Data-driven modeling and stability control for industrial zinc roaster and its edge controller implementation

As large-scale industrial equipment, the operation stability control of the zinc roaster can improve the roasting efficiency and ensure industrial safety. In the industrial field, the roaster temperature is generally controlled manually, but automatic control is an important way to obtain real-time, accuracy, and safety of operation performance. However, existing control methods highly depend on manual experience and knowledge, which decreases their robustness and accuracy. In addition, the complex control algorithm can hardly meet the real-time requirement on the premise of pursuing accuracy. In order to address these puzzles, a data-driven explicit model predictive control solution is proposed in this paper. Specifically, the canonical correlation analysis method is introduced to extract the key controllable variables from the complex and coupled observer variables of the industrial field roaster. Then, the subspace identification method is proposed to obtain the system model with a balance between model complexity and fitting accuracy. Finally, an explicit model predictive control is introduced to simplify the calculation process of online control and simultaneously ensure the real-time stability control of the roaster. The hardware platform deployment experiment is presented to verify the feasibility and real-time performance of the proposed control scheme.