Research On Dynamic Scheduling Of Networked Control System Based On Variable Sampling Period

With the rapid development of modern industrial control systems,networked control systems have become more and more widely used in actual production.The networked control system realizes the real-time data interaction and information sharing between the equipment nodes through the communication network medium,which brings many conveniences to modern production.Nevertheless,after the network intervenes in the control system,a series of new problems have arisen,such as network induced delay,data packet loss and timing disorder.These problems seriously hinder the development of networked control systems and damage the performance of network control systems.Therefore,how to weaken the influence of these problems and ensure that the performance of the networked control system can meet the actual production needs has become a hot research issue in the current control field.The research on the performance of the networked control system is currently mainly carried out from two aspects,the control strategy and the scheduling algorithm.Taking into account the fact that network bandwidth resources are limited and each node competes for limited resources to complete information transmission,this paper conducts research from the scheduling level and allocates node transmission tasks rationally to efficiently use limited network bandwidth resources.So that it can ensure the real-time capability of information transmission in the network.Considering that the sampling period plays an important role in the performance of the networked control system,this paper uses the sampling period as the starting point to study the dynamic scheduling algorithm of variable sampling period to reduce the adverse effects of network intervention on the system performance.The specific research is as follows:Considering the limited network bandwidth resources,a dynamic scheduling algorithm with variable sampling period is proposed for error correction gray neural network to predict network bandwidth.The algorithm first predicts the network bandwidth through an improved gray prediction method,then corrects the gray prediction error through the neural network,and uses the genetic algorithm(GA)to find the best parameters of the neural network,and finally obtains the network bandwidth prediction value.After that,the predicted bandwidth is allocated in the actual operation of the system to realize the dynamic adjustment of the sampling period,and the optimization effect of the scheduling algorithm is verified through simulation.Taking into account the excellent generalization ability of Support Vector Regression(SVR)and the characteristics of small sample requirements,SVR is used to predict the network bandwidth,and the variable sampling period dynamic scheduling algorithm of GA-SVR prediction bandwidth is proposed to make the prediction bandwidth.The value is relatively close to the actual network situation.Finally,the estimated bandwidth based on the real-time error of each loop is allocated online to adjust the sampling period of each loop in real time.The simulation proves that the scheduling algorithm effectively improves the performance index of the system.In the above two dynamic scheduling algorithms,neural network prediction and SVR prediction both parse the original data from different angles,and only contain part of the original data.In order to improve the prediction accuracy,this paper combines the individual prediction methods through certain rules and proposes variable sampling period dynamic scheduling algorithm for combined prediction bandwidth.Combine the two prediction methods,and select the reciprocal error method to determine the weight of each prediction method to improve the accuracy of bandwidth prediction.Finally,the estimated bandwidth is allocated online,and the sampling period of each loop is adjusted.The simulation shows that all system performance indicators are improved under the action of the algorithm.