Fault detection and isolation of sequential multiples faults in dynamic linear systems

This paper presents an adaptive generalized likelihood ratio (GLR) test for multiple faults detection and isolation in dynamic stochastic linear systems. For sequential fault detection, the adaptive GLR test confronts the fault hypothesis Hkhj to the reference hypothesis Hk updating on line after each detection and isolation of one fault. The proposed scheme is based on the design of the fault detection filter and on the GLR test applied on the filter innovation.