The Design And Realization Of Predictive Control Algorithm In Optimal Control Platform

The application of predictive control algorithm to the thermal optimization control platform has important practical significance.On the one hand,it can improve the function of the optimization control platform,and on the other hand,it is also helpful to the popularization and application of predictive control.Based on the full understanding to the software and hardware structure of the thermal optimization control platform,this paper selects the most widely used dynamic matrix control algorithm,and finally encapsulates it in the form of advanced algorithm module.In this paper,dynamic matrix control is extended from single model to multi model,each prediction model presents parallel structure.The optimal controller of each prediction model is designed by quadratic objective function consisting of output error and control constraint.The conditional probability of the matching degree between each prediction model and the actual object is calculated by recursive Bayesian estimation method,and obtain the output weights of the optimal controllers by normalizing the conditional probabilities.The modularization process of dynamic matrix control algorithm adopts the popular object-oriented programming idea in the field of software development,and the programming platform selects the classic Visual C++6.0.The critical technologies for the modularization process include the sampling file management of the prediction model,the refresh of the sampling state,the update of the module configuration comparison file,the realization of the module middle point entry into the system database and the implementation of the matrix operation in the lower controller.Finally,In order to verify the control effect of DMC module,the transfer function relational model between the swing angle of burner and the temperature of reheated steam is selected.In the optimization control platform,we build the simulation logic using DMC module and PID module to control,which the DMC module operates in two ways: multi model and single model.Debugging shows that DMC algorithm module can achieve better control effect when working in multi model mode,but compared to the PID,the output of the DMC control is more intense.