A novel robust fault identification approach for a class of discrete systems

This paper addresses the problem of robust fault identification for a class of discrete systems with modelling uncertainties and unknown input disturbances and a novel robust fault identification approach is proposed. First, by introducing the Moore-Penrose generalized inverse matrix, the unknown input disturbances is obtained and substituted into the state equation of the systems such that a new description of the systems is obtained. Then, based on the new description, a state observer is constructed and the faults are identified by the least squares method based on the relationship between the faults and the state observation error, and the analysis for the identification error is also given. Finally, the robustness of the identification algorithm to the modelling uncertainties is enhanced by introducing the dead zone function