Reserach And Application Of Multiple Model Predictive Control For Denitration System

Coal is the main energy in our country.Thermal power is the most widely used form of power generation.But along with the generation of electric power,nitrogen oxide(NOx)will be emitted,which will cause air pollution.Our country has proposed high standards and requirements for the emission of NOx along with the high frequency of haze and acid rain.Efficient denitration control system is the key to not only guarantee the concentration of NOx to reach the national standard,but also reduce cost for the power plant.Denitration system is a large delay,nonlinear and multi-mode system,which means it is difficult to control.The denitration control system using traditional PID control strategy often cause large fluctuation of the concentration of NOx.After the study of the existing control strategy of denitration control system and smart control strategy,the method of multiple model predictive control(MMPC)is applied to the denitration control system in this paper.MMPC combines the advantages of both predictive control and multiple model control,which has good robustness and improves the dynamic performance of the denitration system.MMPC is applicable for large delay,nonlinear and multi-mode system.The sub-models are composed of several fixed models and one adaptive model.The fixed models are obtained offline by the identification of the denitration system.The adaptive model can identify the model of denitration system and update the original idetification parameters online,which can accelerate the convergence rate of the adaptive model.Generalized predictive control(GPC)is applied in MMPC.GPC controller is designed for every sub-model.The controller corresponding to the adaptive model can be tuned online,which can improve the control accuracy.A switching mechanism based on the minimum accumulative error is formed to select the most appropriate model and controller to meet the productive purity requirements,and the switching mechanism is improved by hysteretic switching mechanism to avoid frequent switching of the controller.Simulation for the denitration system are presented to compare the performance of PID,multi-model PID and MMPC.The results demonstrate that the performance of MMPC is the best among three control strategies whether in the step response test or in varying load conditions.