Research On Optimization Control Methods For Batch Process Considering The Model Error

Owing to less space equipment and flexible operation, Batch process is widely used by people. With the further improvement of batch process in worldwide, a more pressing need in production automation, optimization of production cost and process projects are proposed by manufacturing enterprises of batch process. Therefore, researching on optimization and control for batch process has important theoretical and practical significance.Flexibility of batch process determines the processing products may change at any time, and do not have a lot of experiment and time used for Model Distinguish, which brings a lot of problems to the build of the accurate model. Therefore, a new study is based on the simplified model, which is easy to cause certain deviations between simplified model and the practical process. The work of this paper mainly studies how to effectively reply to the batch process optimization control in the condition of error or big error model exists. This is considered the batch process error model optimization control method. Using the iterative learning control method to make full use of the limited running time and the repeated operation of batch process, and does not depend on the precise system mathematical model, can deal with nonlinear strong coupling characteristics system with high uncertainty in a very simple way. when the model error is relatively small, this paper based on the partial least square algorithm (MPLS) and iterative learning control method between batch to batch and within batch of process, through the iterative optimization to eliminate the influence from model error and disturbance to the system; As for the optimization control with the index information loss of the batch production process, this paper puts forward a robust optimization strategy, and verified through the batch reactor optimal control simulation. As for the big relative error model, this paper puts forward an iterative optimization control method which introduces the gradient of the error model. Error information is obtained by the error prediction model. The optimal initial value is obtained by the model error interval. The effectiveness of the proposed method is verified through the alcohol fermentation process optimization simulation.