Fault actuators detection based on inputs-outputs data mapping: application to a robot manipulator

Abstract

This paper deals with a fault estimation algorithm to reconstruct faulty inputs of a process under actuators faults. This approach is based on a generic model with outputs/inputs data mapping to reconstruct the faulty inputs. The key idea consists on dynamic parameters identification of this generic model in real time for abrupt and intermittent faults. To apply this approach, these dynamic parameters are estimated basing on the recursive least squares solution obtained from the sampling of the dynamic model along a closed loop tracking trajectory. Indeed, the control loop requires information about the state derivative of the outputs system. For this aim, a dynamic gains sliding mode differentiator is introduced. By using measured signals, the proposed approach is validated on a bi-axis robot manipulator to evaluate its usefulness on actuators faults detection.