Networked Control System Based On Improved Generalized Predictive Control

With the network control system(NCS)has more and more application fields and more and more wide distribution range,it is more and more urgent to improve the performance of NCS.As the most direct and typical problem affecting the performance of NCS,network delay not only negatively affects the dynamic and static performance of NCS,such as the increase of overshoot,the extension of regulation time and the narrowing of stability region,but also it can cause the system to become unstable when the network delay is too large.In order to compensate network delay and improve the performance of NCS,this paper studies NCS based on generalized predictive control(GPC)and improved GPC(implicit GPC and fast GPC).In order to predict network delay in advance and compensate it,this paper designs an EEMD-PACFElman-GWO combined prediction model to predict network delay,and in order to verify the prediction effect of the designed combined prediction model,this paper takes a group offline delay as the prediction object to compare the prediction effect of the combined prediction model and Elman neural network prediction model.For the NCS with exist short delay,this paper presents a design method of NCS with delay compensation based on GPC.Firstly,this method uses the time-stamp method to measure the loopback delay.Then,according to the loopback delay,the delay compensator selecting the appropriate control quantity to act on the system to realize the delay compensation.For the NCS with possible exist long delay.Firstly,this paper combining implicit generalized predictive control(IGPC)with prediction delay to get the IGPC algorithm based on predictive delay.Then,giving a design method of NCS with active delay compensation base on this algorithm and EEMD-PACFElman-GWO combined prediction model.In order to improve the performance of NCS,this paper presents an improved fast GPC algorithm which can improve the control accuracy,output stability and rapidity.The improved fast GPC algorithm introduce variable softening factor to improve the reference trajectory of the fast GPC algorithm.And use the output error sequence of AR prediction model to improve the output model of fast GPC algorithm.For the above research content,this paper uses MATLAB and True Time2.0 toolbox to simulation respectively,and the simulation results show that the above research contents have achieved their respective goals.