Neuro-adaptive optimized control for stochastic systems with state constraints: The non-affine faults case

This paper addresses the issue of neuro-adaptive optimal control for stochastic nonlinear systems with state constraints under the presence of non-affine faults. The presence of non-affine faults poses considerable challenges to the stability and performance of the system. To address these challenges, in this article, an adaptive neural network (NN) control scheme is devised within the identifier-critic-actor architecture, enabling the approximation of unknown dynamics and the design of both virtual and actual optimal controllers. In addition, Barrier Lyapunov Functions (BLFs) are utilized to ensure stability while handling state constraints, and a Butterworth low-pass filter is introduced to compensate for high-frequency noise and non-affine nonlinear faults, enhancing system robustness. Furthermore, a novel hybrid event-triggered control (HETC) strategy is proposed to reduce communication and computation demands while optimizing resource utilization. The suggested control strategy guarantees the boundedness of closed-loop signals and keeps all state variables within predefined compact sets. Lastly, the efficacy of the proposed optimal control method is demonstrated through simulation results.