Analysis of transient data in test designs for active fault detection and identification

Active model-based Fault Detection and Isolation (FDI) methods are of increasing importance in modern cyber-physical systems due to their ability to generate detections as residuals between anticipated and observed information, and, thus, isolate and identify causes of faults with greater confidence than conventional passive FDI techniques. This work focuses on evaluating the effectiveness of active FDI tests which use either steady-state or dynamic information. It is shown that transient information from active FDI tests can improve the identifiability of faults compared to steady-state testing. These tests are designed by casting FDI as an optimization problem that maximizes the Fisher Information Matrix of a system sensed outputs with respect to faults. The identifiability of faults is examined at steady-state and transient FDI tests in a plate fin heat exchanger of an aircraft environmental control system (ECS). In this system particulate fouling needs to be detected, which is challenged by multiple sources of system uncertainty. It is shown that the inclusion of transient information during fault diagnosis increases the confidence in fault identification when using optimal test designs. (C) 2018 Elsevier Ltd. All rights reserved.