Kalman filter design for control surface failure detection and isolation (flight control)

Abstract

The use of Kalman filters for residual generation for control surface failure detection and isolation (FDI) is studied. The FDI algorithm investigated consists of two main components: a Kalman filter for residual generation and a set of pairwise log-likelihood-ratio (LLR) tests for decision processing. The performance of several Kalman filter designs under a severe gust level of 20-fps RMS is evaluated. The performance of the FDI algorithm is measured in terms of the probabilities of detection and false alarm which are closely related to the signal-to-noise ratios (SNRs) of the decision LLRs. The filter design process is aided by evaluating the SNRs of the different Kalman filter configurations. It was found that including the gust-shaping filters in the Kalman filter equations improved the FDI performance under severe wind turbulence. However, all the filter configurations evaluated were sensitive to parameter variations. >