阶梯式模型预测控制中控制权重的下限值研究

Research on Lower Bound of Control Weight in Stair-like Model Predictive Control

为了研究控制权重在预测控制器中起作用的范围与其余控制器参数的关系,对于阶梯式动态矩阵控制,通过一个中间变量,间接研究了控制权重λ的下限值与预测步长和阶梯因子的关系.在一定范围内,该下限值对参数变化很敏感,并出现很陡的峰值,此时可去掉控制加权项,其功能由阶梯因子来部分承担,来改善系统闭环性能.本文的研究结果可以指导控制权重λ的取值以及进一步简化算法.*

The relationship between control weight＇s bound and other parameters should be studied in predictive controller, For stair-like dynamic matrix controller, the relationship among the lower bound of control weight A, the prediction horizon and stair-like factor is studied indirectly through a inter-variable. The lower bound is sensitive to parameters in some range, and a steep peak may appear. In these cases, the weighted control item in the objective function can be omitted, and the control moves are arranged by stair-like factor to improve the closed loop performance. The result can help to set the value of A and simplify the algorithm.