis 197-202 Carcassonne France June 2004
We improve the efficiency of Sampath’s diagnoser approach by exploiting compact symbolic representations of the system and diagnoser in terms of binary decision diagrams. We present an algorithm for synthesising the symbolic diagnoser with promising results on test cases derived from a telecommunication application.
Anika Schumann, Yannick Pencolé, Sylvie Thiébaux
Symbolic models for diagnosing discrete-event systems
16th European Conference on Artificial Intelligence 1085-1086 Valencia Spain August 2004
We improve the efficiency of Sampath’s diagnoser approach by exploiting compact symbolic representations of the system and diagnoser in terms of BDDs. We show promising results on test cases derived from a telecommunication application.
Yannick Pencolé
Diagnosability analysis of distributed discrete event systems
15th International Workshop on Principles of Diagnosis 173-178 Carcassonne France June 2004
This paper addresses the diagnosability problem of distributed discrete event systems. Until now, the problem of diagnosability has always been solved by considering centralised approaches, monolithic systems. These approaches do not use the fact that real monitored systems are generally modelled in a distributed manner. In this paper, we propose a framework for the diagnosability analysis of such systems: we study the diagnosability of the system using the fact that it is based on a set of communicating components. In the case where the system is not diagnosable, we also want to provide more accurate information in order to better understand the causes.
2003
Marie-Odile Cordier, Alban Grastien, Christine Largouët, Yannick Pencolé
Efficient trajectories computing exploiting inversibility properties
14th International Workshop on Principles of Diagnosis 93-98 Washington, DC United States June 2003
A time-consuming problem encountered both in system diagnosis and planning is that of computing trajectories over a behavioral model. In order to improve the efficiency of this task, there is currently a great interest in using modelchecking techniques developed within the area of computer aided verification. In this paper, we propose to represent the system as automata and we define a property called inversibility. This property is used to improve the efficiency of the search algorithm computing trajectories. We present two study cases in diagnosis and planning domains where this approach gives satisfactory results.
Marie-Odile Cordier, Alban Grastien, Christine Largouët, Yannick Pencolé
Calcul de trajectoires utilisant les propriétés d'interversibilité
6ème Rencontres Jeunes Chercheurs en Intelligence Artificielle 15-28 Laval France July 2003
Le temps de calcul des trajectoires sur un modèle de comportement du système est un problème critique rencontré aussi bien en diagnostic qu'en planification. Dans le but d'améliorer l'efficacité de cette tâche, un intérêt croissant est porté aux techniques de model-checking développées dans le domaine de la vérification automatique. Dans cet article, nous proposons de représenter le système par un automate, et nous définissons une nouvelle propriété appelée interversibilité. Cette propriété est utilisée pour améliorer l'efficacité de l'algorithme de recherche calculant les trajectoires. Nous présentons deux exemples dans les domaines du diagnostic et de la planification où cette approche donne des résultats satisfaisants.
Yannick Pencolé
All from one, one for all, failure diagnosis of discrete event system using representatives
15th IEEE International Conference on Tools with Artificial Intelligence 208-212 Sacramento, CA United States November 2003
Failure diagnosis in large and complex systems is a critical and challenging task. In the realm of model based diagnosis on discrete event systems, computing a failure diagnosis means computing the set of system behaviours that could explain observations. Depending on the diagnosed system, such behaviours can be numerous, so that a problem of representing them is induced. The paper discusses about this problem and presents a way of representing a diagnosis by the use of a partial order reduction technique.
2002
Yannick Pencolé, Marie-Odile Cordier, Laurence Rozé
Incremental decentralized diagnosis approach for the supervision of a telecommunication network
41st IEEE Conference on Decision and Control 435-440 Las Vegas, NV United States December 2002
This paper extends the decentralized approach proposed in [5] for diagnosing discrete-event systems. It considers the incrementality issue which is a major one in the context of on-line diagnosis and proposes two solutions.
Yannick Pencolé, Marie-Odile Cordier, Laurence Rozé
A decentralized model-based diagnostic tool for complex systems
International Journal on Artificial Intelligence Tools 11(3) 327-346 September 2002
We address the problem of diagnosing complex discrete-event systems such as telecommunication networks. Given a flow of observations from the system, the goal is to explain those observations by identifying and localizing possible faults. Several model-based diagnosis approaches deal with this problem but they need the computation of a global model which is not feasible for complex systems like telecommunication networks. Our contribution is the proposal of a decentralized approach which permits to carry out an on-line diagnosis without computing the global model. This paper describes the implementation of a tool based on this approach. Given a decentralized model of the system and a flow of observations, the program analyzes the flow and computes the diagnosis in a decentralized way. The impact of the merging strategy on the global efficiency is demonstrated and illustrated by experimental results on a real application.
Armen Aghasaryan, Christophe Dousson, Éric Fabre, Aomar Osmani, Yannick Pencolé
Modeling Fault Propagation in telecommunications networks for Diagnosis Purposes
World Telecommunication Congress Paris France September 2002
This paper describes a formalism to model the behavior of telecommunications networks when a fault occurs and how the effects are propagated across equipment. The objective of such a formalism, which is derived from UML diagram sequences, is to ease the construction and the update of a model corresponding to the supervised network; this model can be used to simulate fault propagation in the network but also to process on-line diagnosis and determine primary causes of a set of observed alarms.
Yannick Pencolé, Marie-Odile Cordier, Laurence Rozé
Une stratégie efficace pour une approche décentralisée du diagnostic de systèmes complexes
13ème Congrès Francophone AFRIF-AFIA de Reconnaissance des Formes et Intelligence Artificielle 259-267 Angers France January 2002
Dans cet article, nous nous intéressons au problème du diagnostic de systèmes à événements discrets complexes tels que les réseaux de télécommunications. En raison de la taille importante de ce type de système, l'utilisation d'un modèle global est impossible. Nous proposons une approche en-ligne s'appuyant sur un modèle décentralisé et nous montrons comment préserver l'efficacité par l'utilisation d'une stratégie adéquate.
Yannick Pencolé
Diagnostic décentralisé de systèmes à évènements discrets : application aux réseaux de télécommunications
June 2002
Le cadre de cette thèse est la surveillance et le diagnostic de systèmes dynamiques complexes tels que les réseaux de télécommunications. Ces systèmes sont composés d'un ensemble d'équipements interconnectés. Des mécanismes liés à des capteurs permettent à un superviseur de recevoir les alarmes émises par tous les composants du réseau et de les interpréter.
L'objectif de ces travaux est de fournir une aide à l'interprétation de ces alarmes afin d'avoir à tout moment une vision de l'état du réseau et de ses dysfonctionnements possibles. L'approche développée est issue des techniques de diagnostic à base de modèles. Elle consiste, à partir d'un modèle du fonctionnement et de dysfonctionnement des composants du réseau, à utiliser efficacement ce modèle pour analyser en-ligne le flux d'alarmes. Dans le cadre de la supervision de tels systèmes, le diagnostic consiste non seulement à établir à partir des observations les états possibles du système à un instant donné mais aussi la propagation des pannes. Nous proposons de représenter ces diagnostics sous forme de systèmes de transitions compacts (transducteurs réduits) basés sur des événements de pannes.
Étant données la complexité et la nature répartie de ces systèmes, nous avons concentré notre étude sur l'élaboration d'une approche décentralisée de diagnostic, fondée sur le principe de "diviser pour régner". Dans un premier temps, nous établissons un ensemble de diagnostics locaux fondés sur des modèles de comportements locaux. Afin d'assurer l'efficacité de ces calculs, nous nous appuyons sur l'approche proposée par M. Sampath et al. qui consiste à construire hors-ligne une structure de données appelée "diagnostiqueur" qui rend le suivi en-ligne et la production d'un diagnostic local possible.
Dans un deuxième temps, l'obtention du diagnostic du système est établi par fusion des diagnostics locaux. Cette fusion est nécessaire car elle permet de valider ou d'invalider les hypothèses locales de diagnostic. Une stratégie de fusion a été mise en place afin d'assurer l'efficacité de cette fusion.
Cette thèse a été effectuée dans le cadre d'un projet RNRT : le projet Magda. Elle a abouti au développement d'une plate-forme pour le diagnostic décentralisé de systèmes dynamiques. Cette plate-forme nous a permis de valider notre approche sur deux types de réseaux : le réseau Transpac et un réseau SDH.
2001
Albert Benveniste, Marie-Odile Cordier, Éric Fabre, Claude Jard, Yannick Pencolé, Laurence Rozé
Diagnostic distribué : Formalismes et algorithmes
2001
Dans ce rapport, nous introduisons le formalisme de description du modèle de réseau SDH (Hiérarchie de Données Synchrones). Ce formalisme est un système de transitions élémentaires appelées aussi pièces. À partir de ce formalisme, nous décrivons les approches proposées par les différents partenaires pour résoudre le problème du diagnostic distribué dans le cadre du réseau SDH étudié.
Marie-Odile Cordier, Philippe Dague, Francine Krief, François Lévy, Aomar Osmani, Yannick Pencolé, Laurence Rozé
Modélisation du réseau transpac et algorithmes de simulation
2001
Dans ce rapport, nous étudions la modélisation du réseau TRANSPAC à l’aide d’automates communicants. À partir de ce formalisme, nous générons un simulateur de pannes qui répond au critère des experts en défaillance de ce réseau. Ce simulateur est ensuite utilisé pour générer des scénarios en simulant des pannes simples et multiples. Les données observables ainsi générées et étiquetées sont la base pour l’apprentissage de scénarios à l’aide de chroniques.
Yannick Pencolé, Marie-Odile Cordier, Laurence Rozé
Incremental decentralized diagnosis approach for the supervision of a telecommunication network
12th International Workshop on Principles of Diagnosis 151-158 Via Lattea Italy March 2001
We address the problem of failure diagnosis in discrete-event systems such as telecommunication networks. We propose to extend the decentralized diagnosis approach proposed in (Pencolé 2000) in order to use it in an incremental way. The incremental approach is needed in order to provide on-line diagnosis and assist supervision operators. The diffculties about the incremental decentralized diagnosis approach are analyzed. Two solutions are proposed and discussed. This incremental approach has been experimented on telecommunication network.
Yannick Pencolé, Marie-Odile Cordier, Laurence Rozé
A decentralized model-based diagnostic tool for complex systems
13th International Conference on Tools with Artificial Intelligence 95-102 Dallas, TX United States November 2001
We address the problem of diagnosing complex discrete-event systems such as telecommunication networks. Given a flow of observations from the system, the goal is to explain those observations by identifying and localizing possible faults. Several model-based diagnosis approaches deal with this problem but they need the computation of a global model which is not feasible for complex systems like telecommunication networks. Our contribution is the proposal of a decentralized approach which permits to carry out an on-line diagnosis without computing the global model. This paper describes the implementation of a tool based on this approach. Given a decentralized model of the system and a flow of observations, the program analyzes the flow and computes the diagnosis in a decentralized way. We also present experimental results based on a real system.
2000
Yannick Pencolé
Decentralized diagnoser approach: application to telecommunication networks.
11th International Workshop on Principles of Diagnosis 185-192 Morelia Mexico 2000
This paper presents a general method for the diagnosis of large systems, such as telecommunication networks. Because of the size of the system, the model we use is
decentralized. In order to increase the efficiency of the diagnosis, the method combines two basic techniques of diagnosis: diagnosers and simulation-based techniques.
We propose the construction of diagnosers based on local behaviors to compute local diagnoses. Then, we propose a coordination of local diagnoses based on a
strategy which minimizes the computation for the coordination.
Yannick Pencolé
Approche diagnostiqueur décentralisé : application aux réseaux de télécommunications
5èmes Rencontres nationales des Jeunes Chercheurs en Intelligence Artificielle 309-322 Lyon France September 2000
Cet article décrit une méthode de diagnostic pour des systèmes de grande taille tels que les réseaux de télécommunication. Cette méthode utilise des techniques à base de modèle. La taille du système étant importnate, nous ne pouvons pas fonder le calcul d'un diagnostic sur la connaissance d'un modèle global du système. Le modèle utilisé est donc décentralisé. Nous proposons de construire des diagnostiqueurs fondés sur des comportements locaux dont le but est de déterminer des diagnostics locaux. Ensuite, un système de coordination a pour but de réunir les différents diagnostics locaux et de produire un diagnostic global du système. Cette coordination est effectuée à l'aide d'une stratégie de reconstruction qui minimise le calcul du diagnostic global.
Yannick Pencolé
Dyp : a centralized diagnoser approach demonstration software (In Newletters, CDROM).
September 2000