Special section on ``Applications of probabilistic graphical models in dependability, diagnosis and prognosis''
Résumé
In order to manage and to maintain in operating condition industrial artefacts during the lifecycle, engineers usually produce models for the prediction or the prognosis of functioning or dysfunctioning states of the system. These models are extracted from knowledge of the systems and depend on the objective of the model. In dependability analysis, models may serve to several aims:-To assess the impact of maintenance activities on the ability to maintain operational conditions and to aid decisions during maintenance.-To assess the impact of control and pilotage activities on wear, degradations (faults) or failures prognosis of all or part of the system; thus, they satisfy the main goals as service quality or low-risk situations for users, staff, environment, etc.-To assess the efficiency of means to warrant an acceptable level of risk, whatever the operational constraints and environmental perturbations. There are several domains of the application of dependability. These domains can be associated with different aspects of the system behavior such as management modes, governance, human factors, extreme events or rare events and their consequence over society, maintenance, control and supervision or risk reductions of socio-technical systems, etc. Nevertheless, most engineers have neither the tools nor the methods to effectively understand the whole set of information (knowledge and evidence) according to the operational constraints and disturbances that condition the functioning of complex socio-technical systems. The phenomena encountered are so complex as a result of their heterogeneity and the number of nested mechanisms of different natures that it is often quite difficult to reach the required objectives or levels of performance. Moreover, there is usually no exact analytical model able to describe the whole phenomenon encountered. It is also impossible to know exactly all the states of the system and to observe all the component states at each point in time, in order to determine the optimal decision. The engineers should bear in mind that all models are biased and partial. As a result, engineers need new methods to solve these modeling problems. Nowadays, it is necessary to model systems and components with a finite but unbounded set of states or performance levels i.e. systems with multiple states. In addition, the component behaviors are conditioned by the operational constraints and environmental disturbances of the system. In such cases, dependability assessment becomes difficult because it should take into account the combining effects of dependent failures due to constraints, disturbances and the intrinsic multi-state nature of system components. All of this results in an increasing number of scenarios to model and evaluate. The consequence is a cumbersome activity for the analyst, often enforcing bias and partiality. However, quantitative assessment is necessary to warrant the viability of systems and their performance regarding risk, dependability or control. It is thus necessary to handle an uncertain representation of the system in order to describe its working and/or faulty behavior. This imperfect perception goes toward a probabilistic view of system states. The main difficulties are the integration of a huge quantity of information to model industrial or socio-technical systems subject to a large set of interactions with their environment. In order to contribute to solving this modeling problem, this section focuses on the use of graph theoretical and probabilistic approaches based on Probabilistic Graphical Models (PGM), in maintenance, in risk analysis and management, and in control theory as well.
Domaines
Automatique / Robotique
Origine : Fichiers produits par l'(les) auteur(s)
Loading...