Safe Controllability using Online Prognosis

IEEE Transactions on Automatic Control, 2010

We study the prognosis of failures, i.e., their prediction prior to their occurrence, in discrete event systems in a decentralized setting where multiple prognosers use their local observations to issue local prognosis decisions. We define the notion of correctness of a decentralized set of prognosers in terms of "no missed detections" (each failure is prognosed prior to its occurrence) and "no false alarms" (an incorrect prognostic decision is never issued), and introduce the notion of coprognosability as an existence condition. When specialized to the centralized case (i.e., the case of a single prognoser), this condition turns out to be weaker than the one introduced by Genc and Lafortune in 2006 since a uniform bound on the number of steps within which a failure will occur is not required. For comparison, we also introduce the stronger notion of "uniformly bounded coprognosability" and identify the subclass of decentralized prognosers for which it serves as an existence condition. We show that the two notions coincide when the underlying system and its nonfailure specification possess finite-state representations, and present a verification algorithm whose complexity is polynomial in the sizes of the system being prognosed and its nonfailure specification, and is exponential in the number of the local prognosers. We also introduce the notion of reaction bound for coprognosis as the earliest time beyond a prognostic decision when a failure can occur, and present an algorithm for computing it. The complexity of this algorithm is identical to that of the verification algorithm. An algorithm with complexity linear in the size of the specification and the number of local prognosers is also presented for an online prognosis of failures. We show that the notions of coprognosability and its uniformly bounded version are in general incomparable with the notion of codiagnosability (that guarantees a uniformly bounded delay detection of a failure by a local diagnoser). When the system cannot execute an unbounded sequence of unobservable events, uniformly bounded coprognosability implies codiagnosability, whereas coprognosability and codiagnosability remain incomparable.

Proceedings of the 2001 American Control Conference. (Cat. No.01CH37148), 2001

Faults in automated processes will often cause undesired reactions and shut-down of a controlled plant, and the consequences could be damage to technical parts of the plant, to personnel or the environment.Faulttolerant control combines diagnosis with control methods to handle faults in an intelligent way. The aim is to prevent that simple faults develop into serious failure and hence increase plant availability and reduce the risk of safety hazards. Fault-tolerant control merges several disciplines into a common framework to achieve these goals. The desired features are obtained through on-line fault diagnosis, automatic condition assessment and calculation of appropriate remedial actions to avoid certain consequences of a fault. The envelope of the possible remedial actions is very wide. Sometimes, simple re-tuning can suffice. In other cases, accommodation of the fault could be achieved by replacing a measurement from a faulty sensor by an estimate. In yet other situations, complex reconfiguration or online controller redesign is required. This paper gives an overview of recent tools to analyze and explore structure and other fundamental properties of an automated system such that any inherent redundancy in the controlled process can be fully utilized to maintain availability, even though faults may occur. 1 .

2009 American Control Conference, 2009

We consider robust decentralized diagnosis of discrete event systems, where the goal is to detect the occurrence of unobservable fault events using a set of local diagnosers that are themselves subject to failures. We introduce a formal notion of robust decentralized diagnosability, called robust codiagnosability, and study its properties. Two different tests of robust codiagnosability are presented; one uses diagnoser automata and the other uses verifier automata. We also revisit the problem of centralized diagnosability and study the problem of diagnosability under partial observation, where the set of observable events is reduced; in this regard, we introduce the notions of partial diagnosers and indeterminate hidden cycles, which are subsequently used in the study of robust codiagnosability.

IEEE Transactions on Automatic Control, 1995

Abstruct-Fault detection and isolation is a crucial and challenging task in the automatic control of large complex systems. We propose a discrete-event system (DES) approach to the problem of failure diagnosis. We introduce two related notions of diagnosability of DES's in the framework of formal languages and compare diagnosability with the related notions of observability and invertibility. We present a systematic procedure for detection and isolation of failure events using diagnosers and provide necessary and sufficient conditions for a language to be diagnosable. The diagnoser performs diagnostics using online observations of the system behavior; it is also used to state and verify off-line the necessary and sufficient conditions for diagnosability. These conditions are stated on the diagnoser or variations thereof. The approach to failure diagnosis presented in this paper is applicable to systems that fall naturally in the class of DES's; moreover, for the purpose of diagnosis, most continuous variable dynamic systems can be viewed as DES's at a higher level of abstraction. In a companion paper [20], we provide a methodology for building DES models for the purpose of failure diagnosis and present applications of the theory developed in this paper.

European Conference on Artificial Intelligence, 2010

Diagnosability is the property of a given partially observable system model to always exhibit unambiguously a failure behavior from its only available observations in finite time after the fault occurrence, which is the basic question that underlies diagnosis taking into account its requirements at design stage. However, for the sake of simplicity, the previous works on diagnosability analysis of discrete event systems (DESs) have the same assumption that any observable event can be globally observed, which is at the price of privacy. In this paper, we first briefly describe cooperative diagnosis architecture for DESs with autonomous components, where any component can only observe its own observable events and thus keeps its internal structure private. And then a new definition of cooperative diagnosability is consequently proposed. At the same time, we present a formal framework for cooperative diagnosability checking, where global consistency of local diagnosability analysis can be achieved by analyzing communication compatibility between local twin plants without any synchronization. The formal algorithm with its discussion is provided as well. 2 PRELIMINARIES In this section, we first describe how to model DESs with autonomous components and then give some important concepts before proposing cooperative diagnosis architecture for such systems. 2.1 System model We consider a distributed DES composed of a set of autonomous components {G 1 , G 2 ,..., G n } that communicate with each other by communication events. Moreover, any component can only observe its own observable events and thus can keep its internal structure private. This kind of system is modeled by a set of FSMs with each one representing the local model of one component.

2007 European Control Conference (ECC), 2007

This paper reports a low-cost online fault detection approach for supervisory controllers in the framework of Supervisory Control Theory (SCT). For the cases when sensors dedicated to fault detection increase significantly the cost of controllers, or failure events are even impossible to detect by a direct way, methods based on the well-known watchdog structures are proposed. To successfully integrate watchdogs in the SCT framework, their discrete-event model is defined, and fault-detection techniques proposed in this paper are based on the extension of controller models previously designed using conventional supervisory synthesis methods. Fault-detection strategies are presented for centralized and distributed supervisory control environments, in the latter case providing solutions for avoiding problems according to fault propagation. Proposed techniques give full authority to the system designer in defining failure handling procedures and are proved not to influence the operation ...

HAL (Le Centre pour la Communication Scientifique Directe), 2014

The paper deals with an online safety mechanism to define interactions between a diagnoser and a control filter for fault tolerant control of manufacturing discrete systems. The diagnoser observes the plant behavior whereas the control filter ensures the safety from the controller. This online interaction is based by events communication where the control law is never reconfigured. The proposed approach is applied to CISPI platform from the CRAN laboratory (Research Center for Automatic Control of Nancy).

This article studies the supervisory control problem of discrete event systems (DES) with state-dependent controllability and observability. The new model is novel because the controllability and observability of an event is changeable in the lifetime of system evolution, and dependent on the system state. Two fundamental problems are discussed in detail with the new model: supervisor existence problem and supervisor synthesis problem. We derive a necessary and sufficient condition for the existence of the supervisor, and two algorithms are next developed to check the properties of the specification and to synthesize the required supervisors if such properties hold. Two examples are presented to illustrate how the new model is applied to practical computing. One example refers to state-dependent controllability, and it is introduced with the background of process control and ISR management in operating systems. The other example refers to state-dependent observability, and is presented with the background of flexible manufacturing systems.

2018 International Conference on Control, Automation and Diagnosis (ICCAD), 2018

This paper presents an approach of a safe control synthesis of Timed Discrete Event Systems, based on timed properties. This synthesis aims to realize a fault-tolerant control during the detection of sensor faults. To establish this synthesis, the proposed approach relies on the modularity of the manufacturing systems to avoid the combinatorial explosion recurrent to several approaches. An example of a manufacturing system illustrates our remarks.

2013 IEEE Conference on Computer Aided Control System Design (CACSD), 2013

This paper deals with the detection of (permanent) fault in the setting of stochastic discrete-event systems (DESs) under partial observability of events. Prior works have only studied the verification of the stochastic diagnosability (S-Diagnosability) property. To the best of our knowledge, this is a first paper that investigates the online detection schemes and also introduces the notions of their missed detections (MDs) and false alarms (FAs), and we establish that S-Diagnosability is a necessary and sufficient condition for achieving any desired levels of MD and FA rates. Next we provide a detection scheme, that can achieve the specified MD and FA rates, based on comparing a suitable detection statistic, that we define, with a suitable detection threshold, that we algorithmically compute. We also algorithmically compute the corresponding detection delay bound. The detection scheme also works for non-S-Diagnosable systems, with the difference that in this case only any FA rate can be met, and there exists a minimum MD rate that increases as FA rate is decreased.