摘要
通过动态监测板式换热器冷却水污垢热阻及影响污垢热阻的松花江水水质参数(如pH值、溶解氧、铁离子、氯离子、细菌总数、浊度、电导率、化学需氧量、碱度和硬度等)变化。采用BP神经网络主成分分析、主成分回归、全要素BP神经网络三种预测方法建立板式换热器污垢热阻预测模型,选取1-15号样本为训练或回归拟合样本,16-20号样本为测试样本,并将三种方法的预测结果进行了对比。结果表明,三种方法均可对板式换热器污垢特性进行有效预测,而基于主成分分析的BP神经网络方法的预测结果误差小,优于另外两种方法。
The fouling resistance of cooling water in a plate heat exchanger and variation of water quality parameters （ such as pH, dissolved oxygen, Fe3＋, CI-, total bacteria, turbidity, electrical conductivity, COD, alkalinity, hardness and so on）which influences the fouling resistance were dynamically monitored. Defining Samples 1-15 as the training or regression samples as well as Samples 16-20 as the testing samples,three forecasting models of fouling resistance in the plate heat exchanger were established by the BP neural network based on the principal component analysis, the single principal component regression and the total factors＇ BP neural network respectively. Besides,forecastlng results lay mree memods were contrasteu with total factors＇BP neural that the fouling characteristic of the plate heat exchanger could be predicted by aforesaid three methods effectively and the BP neural network based on the principal component analysis whose forecasting error is the least is better than the other two methods.
出处
《东北电力大学学报》
2014年第2期1-6,共6页
Journal of Northeast Electric Power University
基金
国家重点基础研究发展规划基金项目(2007CB206904)
东北电力大学博士启动基金项目(BSJXM-200916)
关键词
污垢热阻
预测
BP神经网络法
主成分分析
Fouling resistance
Forecasting
BP neural network
Principal component analysis
