阶跃基权系数时变的预测函数控制

摘要

研究预测函数的精度优化问题,传统预测函数控制基函数为全局函数,过程预测轨迹与参考轨迹在预测时域内的拟合只有有限个拟合点,无法实现在整个预测时域内的整体优化目标.针对实际中无法实现全局拟合的情况,提出一种基函数进行加权来提高过程预测轨迹与参考轨迹逼近程度的新方法.根据参考轨迹的特点,采用阶跃基函数,其加权系数随着预测时步的增加而减小,在预测时域内使过程预测轨迹逼近参考轨迹,提高整体优化目标,减小了第一步预测控制量与理想控制量的差.每次控制量计算先进行基函数的权系数修正,保证过程预测轨迹与参考轨迹的尽可能全局拟合.上述预测函数控制策略用于某水厂混凝投药过程控制,在Matlab上仿真表明,具有更好的跟踪性和抗模型失配能力,比传统预测函数控制算法具有更好的控制效果.%Traditional predictive functional control can only fit limited points in the time domain of prediction, and is unable to optimize the target function in the whole time domain of prediction. This paper proposed a predictive functional control. According to the characteristics of reference trajectory, step basis function with changing weight coefficient was adopted, which can improve the optimization of the target function in the whole and reduce errors of control variable. The weight coefficient of basis function was modified before control variable computation, and the predictive function control was applied to the dosing of coagulant. The Matlab simulation results show that the predic tive functional control strategy base on changing basis function is better than common function strategy in tracking speed and anti-model mismatch ability.