Abstract: Hierarchical model predictive control(HMPC) strategy based on model switching is proposed to solve the problem of optimal working point deviation calculated by plant-wide optimization. In the plant-wide optimization layer, SQP algorithm is used to calculate the optimal point under maximize economic benefits. In the control layer, the whole workspace is divided into several subintervals, in the subintervals the corresponding sub models are set up with the minimum gain method, last MPC (model predictive control) controller is designed under sub model selected by output error index to track the optimal point. In a word, HMPC strategy based on mode switching can realize system optimization in the economic and control performance, and the comprehensive economic benefits of the system is improved. In this paper, simulations show that HMPC strategy based on model switching is effective in improving economic benefits. The comparison between the multi-model control strategy and the single model control strategy show that the system control performance is improved by HMPC strategy based on model switching.

Key words: Hierarchical predictive control, economic benefit; multi-model predictive control, model switching , setpoint tracking 

摘要： 针对商品价格波动引起整厂优化最优工况点偏移问题，提出了一种基于多模型切换的分层预测控制策略。优化层用SQP算法确定经济效益最大时的关键变量设定值。控制层将对象的整个工作区间划分为几个子区间，在子区间用小增益法建立相应子模型，根据输出误差指标切换模型，用子模型代替当前模型，跟踪设定值。基于多模切换的分层预测控制策略能够同时达到系统的经济优化和控制性能优化，提高了系统的综合经济效益。仿真结果验证了多模型预测控制策略在物料价格波动下可以保证生产过程的经济效益最大化的有效性，并对单模型控制策略和多模型控制策略进行了比较，分析了多模型控制策略对系统控制性能的影响。

关键词: 分层预测控制, 经济效益, 多模型预测控制, 模型切换, 设定值跟踪