PROJECT TITLE :

Controller Integrity Monitoring in Adaptive Learning Systems Towards Trusted Autonomy

ABSTRACT:

This paper presents a controller integrity monitoring methodology for a class of second-order nonlinear uncertain systems incorporating neural network-primarily based adaptive control algorithms. The adaptive neural network model is utilized to make sure strong tracking performance in the presence of bound modeling uncertainty underneath thought throughout the offline controller design method. Primarily based on Lyapunov stability analysis, an online controller integrity monitoring scheme is developed to detect the prevalence of controller software/algorithm faults and unanticipated physical element faults, which could cause unstable learning behaviors and malfunctions of the adaptive controller. Adaptive thresholds for detecting controller malfunctions are derived, guaranteeing the robustness with respect to modeling uncertainty and neural network approximation error. Additionally, the detectability conditions are investigated, characterizing the category of detectable software faults and unanticipated physical faults. An higher certain on the fault detection time is additionally established. Simulation results are shown to illustrate the effectiveness of the proposed technique.