Directional expansion‐based fault diagnosis algorithm using orthotopic and ellipsoidal filtering

Abstract

For solving the problem of fault diagnosis for systems with unknown but bounded noises, a directional expansion-based fault diagnosis algorithm using orthotopic and ellipsoidal filtering is proposed in this study. In this fault diagnosis process, whether the system is faulty or not is checked by detecting whether the ellipsoidal feasible set is empty. When the ellipsoidal feasible set is empty, the specific faulty components are isolated based on the empty state of the test ellipsoidal sets. The orthotopic set is expanded in the faulty directions alone to minimise the set containing the fault parameter vector. Finally, comparative analyses of the proposed fault diagnosis algorithm and global expansion-based fault diagnosis algorithm using the orthotopic and ellipsoidal filtering in both numerical and case simulations prove the effectiveness and practicability of the proposed algorithm and show the advantages of this algorithm in view of convergence rate and fault identification speed.