Control and diagnosis of real-time systems under finite-precision measurement of time
Is Version Of
A discrete event system (DES) is an event-driven system that evolves according to abrupt occurrences of discrete changes (events). The domain of such systems encompasses aspects of many man-made systems such as manufacturing systems, telephone networks, communication protocols, traffic systems, embedded software, asynchronous hardware, robotics, etc.
Supervisory control theory for DESs studies the existence and synthesis of the supervisory controllers, namely, supervisors that restrict the system behaviors by dynamically disabling certain controllable events so that the controlled close-loop system could behave as desired. Extensive work on supervisory control of untimed DESs exists and the extension to the timed setting has been reported in the literature. In this dissertation, we study the supervisory control of dense-time DESs in which the digital-clocks of finite-precision are employed to observe the event occurrence times, thereby relaxing the assumption of the prior works that time can be measured precisely. In our setting, the passing of time is measured using the number of ticks generated by a digital-clock and we allow the plant events and digital-clock ticks to occur concurrently. We formalize the notion of a control policy that issues the control actions based on the observations of events and their occurrence times as measured using a
digital-clock, and show that such a control policy can be equivalently represented as a "digitalized"-automaton, namely, an untimed-automaton that evolves over the events (of the plant) and ticks (of the digital-clock). We introduce the notion of observability with respect to the partial observations of time resulting from the use of a digital-clock, and
show that this property together with controllability serves as a necessary and sufficient condition for the existence of a supervisor to enforce a real-time specification on a dense-time discrete event plant. The observability condition presented in the dissertation is very different from the one arising due to a partial observation of events since a partial
observation of time is in general nondeterministic (the number of ticks generated in any time interval can vary from execution to execution of a digital-clock). We also present a method to verify the proposed observability and controllability conditions, and an algorithm to
compute a supervisor when such conditions are satisfied. Furthermore we examine the lattice structure of a class of timing-mask observable languages, and show that the proposed observability is not preserved under intersection but preserved under union.
Fault diagnosis for DESs is to detect the occurrence of a fault so as to enable any corrective actions. It is crucial in automatic control of large complex man-made systems and has attracted considerable attention in the literature of reliability engineering, control and computer science.
For the event-driven systems with timing-requirements such as manufacturing systems, communication networks, real-time scheduling and traffic systems, fault diagnosis involves detecting the timing-faults, besides the sequence-faults. This requires monitoring timing and sequence of events, both of which may only be partially observed in practice. In this dissertation, we extend the prior works on fault diagnosis of timed DESs by allowing time to be partially observed using a digital-clock which measures the advancement of time with finite precision by the number of ticks. For the diagnosis purposes, the set of nonfaulty timed-traces is specified as another timed-automaton that is deterministic. We show that the set of timed-traces observed using a digital-clock with finite precision is regular, i.e., can be represented using a finite (untimed) automaton. We also show that the verification of
diagnosability (the ability to detect the execution of a faulty timed-trace within a bounded time delay) as well as the off-line synthesis of a diagnoser are decidable by reducing these problems to the untimed
setting. The reduction to the untimed setting also suggests an effective method for the off-line computation of a diagnoser as well as its on-line implementation for diagnosis. The aforementioned results are further extended to the nondeterministic setting, i.e., diagnosis of dense-time DESs using digital-clocks under nondeterministic event observation mask. We introduce the notion of "lifting" (associating each event with each of its nondeterministic observations), and show that diagnosis of dense-time DESs employing digital-clocks to observe event occurrence times under nondeterministic event observation mask can be reduced to that of the deterministic setting, i.e., diagnosis of the lifted dense-time DESs under the deterministic lifted event observation mask, and hence can be further reduced to diagnosis of the untimed setting.