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

作　　者：哀微[1] 朱学峰[1] 

机构地区：[1]华南理工大学自动化科学与工程学院,广东广州510640

出　　处：《控制理论与应用》2011年第3期335-342,共8页Control Theory & Applications

基　　金：广东省科技计划资助项目(2005B10201005);2008佛山市禅城区产学研资助项目(2008B1034);华南理工大学学生研究计划(SRP)资助项目(Y1090060)

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.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（IFF） 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 IPT 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.

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 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]