Multi-Model Based Fault Detection and Diagnosis of a Hydraulic Servo Axis

Abstract: This contribution describes the design and implementation of a multi-model based fault detection and diagnosis system for a hydraulic servo axis. Multiple models are operated in parallel to the plant similar to a bank of observers. Beside one model mimicking the fault free process, there are additional models describing the process behavior in the presence of both actuator and process faults. For each model the squared error between the model output and the plant output is determined. The model which (over a period of time) shows the best match with the process is assumed to currently best describe the process behavior. The fault injected into the model with the best match is thus at the same time the diagnosis, i.e. the fault which is assumed to be present in the real system. This system has been tested exhaustively at a testbed. The result of the tests is shown in the accompanying video.