Model based multiple faults detection and isolation for electro-hydrostatic actuator

In this paper, electro-hydrostatic actuator (EHA) model is presented in behavioral form using bond graph. Bond graph methodology uses physical knowledge of the system to obtain the model which results in precise modeling and hence improve the monitoring and consequently ensure better diagnosis of the online system. The subject system model is evaluated through powerful software for any possible error. Real time values for the EHA have been used and the fault free behavior of the system is then checked in the presence of external load at output shaft of EHA. Furthermore, possible multiple faults have been incorporated into the system and examined by fault monitoring algorithms. Diagnostic model of the actuator is obtained through causality inversion method, the constraint relations are derived and fault signature matrix (FSM) is finally developed. The constraint relations are dependent on the model structure and number of causality inversion junctions. The behavioral and diagnostic models are coupled for simulation purposes. The hydraulic pump internal leakage, hydraulic lines pressure loss and hydraulic jack internal leakage faults are induced individually and also collectively at different intervals of time and simulated in the actual system model and consequently detected and isolated successfully by this methodology.