Optimizing and tracking of temperature control trajectory of batch process

Abstract

The optimal operation trajectory of batch process is always determined by experienced workers according to rich experiences. Obviously it is time-consuming and laborious. In order to overcome this, a simple and fast mathematical method is given to obtain the optimal trajectory based on process neural network (PNN), which can extend the traditional artificial neural network (ANN) in the time domain and is especially suitable for the process of serious nonlinear, long prediction and unable be modeled by traditional ANN. Then based on the PNN model, a quadratic form performance index function is presented to obtain the optimal operating trajectory. Finally to track the optimal operating trajectory, internal model control system constituted with filter, internal model and controller model based on BP neural network is proposed. The proposed scheme is successfully applied to a mercaptobenzothiazole batch process.