Incipient actuator fault identification with fuzzy relational model and adaptive rules

Abstract

In this paper, a novel algorithm is proposed to identify the incipient actuator fault with the fuzzy relational model and adaptive updating rules. Normally, it is difficult to model a nonlinear, information poor system with physical model or mathematical model. Hence a black-box with input/output measurements data is used to develop the system behaviour. A fuzzy relational model technique is adopted here via the global least square identification algorithm to identify the model performance. The fault is detected via the discrepancy between the system output and model output. With the aid of augmented error technique from model reference adaptive control, the fault can be reconstructed and fed back to the control input to compensate the fault influence on the model. An cooling-coil subsystem is used as an example to perform the experiments for the incipient actuator fault identification. Simulation experiments results have been verified the effectiveness of the proposed method.