An efficient method for determining fault isolability properties based on an augmented system model

摘要

This paper presents a novel approach based on an augmented system model to rapidly determine the fault isolability properties. In conventional structural analysis approaches, fault variables are not classified for fault isolation. That usually causes large amounts of computation. In our work, through only one decomposition of an augmented system model, the computational dependencies between the fault variables are disclosed. Then, some fault isolability properties are directly extracted, and the corresponding fault equations are overall classified. That significantly reduces the search space for non-isolable faults. Later, using the structural properties of the monitored system model, the rest of the fault isolability properties are derived. Compared with previous approaches, our approach does not need to build the fault signature matrix or evaluate all pairs of faults for fault isolation. A new algorithm based on the approach is proposed and the simulations of a four-tank system indicate its validity and efficiency. (C) 2020 European Control Association. Published by Elsevier Ltd. All rights reserved.