Research On Adaptive Tracking Control Of Cyber-Physical System Based On Incomplete Measurement

As the research objectives and scope of the control field continue to expand,the control system also expands from linear to nonlinear.At the same time,due to the continuous development of computer network technology and nonlinear control theory,Cyber-physical System(CPS)has attracted more and more attention of scholars,It is a next-generation intelligent system integrating control,communication and computing.In real life,an important feature of CPS includes the interaction between physical processes and network dynamics.By exchanging the information obtained from the network side,the performance of the CPS can be improved through the cooperation of the subsystems.However,due to the physical and technical limitations of the communication network,problems such as packet loss,saturation,and delay will occur during the transmission process,which inevitably reduces the control performance and even makes the system unstable.Therefore,this thesis designs an adaptive event-triggered neural network tracking controller for strict-feedback and pure-feedback CPS.The controllers are based on backstepping control theory and event-triggered strategy to ensure the stability of the system.The research contents are as follows:1.Introduce the research directions and urgent problems of experts and scholars in the CPS in recent years,and propose new solutions to the existing CPS stability problems at this stage.The relevant theorems and knowledge points involved in this thesis are introduced.2.The adaptive event-triggered neural network tracking control problem for a class of strict-feedback CPS with incomplete measurements is studied.In incomplete measurements,system performance is degraded and state variables are not available due to data transfer problems.To solve these problems,radial basis function neural network control is introduced to approximate unknown nonlinear functions in CPS.An adaptive controller is constructed based on backstepping control theory and event-triggered strategy to ensure that all closed-loop signals are uniformly ultimately bounded in mean square and avoid Zeno-behavior.Finally,the effectiveness of the proposed control method is proved by simulation.3.The adaptive event-triggered neural network tracking control problem for a class of pure-feedback CPS with incomplete measurements is studied.Since the state variable becomes unavailable or distorted in incomplete measurements during data transmission,this may degrade the performance of the system.To solve these problems,radial basis function neural networks are used to approximate unknown nonlinear functions in CPS,and the Butterworth low-pass filter is introduced to estimate unmeasurable states,which are used to construct neural network observers.By constructing the Lyapunov function,the tracking error of the controller is confined within a small boundary.Based on the backstepping control theory and event-triggered strategy,the control signal of the fixed threshold strategy is obtained,and two adaptive controllers for the CPS are established,which ensures that all closed-loop signals are uniformly ultimately bounded in mean square and avoid Zeno-behavior.Simulation results confirm the feasibility and effectiveness of the controller.