Adaptive Sliding-Mode Path-Following Control of Cart-Pendulum Robots with False Data Injection Attacks

This paper addresses the displacement path-following problem for a class of disturbed cart-pendulum systems under the fake data injection (FDI) actuator attacks. A filter operator is proposed to estimate the weight vector caused by unknown attacks and disturbances, so that the actuator attacks can be parameterized using neural networks. Then, combined with filter signals and based on adaptive neural network and integral sliding-mode techniques, robust path-following control schemes are proposed to withdraw the impacts of disturbances and FDI attacks. The uniformly ultimately bounded stability results of the closed-loop cart-pendulum system with neural network weight estimations and sliding functions are achieved based on Lyapunov stability theory. Finally, a simulation model of a material robot is used to verify the proposed control strategy.

Automated summary

This paper investigates the displacement path-following problem for a class of disturbed cart-pendulum systems under fake data injection (FDI) actuator attacks. A filter operator is proposed to estimate the weight vector caused by unknown attacks and disturbances, enabling parameterization of the actuator attacks using neural networks. Robust path-following control schemes are then proposed, leveraging adaptive neural network and integral sliding-mode techniques, to counteract the impacts of disturbances and FDI attacks. The closed-loop cart-pendulum system with neural network weight estimations and sliding functions achieves uniformly ultimately bounded stability results based on Lyapunov stability theory. Finally, a simulation model of a material robot is employed to validate the proposed control strategy.