水质参数影响板式换热器污垢特性的神经网络分析  

An Analysis of Neural Networks of Water Quality Parameters' Effect on Fouling in Plate Heat Exchangers

作　　者：徐志明[1] 郭进生[1] 黄兴[1] 张仲彬[1] 

机构地区：[1]东北电力大学能源与动力工程学院,吉林吉林132012

出　　处：《中国农村水利水电》2010年第12期69-71,77,共4页China Rural Water and Hydropower

基　　金：国家重点基础研究发展规划基金项目(2007CB206904)

摘　　要：搭建了板式换热器污垢热阻实时监测系统平台,采用多项式拟合的方法分析出松花江水中电导率、铁离子、细菌总数、pH值、浊度等水质参数与污垢的形成过程密切相关;将其和时间作为输入量,污垢热阻作为输出量,建立基于BP神经网络的冷却水污垢热阻预测模型。结合绝对误差和相对误差的分析知:预测值准确率达到90%以上,网络泛化能力的可靠性强;分析比较多次预测结果表明,该模型不但适用于流速、换热器特性参数等为定值的情况,而且当这些参数发生改变时,该模型也适用。A fouling test system in plate heat exchangers is established and the water quality parameters of Songhua River such as conductivity,iron,total bacteria,pH and turbidity are closeted with the fouling by using the polynomial matching method.This paper selects these water quality parameters and time as the input variable,take the fouling resistance as the output variable,finds a prediction model of cooling water fouling resistance based on BP neural network.Combining the absolute error and relative error analysis,the prediction accuracy is up to 90% and the generalization ability of network is reliable.The results show that the model is applicable not only to the conditions that water flow,geometrical specifications of plate heat exchangers are constant,but also to the conditions when they change.

关 键 词：水质参数 板式换热器 污垢热阻 神经网络 

分 类 号：TK124[动力工程及工程热物理—工程热物理]