A novel control strategy based on temperature dynamic decoupling and belief expert rule base for large-scale vertical quench furnaces

The large-scale vertical quench furnace has large volume, complicated structure, strong coupling property, and a variety of heat transfer characteristics. It is hard to achieve stability control. And in the actual production process, due to the serious variety of ambient temperature, its temperature fluctuate greatly In this paper, a Kernel Extreme Learning Machine (KELM) prediction method is proposed to predict the decoupling parameters in real time by predicting the temperature difference between adjacent two zones at the current time to achieve decouple the temperature of each zone. Then, the Belief Expert Rule Base (BERB) PID control algorithm is utilized to control the temperature precisely. The results show that the method can effectively improve the homogeneity of the temperature in the furnace, and the method is better than the existing method when the ambient temperature changes greatly.