Actuator fault diagnosis for electronic-control systems based on multiple-model structure

In this paper, the diagnosis problem of actuator lock-in-place for electronic-control systems is studied. Based on multiple-model structure, an adaptive unknown input observer (UIO) approach was developed to detect actuator faults. For a model with uncertain parameters, we developed an adaptive approach to estimate the unknown parameters on-line. According to lyapunov theory, the proposed adaptive algorithms guarantee the residual signals converge when a model matches the system. Simulation results on an automobile engines example show satisfactory performance of the algorithm.