Modeling and optimal-setting control of blending process in a metallurgical industry

This paper presents a kind of hierarchical inference strategy integrating quality prediction model for the optimal-setting control of blending process in alumina metallurgical industry. By integrating mechanistic model with intelligent compensator based on neural networks and feedback of indices, the prediction model is built to predict raw slurry quality. The target of raw slurry quality is online adjusted according the historical production of blending process and the current requirements of downstream process. Based on the biases between the prediction results and the target values of quality, a hierarchical inference strategy is proposed to determine an optimal set-point of raw material proportioning (RMP) in real time. The practical running results show that the eligibility rate of raw slurry is effectively improved, and the blending process is successfully simplified as well as the energy consumption is obviously reduced.