Adaptive Fault-tolerant And Resilient Control For Cyber-physical Systems

Cyber-physics systems have attracted the attention of many scholars at home and abroad due to their significant potential in analyzing practical engineering systems.Cyber-physics system is a multi-dimensional complex system that integrates computing,network,and physical environments,covering many internal system engineerings,such as embedded computing,unique environment awareness,network control,and network communication.Therefore,cyber-physics systems need more prosperous control schemes to maintain their controllability.Researchers favor adaptive control strategies due to their unique flexibility in updating control parameters.However,traditional adaptive control needs to improve its application due to the need for accurate system models.Neural networks and fuzzy logic systems can approximate the unknown nonlinear characteristics of systems,thereby providing great convenience for the design of system controllers.To ensure the smooth operation of the information physics system and improve the system’s anti-interference ability,this thesis studies the stability control scheme design of the cyber-physics system.The main contributions are summarized as follows:(1)An event-triggered adaptive controller design study was conducted for a class of cyber-physics systems with actuator and sensor failures.Firstly,a neural network state observer is designed to estimate the state parameters,which overcomes the influence of unknown system state parameters on controller design;A sliding mode observer is designed to estimate the output state of the navigator for each follower in response to the communication topology dependency problem among multiple systems.Then,to solve the differential explosion problem in traditional adaptive control methods,we adopted instruction filtering technology,thereby reducing the computational burden of the system.In addition,we have designed a compensation mechanism and a switching threshold event triggering strategy to solve better the problems of sensor and actuator failures and communication resource waste in the system.We propose an adaptive fault-tolerant control strategy to achieve the leader-following consensus for cyber-physics systems with actuator and sensor failures.(2)An event-triggered adaptive resilient controller design study was conducted for a class of cyber-physics systems subject to spoofing attacks and state delays.Due to the simultaneous occurrence of deception attacks,state delays,and unknown external disturbances,the actual system state is unavailable,and the control coefficient is unknown.First,we adopted the Nussbaum gain function instead of the actual control gain.We applied the attacked system variable to the controller design,breaking through the dependence on the real state of the system.In addition,a disturbance observer is constructed based on the system state under deception attacks,further improving the robustness of the system.In order to eliminate the impact of state delays,a suitable Lyapunov Krasovskii function was constructed during the backstepping design process.Then,by using the event-triggering mechanism,the possibility of external attacks on system information is further reduced.Based on the above,we propose an adaptive resilient control strategy to achieve cooperative control of cyber-physics systems with deception attacks and state delays.(3)Based on content 2,we have put forward higher requirements for the steady-state performance of the system.An adaptive fuzzy finite time elastic controller based on event triggering was designed for a class of switched information physics systems with nonlinear actuator hysteresis under deception attacks.Firstly,the nonlinear uncertainties in the system are eliminated by using a fuzzy logic system,simplifying the design process of the system controller.In addition,we use finite-time instruction filters to eliminate the exponential explosion caused by adaptive backstepping,improve the convergence speed of filtering errors,and design a filtering error compensation system,further improving the control performance of the system.Then,we combine the system’s external disturbances with the actuator’s nonlinear characteristics to construct a synthetic disturbance observer,which improves the system’s anti-interference ability.Finally,based on the command Lyapunov function method,under the rule of arbitrary system dynamic switching,a sufficient condition is obtained to ensure that all signals in the closed-loop system are uniformly bounded,and an upper bound of the system convergence time is given.