Time-delay Prediction And Compensation Of Networked Control Systems

Networked control system (NCS) is a real-time closed-loop feedback controlsystem.It isan emerging cross research field of computer technology,communicationstechnology and control technology.Compared with the traditional controlsystem,NCS has the advantages of resource sharing, remote operation and control,high reliability and easy maintenance. Butduring the data transmission in thenetworked control systems,time delay and packet lossare alwaysoccurred, whichmake the analysis of the systems more difficult and generally reduce theirdynamicperformance.In this thesis, we investigated the prediction and compensation oftransmission delaysand had the following results:1.We developed a network time delay model and proposed severaltime delayprediction methods. First, the causes and the composition of time delay wereanalyzed.An experimental platform was built to obtain the distribution of the actualnetworkedtime delays andthe delay models were discussed. Based on the networkeddelay model, we used LMI, artificial neural networks and support vector machine topredict the value of time delay.The prediction results were analyzed and compared aswell.2.A hybrid delay compensation control strategywas proposed. In the Hybriddelay compensation control, the delay predictive value as taken a controller input. Fordifferent delay values we used different control strategies.If the delay value issmall,the Fuzzy-PID was used to compensatethe time delay. On the other hand, if thedelay value is large, an improved JGPC method was proposed to compensate the timedelay. Simulation results showed that the proposed hybrid control strategy caneffectively improve the compensation effect and the stability of the system.3.In order to attenuatethe effect of thetime delay prediction accuracyduring thedelay compensation, an improved hybrid delay compensation control methods wasproposed, whichwas not based on the delay prediction and need notto switch controllers constantly. Simulation results showedthat the improved method was moreeffective to compensate the time delay.4.The Smith predictor estimated was usedin the network delay compensationcontrol.Based on the assumption that the controlled object is known,for the nonlinearnetwork time delay, a self-tuning Smith predictor was proposed to compensate for thedelay. Depend on the delay forecast we can adjust the Smith predictor model toget thefull delay compensation.5.We proposed an improved Smith predictor delay compensation control strategy.Double-ended structure Smith predictor was usedas the system controller and theMRAC was used as an adjustment controller when model mismatch. The proposedmethod can make the Smith predictor model and the actual model matched perfectly.Simulation results illustrated that the improved Smith predictor did not affected bythedelay prediction accuracy and had better dynamic performance.