基于数据驱动的控制器综合性能优化

Comprehensive Data-Driven Controller Performance Optimization

为了提高过程回路的控制性能,提出一种基于数据驱动的控制器综合性能优化方法。首先,通过分析虚拟参考反馈校正方法与内模控制器设计过程的联系,确定数据驱动控制器校正所需参考模型的结构;然后,定义一种结合绝对误差积分和最小方差性能基准的综合性能指标,再基于该指标确定参考模型的未决参数;最后,利用控制系统闭环数据实现对控制器参数的校正。通过数值仿真和工业实例验证了所提方法的有效性,并将其与内模控制(IMC)-比例积分微分(PID)方法进行比较,验证了该方法的优越性。

To improve the control performance of process loops,a data-driven controller-based comprehensive performance optimization method is proposed.Firstly,the structure of the reference model required for data-driven controller calibration is determined by analyzing the connection between the virtual reference feedback calibration method and the internal-mode controller design process.Secondly,a comprehensive performance index combining the absolute error integral and the minimum variance performance benchmark is defined,and then the pending parameters of the reference model are determined based on this index.Finally,the calibration of the controller parameters is realized by using the closed-loop data of the control system.The effectiveness of the proposed method is verified by numerical simulations and industrial examples,and its superiority is verified by comparing it with the internal model control(IMC)-proportional integial differential(PID)method.