Aircraft Health Monitoring System Using Multiple-Model Adaptive Estimation

This paper proposes Multiple-Models Adaptive Estimation (MMAE) for failure detection and identification of aircraft components, i.e, flaps, landing gears. The MMAE FDI consists of parallel Kalman filters and each Kalman filter is constructed to represent a specific failure mode including the nominal mode. The Kalman filter residuals are post processed to produce the log-likelihood function values using sliding window methods, and posterior probabilities. The hypothesis with the maximum log-likelihood function values is declared the most possible mode of the system at the current decision time, and the probability-weighted average state estimate is calculated. This method is applied to the DC motor system, and evaluated the performance with sensors failures. Simulation results show that the MMAE is simple to implement and effective in fault detection and identification.