RBF-ARX Model Based Robust Predictive Control for the Coal Gasification Process

To solve the problem of performance degradation of linear model predictive control of the coal gasification caused by coal quality, a covariance-constrained minimum variance robust predictive control method based on the RBF-ARX (Radial Basis Function-Auto Regressive with Extra Inputs) model is proposed in this work. Historical data containing coal quality variation information is utilized in the proposed method, and the RBF-ARX model for the coal gasification process is identified through the application of the Zhang Neural Network (ZNN) method and the Least Squares Method (LSM). A multi-step prediction equation is constructed based on the identified RBF-ARX model, and a robust minimum variance covariance-constrained control approach is developed to solve the receding horizon predictive control (RHP) action. The simulation and industrial application results have verified the feasibility and effectiveness of the proposed method.