An improved neural network state observer-based optimized control for time-delay robotic manipulator system under deception attacks

Robotic manipulators have become the critical equipment in contemporary industrial applications due to remarkable path tracking robustness and enduring mission execution capabilities. This paper proposes an optimized tracking control strategy for time-delay n-link robotic manipulator under the unknown deception attacks and external disturbances. To counteract the unavailability of real system states caused by deception attacks and external disturbances, an improved neural network state observer is designed to compensate for the system states, effectively mitigating the impact of deception attacks and external disturbances for robotic manipulators. The proposed strategy addresses the adversarial effects of deception attacks that inject false data and eliminates the adverse effects of external disturbances. Subsequently, the unknown time delays under deception attacks are handled by designing appropriate Lyapunov–Krasovskii functions, which effectively compensate for the time delay of robotic manipulators. Meanwhile, a modification term is constructed by a simplified reinforcement learning algorithm in the process of optimized control, which prevents training termination and diminishes the energy consumption when robotic manipulators execute tasks. The optimized control strategy improves robust tracking performance, which guarantees that all signals in the robotic manipulators remain bounded. Finally, simulation results validate the feasibility of the proposed control strategy.