水厂混凝投药大滞后过程的数据驱动直接控制方法

Data-driven direct control method for coagulation dosing large time-delay process in water works

中文关键词  混凝投药  大时滞系统  Smith预估器  迭代反馈整定  数据驱动控制

英文关键词  coagulation dosing process  large time-delay  Smith predictor  iterative feedback tuning  data-driven control

中文摘要

      以自来水厂混凝投药大时滞过程为研究对象, 在迭代反馈整定(IFT)方法的基础上, 结合Smith预估控制结构, 提出了一种混凝投药过程数据驱动直接控制算法. 着重针对过程大时滞的特点, 在性能指标中加入了预估误差惩罚因子. 提出了一种新的步长设置方法, 使得步长的下降速率可调. 设计了3个闭环实验来求取性能指标梯度向量的无偏估计, 完全利用输入输出数据对广义控制器参数实现了自整定. 通过水厂混凝投药实测模型仿真验证了该方法在抗滞后方面表现出较强的收敛性和稳定性, 降低了对迭代初值的要求. 该方法既可以用于初始控制器的整定, 还可以用于投药过程中控制器的在线自整定.

英文摘要

      To deal with the large time-delay in the coagulation dosing process in water works, we, in combining with Smith predictor, propose a novel data-driven direct control method based on the iterative feedback tuning(IFT) method. In considering the characteristics of the large time-delay, we add to the performance index a penalty factor of predictive errors in the controller design. A novel step size setting scheme is proposed in which the rate of descent of the step size is adjustable. Three closed-loop experiments are carried out to perform the gradient descent optimization of the cost function which is computed totally from the I/O data in the close-loop system. The parameters of controller and predictor are automatically tuned for an unknown plant. Simulation studies with a practical coagulation dosing process model are conducted. Results show that the convergence and stability of this algorithm outperform classical IFT in large time-delay applications and with less dependence on the initial parameters. This scheme can be applied to the initial tuning of the controller or for the automatic tuning during the dosing process.