Fault detection and isolation is a crucial challenging task in the automatic control of large complex systems. We propose discrete-event system (DES) approach to problem failure diagnosis. introduce two related notions diagnosability DES's framework formal languages compare with observability invertibility. present systematic procedure for events using diagnosers provide necessary sufficient conditions language be diagnosable. The diagnoser performs diagnostics online observations behavior;...

Detection and isolation of failures in large, complex systems is a crucial challenging task. The increasingly stringent requirements on performance reliability technological have necessitated the development sophisticated systematic methods for timely accurate diagnosis system failures. We propose discrete-event (DES) approach to failure problem. This applicable that fall naturally class DES; moreover, purpose diagnosis, continuous-variable dynamic can often be viewed as DES at higher level...

The problem of verifying the properties diagnosability and I-diagnosability is considered. We present new polynomial-time algorithms for deciding I-diagnosability. These are based on construction a nondeterministic automaton called verifier.

The need for accurate and timely diagnosis of system failures the advantages automated diagnostic systems are well appreciated. However, diagnosability considerations often not explicitly taken into account in design. In particular, design controller that subsystem decoupled, this may significantly affect properties a system. authors present an integrated approach to control diagnosis. More specifically, they diagnosable by appropriate controller. This problem, which refer as active is...

A supervisory control scheme which is based on limited lookahead policies proposed, using an online where after the occurrence of event, next action determined basis N-step-ahead projection behavior process (represented as N-level tree). This procedure then repeats execution event. Different attitudes can be adopted to calculate a for given tree in order resolve uncertainties about beyond N steps. Two such attitudes, termed conservative and optimistic. are studied. Results presented...

The problem under consideration in this paper is that of enforcement by supervisory control a given property on partially-observed discrete-event system. We present general methodology applicable to large class properties previously studied (individually) the literature. These include, but are not restricted to, safety, diagnosability, opacity, detectability, anonymity and attractability. When system does satisfy considered property, objective synthesize supervisor restricts system's...

In this paper, we analyze the diagnosability properties of labeled Petri nets. We consider standard notion languages, requiring that every occurrence an unobservable fault event be eventually detected, as well stronger in <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> steps, where detection must occur within a fixed bound occurrences after fault. give necessary and sufficient conditions for these two notions both bounded unbounded nets then...

We present new results on the synthesis of safe, non-blocking, and maximally permissive supervisors for partially observed discrete event systems. consider case where legal language is a non-prefix-closed sublanguage system language, i.e., non-blockingness must be ensured in addition to safety. To solve this problem, we define bipartite transition system, called Non-blocking All Inclusive Controller (NB-AIC), that embeds all safe non-blocking supervisors. an algorithm construction NB-AIC...

We consider feedback control systems where sensor readings may be compromised by a malicious attacker intending on causing damage to the system. study this problem at supervisory layer of system, using discrete event techniques. assume that can edit outputs from sensors system before they reach controller. In context, we formulate synthesizing supervisor is robust against class attacks and present solution methodology for problem. This blends techniques games automata with imperfect...

The decentralized control problem for discrete-event systems addressed in this paper is that of several communicating supervisory controllers, each with different information, working concert to exactly achieve a given legal sublanguage the uncontrolled system's language model. A novel information structure model presented dealing class problems. Existence results are cases when controllers do and not anticipate future communications, synthesis procedure case communications. Several...

This paper studies online fault detection and isolation of modular dynamic systems modeled as sets place-bordered Petri nets. The common places among the set nets modeling a system capture coupling various components. transitions are labeled by events, some which unobservable (i.e., not directly recorded sensors attached to system). events whose occurrence must be diagnosed have transition labels. These model faults or other significant changes in state. existing theory diagnosis...