基于广义回归神经网络的污垢热阻预测  

Prediction of Fouling Thermal Resistance Based on Generalized Regression Neural Network

作　　者：王久生 张春波 苏涛 曹生现[2] WANG Jiu-sheng;ZHANG Chun-bo;SU Tao;CAO Sheng-xian(Jilin Electric Power Research Institute Co. Ltd., Changchun 130021, China;School of Automation Engineering, Northeast Electric Power University, Jilin 132012, China)

机构地区：[1]吉林省电力科学研究院有限责任公司,长春130021 [2]东北电力大学自动化工程学院,吉林132012

出　　处：《科学技术与工程》2019年第34期169-173,共5页Science Technology and Engineering

基　　金：吉林省电力科学院有限公司科技项目(SGTYHT/17-JS-199)资助

摘　　要：在生产中,换热器受到污垢沉积的影响较大。为了研究污垢在换热设备中的变化趋势,实验模拟了金属-水-蒸气体系下的换热器动态循环系统,测量了流经换热管中冷却水的各种水质参数,并结合实验数据,建立了基于广义回归神经网络(generalized regression neural network,GRNN)的换热管污垢热阻预测模型。通过交叉验证确定了最佳平滑系数为0.2,预测样本与实测样本具有较高的拟合精度,其相对误差最大为8.91,符合工程要求,证明该方法是可行的。In production,the heat exchanger is greatly affected by fouling deposition.In order to study the changing trend of fouling in heat exchanger equipment,the dynamic circulation system of heat exchanger in metal-water-vapor system was simulated.Various water quality parameters of cooling water flowing through heat exchanger tube were measured.Combining with experimental data,a prediction model of fouling thermal resistance of heat exchanger tube based on generalized regression neural network(GRNN)was established.The optimal smoothing coefficient 0.2 is determined by cross-validation.The predicted sample and the measured sample have higher fitting accuracy,and the maximum relative error is 8.91,which meets the engineering requirements.It proves that this method is feasible.

关 键 词：换热器管 水质参数 污垢热阻 神经网络 

分 类 号：TK172.4[动力工程及工程热物理—热能工程]