Research On The Application Of Improved Generalized Predictive Control Algorithm In Networked Control Systems

Networked control systems employs shared network instead of private lines amongpoint-to-point nodes in the traditional control systems. Several advantages such as ease ofinstallation and maintenance, reduced cost, resource sharing emerge, result in wild application infields of remote medical treatment, remote teaching, aircraft and spacecraft and so on. However,because of the introduction of network, delay and packet dropout are inevitable when datatransmits over the network. Delay and packet dropout can degrade the control performance of thenetworked control systems, even destabilize the system, and make traditional control theoryunsuitable to apply in the networked control systems. Novel control schemes must be designed todirect towards the network features such as delay, packet dropout, packet out-of-order and so on.According to the theory of predictive control,the theory based on generalized predictivecontrol is introduced.And then,I do some research on the improved generalized predietivecontrol which can simplify the control algorithm and inhibit output overshoot.Aim to the delay inthe networked control system,this paper give a method that can realize compensation of thenetwork delay by setting network delay compensator actuator.For the robustness of networkedcontrol system(NCS), a predictive control method based on online least squares support vectormachine(LS-SVM)is proposed.The method based on LS-SVM is only to solve the problem aboutanalytical formula optimization,It can be used to establish model which the model’s order isunknown.Then,using linear fernel function, training data window and training data threshold todevelop an online training method suitable for the networked control system.Finally,this paper does some simulations on the networked control system which have timedelay and packet loss, the result shows that the presented method is rapid,accurate and robust.