Actuator Failure Identification and Compensation Through Sliding Modes

Abstract

The actuator failure compensation problem is addressed in this brief. It is considered an uncertain linear plant, which is supposed to undergo unknown failures causing the plant input components to be stuck at some uncertain but bounded time functions. A sliding-mode-based control policy is presented, guaranteeing the detection of the fault and the identification of the failed component by means of a suitable test input. Once the failed component has been identified, the control law is reconfigured, redistributing the control activity among the controllers still working. The proposed controller has been tested by simulation on a benchmark problem