基于数据驱动的多工况过程控制性能评估研究

The performance assessment of multiple mode MIMO industrial process based on Data-driven controller

为了解决如何选择合适的基准数据与实时数据进行控制性能评估问题,提出一种基于多变量分类的方法.用一种综合相似因子来衡量当前实时数据和某一工况下数据的相似性,确定实时数据所属的工况,而后再进行控制性能评估,通过仿真结果验证了新方法的有效性.

To choose the suitable benchmark data for data-driven controller performance monitoring from data based on kinds of multiple mode in the history database, a new method based on multiple variables classification is proposed. First the method provides a combined similarity factor to measure the similarity between realtime acquisition data and data from some mode in the history database, and then it determines the mode that real-time data belongs to, finally it assesses the controller performance. Simulation study demonstrates the efficiency of the proposed approach.