Adaptive Neural Sliding Mode Control-Based Real-Time Security Reachable Set Control of Markov Jump Cyber–Physical Systems Against Actuator Attacks
针对受时变延迟和执行器攻击的非线性半马尔可夫跳变信息物理系统，提出一种自适应神经网络滑模控制策略，确保系统状态在预设安全可达集内有界且稳定。
This article addresses the real-time security reachable set (RS) control problem for nonlinear semi-Markov jump cyber–physical systems (s-MJCPSs) subject to multiple time-varying delays and actuator attacks. Then, based on the neural network (NN) approximation approach and employing a sliding mode control (SMC) strategy, an adaptive NN SMC scheme is proposed to tackle challenges arising from nonlinear attack functions and mode jumps. The proposed strategy ensures s-MJCPSs stability and mean-square boundedness of system states within a predefined RS. Sufficient conditions for real-time security RS control are derived. Finally, an example is presented to demonstrate the effectiveness of the proposed strategy in realizing adaptive real-time RS control.