Performance Assessment And Remedy For A SCR Denitration System Under Model Predictive Control

In recent years,with the increasingly strict national emission standards for pollutants from coal-fired power plants,research on the control of selective catalytic reduction(SCR)denitration systems in coal-fired power plants has become particularly important.The SCR denitration system is a typical large-delay,large-inertia object with certain non-linearity.Model predictive control can directly deal with process constraints and time delay characteristics,and has inherent fault tolerance and implicit decoupling capabilities.It has broad application prospects in the control of SCR denitration system.However,for the SCR denitration process,as the running time continues to increase,the catalyst will continue to be consumed,and the dynamic characteristics of the denitrification process will continue to change,therefore the established denitration process model will deviate from the actual denitration process apparently.MPC is essentially a model-dependent control method,so it has relatively high requirements for the accuracy of the prediction model,and the occurrence of modeling errors will greatly affect the control effect.If the corresponding measures are not taken in time,it will have a great impact on the SCR denitration efficiency and environment.This paper studies the performance assessment,diagnosis and remedy of the SCR denitration predictive control system from the perspective of the maintainability of the control system.First,the performance assessment of the SCR denitration predictive control system based on the nominal sensitivity function is studied to monitor the control performance of the SCR circuit in time.When the control performance decreases,the performance diagnosis method of the SCR denitration predictive control system based on the relative sensitivity index is conducted.Its main task is to distinguish whether the performance degradation is due to model mismatch or unmeasured disturbance.And when a model mismatch occurs,the model mismatch diagnosis method based on the process model ratio can be used to further determine the type of model mismatch.Finally,through the auto-covariance-based performance repair method,the estimation of the model mismatch is obtained and the performance repair is achieved.