Unknown input observer based fault class isolation and estimation

The advance of observer-based fault diagnosis is outlined and the idea of fault class isolation (FCI) is introduced and studied. According to the requirements of some missile actuator test, a new scheme to implement FCI and fault estimation between motor and sensor is proposed based on unknown input observer (UIO), whose procedure completes in three steps. Firstly, the open loop model from motor to sensor is constructed and its parameters are estimated via least squares principle. Secondly, the design procedure of the UIO is deduced and elaborated, and two principles for FCI and fault estimation between motor and sensor are presented. Finally, wavelet filtered is introduced and studied so as to de-noise the estimated faults. Simulation demonstrated that the scheme could effectively isolate constant fault(s) as well as sudden fault(s) between motor and sensor while the fault(s) could be estimated precisely. The study indicates that the presented scheme is feasible and appropriate for actuator fault diagnosis.