Residual Fault Detection Filter Design in the Finite Frequency Domain Based on LMI Optimization

In the observer based fault detection methods, the parameter be designed and optimized is the observer gain matrix. However, an observer cannot be optimized to achieve all the expected performances. A novel fault detection filter designed method is presented in this paper, where the filter is cascaded after the observer based fault detection system. The filter is designed to be sensitive to the faults but robust to the disturbances and model uncertainties in the finite frequency domain. For the proposed filter, there are four matrices need to be optimized and the performance would be enhanced significantly. The effectiveness of the proposed method is demonstrated by applying it to a railway suspension system.