Theory And Application Of Economic Model Predictive Control For Batch Processes

With the diversification of market demand,batch processes have attracted extensive atten?tion from both industry and academia,and its position in industrial production process is becoming more and more important.With the development of industrial control technology,especially the advanced control technology based on model predictive control(MPC),the control theory of batch processes has been gradually mature.In recent years,some researchers have shown great inter-est in the economic performance of processes,and proposed economic model predictive control(EMPC)algorithm to improve economic performance.However,its main theory and application are for continuous processes.Focusing on the characteristics of batch process which is different from those of continuous processes,this paper re-examines the impacts of process disturbances,market fluctuations,decay of processes and other factors on processes from the perspective of economic performance,studies the economic performance of batch processes,fully exploits the space for economic performance improvement,and achieves the following results:1.According to the contributions of process disturbances to economic performance,they can be divided into favorable disturbances and unfavorable disturbances.An alternate feedback control strategy based on disturbance is proposed to deal with different types of distur-bances in different ways.The feedback control strategy acts on the process disturbances directly,and different feedback control laws are selected to deal with the different types of disturbances.At the same time,an economic integration robust design,which consid-ers both economic performance optimization and the closed-loop dynamics of processes,is proposed to pay attention to economic performance and improve the robustness of the con-trol system in dealing with process disturbances.The optimal parameters of the feedback control strategy are obtained by optimizing the economic integration robust design.The feedback control strategy is able to improve the average economic performance of batch processes by utilizing process disturbances;2.According to the repetitive characteristic of batch processes and the existence of unknown process disturbances,an economic MPC algorithm based on iterative learning control(ILC)is proposed.By gradually learning the changes of process disturbances along batch direction and time direction,the control actions are modified according to the economic performance index.Thus,batch processes run at operating conditions with better economic performance in actual,improving the economic performance of batch processes.The economic MPC based on ILC is divided into two parts:batch-to-batch and within-batch cases.The for-mer learns the repeatability of batch-direction process disturbances to guide the subsequent batches,while the latter learns the randomness of time-direction process disturbances to avoid affecting subsequent batches;3.According to the limited batch running time and the characteristic that products can only be obtained at the end of batch processes,an economic MPC based on the average economic performance and unfixed terminal time is proposed.The algorithm takes the terminal time as an optimization variable,so that it can choose the batch running time reasonably according to the change of process disturbances.Besides,the average economic performance is taken as the obj ective function,which not only considers the economic benefits of a single batch,but also considers the time cost.As a result,the economic benefits per unit time is the highest.Aiming at the problem that the process is highly disturbed and the terminal constraints are not reachable or the economic benefits can not be guaranteed,the early stop mechanism is proposed,which runs online to judge if the current batch should be stopped immediately to reduce the economic losses caused by the heavy disturbances.4.According to the special case of batch process,Run-to-Run process,a method of partial least square(PLS)modeling based on decay is proposed.By adding decay terms to the internal model of the standard PLS,the problem of multi-variable process coupling and the problem of process decay modeling are solved simultaneously.Aiming at the problem that process decay may cause the deviation of optimal operating condition,the economic Run-to-Run control is proposed to replace the original tracking Run-to-Run control.With the decay of processes,the optimal economic performance can be obtained by moving the operating condition step by step according to the economic performance index.