An Alternative Approach to Implementation of the Generalized Likelihood Ratio Test for Fault Detection and Isolation

In this study an alternative approach to implementation of the generalized likelihood ratio (GLR) test for, detection and isolation of the fault in linear systems is proposed. The proposed approach offers the following advantages: 1) It overcomes the Shortcomings of the previously suggested methods by accurately detecting the time of occurrence of the fault; 2) It uses statistical fault detection and confirmation tests to obtain a crude estimate of time of occurrence of the fault (TOF) and then refines the estimated TOF using an extended data window and the GLR test; and 3) It avoids performing the isolation in case the number of data points is not enough and hence the number of misclassifications is significantly reduced. The newly proposed method is evaluated by application to a benchmark CSTR problem using Monte Carlo simulations, and the results. reveal that this method can estimate the time of occurrence of the fault and the fault magnitude more accurately compared to a previous approach applied to the same benchmark problem. Simulation results on a benchmark problem also show significantly. lower misclassification rates.