Multiobjective predictive control problems in industrial manufacturing processes are characterized by no specific mechanism model, which poses a challenge to traditional methods. To address these problems, a multi-level data-driven predictive control method is proposed based on Willems' fundamental lemma. Firstly, a multi-level data-driven predictive control scheme is developed according to different priorities of objectives. Furthermore, a hierarchical solving method is introduced to solve the multi-level problem. Finally, the effectiveness of the proposed method is verified by a finishing rolling case.

data-driven predictive control,industrial manufacturing processes,multi-objective optimization problem,Willems' fundamental lemma