Event-Triggered Reliability Control for Fault Tolerant Tracking of Nonlinear Systems via Adaptive Dynamic Programming

This study proposes a novel methodology driven by adaptive dynamic programming (ADP) to achieve optimal fault-tolerant tracking control (FTTC) in affine nonlinear systems (ANSs) subject to actuator failures. To begin with, a key innovation lies in constructing an extended system model that integrates tracking error dynamics and desired trajectory, establishing a basis for control synthesis. For this extended system, a critic neural network (NN) with adaptive weights is designed to approximate the discounted value function, enabling online optimization of control policies. Additionally, to address practical network constraints, a bandwidth-efficient event-triggered scheme (BEETS) is introduced to optimize limited bandwidth resources in networked control setups. Meanwhile, an NN based fault observer realizes real-time identification and dynamic compensation of actuator faults, mitigating their adverse effects on system performance. The resulting FTTC framework, which merges ADP-based optimal control with a fault compensator, ensures robust fault tolerance under both constant and time varying fault scenarios. Through rigorous Lyapunov stability assessment, uniform ultimate boundedness (UUB) is proven to be achieved for the closed-loop nonlinear system, guaranteeing stable tracking performance. The validity of this algorithm is verified through rigorous theoretical analysis and captivating numerical simulations.