Model Predictive Control Based on Bayesian Forecasting for a Nitrobenzene Rectifying Tower

Step test method is a commonly used system identification method in the plant advanced process control application. Bayesian forecasting method is a kind of method based on subjective experience information and historical data. In this paper, a model identification method based on Bayesian forecasting is proposed. Combining the operators experience information and historical model data, the Bayesian method is used to predict the process model, and the model monitoring is used to abandon the abnormal data. Finally, the MPC controller is designed and the practical application of the top temperature control of the nitrobenzene rectifying tower proves the effectiveness of this method.