Online Actuator Failure Detection using Direct Approach

A novel actuator failure detection algorithm is developed in this paper. An actuator failure is considered to occur when it produces an input to the structure that is dierent from the commanded input. In this paper, an error function, one for each actuator, is developed to monitor the working status of the examined actuator in real-time, regardless the status of other actuators. Non-zero signal profile in the error function indicates the time instant of failure of the examined actuator (for measurement noise free case). The coecients of the error function are calculated directly from the healthy input data (from the examined actuator) and all outputs without having to identify the state-space model of the system. Thus the need to know the state-space model of the plant is bypassed in the presented direct approach. Experimental results from a NASA eight bay truss show that the direct method can successfully isolate and identify the failure of the examined actuator in real-time.