The Analysis And Compensation Of The Random Communication Delay In Networked Control Systems

With the mutual penetration and development of control,network and computer technology,the control system shows the characteristics of more complex structure and wider spatial distribution.It requires to achieve decentralized control completely and external connection with Web and Internet.Then,Network Control System came into being.The limited network bandwidth leads to unavoidable communication delay,packet loss and disordered,for which the system loses constant,integrity,causality and uncertainty.At present,more and more scholars focus on designing stable and reliable systems with high performance.As the main issue,the research of characteristic analysis and compensation on communication delay has an important theoretical significance and wide application value,which is also the research direction of this paper.The main contents of this thesis are as follows:Aiming at the compensation issue of communication delay in networked control systems,a method to design the time-delay compensation controller is proposed based on generalized predictive control(GPC).It consists of a network predictor and a time-delay compensator,which is used to make up for the delay with different characteristics.Meanwhile,the control parameters depend on random delay is given to improve the system performance and the stability conditions of the closed-loop system are also constructed.Using TrueTime toolbox to simulate the real network communication,both of fixed delay and stochastic delay in networked environment are considered to analyze the compensating performance and the corresponding effect of different control parameters.The simulated results illustrate a better performance of the varying-parameter compensation controller than the fixed one.Aiming at the uncertain problems caused by the random delay of forward channel in networked control systems,an online predictive model is proposed based on autoregressive model(AR)on account of timing analysis.The delay series is collected through the feedback channel,AR model is established after pretreating the delay series and the parameters are updated online by recursive least-square algorithm.Finally,rolling forecast is considered to realize multi-step prediction to achieve the delay.TrueTime toolbox is used to establish the simulation system,the experimental results suggest that this online predictive model possesses a better predictive performance compared to the one based on GRNN.Based on the online multi-step prediction model,an improved compensation controller for the stochastic delay in networked control systems is proposed.The interval partition is used to describe the randomness of communication delay to obtain a reliable prediction value of loop delay.To compensate for the loop delay,the controller mainly contains a state predictor,a network predictor and a time-delay compensator.The state predictor predicts the delay of the forward channel and estimates the state of the controlled object for the network predictor to obtain optimal control increment sequence.The method of setting control parameters is given as well.Constructing the simulation system with TrueTime toolbox,the results show that the compensation controller proposed here is effective for the random delay and obtains a better performance compared to the traditional one.