Abstract

A framework for on-line passive fault diagnosis in hierarchical discrete-event systems is proposed. In this approach, the system model is broken to simpler substructures called D-holons. A state based diagnoser is constructed for each D-holon. Fault diagnosis is accomplished using the state estimates provided by the D-holon diagnosers. The diagnosers may communicate among each other in order to update their state estimates. At any given time, only a subset of the diagnosers are active, and as a result, instead of the entire model of the system, only the models of the D-holons associated with the active diagnosers are used. Therefore, only part of the system model needs to be stored in computer Random Access Memory (RAM). This reduces RAM requirements and thus, could be useful in complex multi-phase systems. The concept of D-holon provides a suitable tool to study failure diagnosability in cases where components are active in certain phases of operation and inactive in other phases. This resulted in the introduction of the concept of phase-diagnosability. A set of necessary and sufficient conditions for phase-diagnosability is obtained. Furthermore, in order to reduce the computational complexity of the diagnosis process, a set of sufficient conditions is provided under which the diagnosis process becomes semi-modular. It is shown that the computational complexity of constructing (time) and storing (space) the transition systems required for diagnosis in the proposed semi-modular approach is polynomial in the number of system components, whereas in the original monolithic approach the computational complexity is exponential.