Model based fault tolerant control using the marginalized likelihood ratio test

A new methodology is presented in this paper which incorporates the marginalized likelihood ratio (MLR) test for online fault detection and isolation. The proposed methodology reduces the number of optimization problems required for isolating the fault by means of a simple integration scheme. Moreover, the dependency on the accuracy of the statistical fault detection and confirmation tests is relaxed using this technique. The proposed method is capable of detecting and isolating the fault simultaneously and hence significantly reduces the computational effort.