Modeling and Fault Diagnosis Using Mixed Approach: Hybrid Dynamical Systems with Unknown Disturbances

Abstract

A hybrid dynamical system (HDS) results from the interaction between a continuous dynamics and a discrete. The continuous dynamics represents the system behaviour which may change from a functioning mode to another depending on the discrete dynamics. The HDS is often subject to disturbances arising from external phenomena due to the environment, or internal phenomena linked to changes in the system. This paper proposes a diagnosis approach of disturbed complex HDS by combining using State Observers and Timed Automata (TA). The proposed method combines the advantages of Observer and TA in order to obtain the best performance in the presence of disturbances, particularly in the fault location phase. Thus, for complex systems, the observer-based approach is used to generate fault indicators for fault detection and the fault isolation procedure consists in analysing the fault signatures matrix. The TA, which are commonly used in discrete event systems enables to take into account temporal aspects and to follow up the system dynamic evolution as well as fault propagation effects. To refine fault location, a diagnoser based on TA is built. The simulation results show the dynamic behaviour of the system variables with performance evaluation of the proposed fault diagnostic approach.