关于净水混凝投药控制器系统的设计  

Design of Coagulant Dosing Control System of Water Purification

作　　者：徐少川[1] 刘东昆 井元伟[1] 刘宝伟[1] 

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114000

出　　处：《计算机仿真》2016年第5期339-343,共5页Computer Simulation

基　　金：国家自然科学基金(61273011)

摘　　要：出水浊度是净水混凝投药过程的重要指标,由于过程受多变量影响,且滞后时间较长,故工业控制中常用的反馈控制器无法实现对出水浊度的精确控制。为了将出水浊度控制在设定的范围内,并且有效地减少投药量,设计了一种基于小脑神经网络的前馈控制器,并应用于混凝投药控制系统。同时,为了提高CMAC神经网络的泛化能力和学习精度,提出了感知器和权值地址的隶属度的概念,对CMAC神经网络的映射过程做出了改进。仿真结果表明,前馈控制器将出水浊度有效地控制在设定的范围0.7～0.9NTU内,并且实现了投药量的优化。表明前馈控制器具有较强的实际应用价值。Turbidity of effluent is an important index of coagulant dosing of water purification process.Because it is a multivariable and large time delay system,feedback controller,which be always used in the Industrial control,could not control turbidity of effluent in setting range.In order to control the turbidity in the setting range and effectively reduce the dosage,a kind of feedforward controller that based on the CMAC neural network is designed.And,in order to improve the generalization and learning precision of CMAC neural network,the concepts of the membership of perception and the address of weights is proposed,then the mapping process and the weights training process of CMAC neural network is improved.The simulation result shows that the turbidity of effluent could be controlled in the setting range that 0.7- 0.9 NTU and dosage could be optimized.It shows that this feedforward controller practical application value.

关 键 词：混凝投药 前馈控制器 小脑神经网络 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]