水电站混凝土浇筑用冷水管结垢系数的神经网络预测模型  

Study on Fouling Coefficient of Cold Water Pipes for Dam Concrete Pouring Based on Neural Network Prediction Model

作　　者：朱敏[1] 赵玮 任万英 ZHU Min;ZHAO Wei;REN Wanying(Department of Hydraulic and Civil Engineering,Inner Mongolia Technical College of Mechanics&Electrics,Hohhot 010070,Inner Mongolia,China;Daxin Construction and Development Co.,Ltd.,Hohhot 010020,Inner Mongolia,China;School of Electric Power,North China University of Water Resources and Electric Power,Zhengzhou 450003,Henan,China)

机构地区：[1]内蒙古机电职业技术学院水利与土木建筑工程系,内蒙古呼和浩特010070 [2]达信建设发展有限公司,内蒙古呼和浩特010020 [3]华北水利水电大学电力学院,河南郑州450003

出　　处：《水力发电》2020年第10期68-72,共5页Water Power

基　　金：河南省高等教育教学改革项目(ZZJG-C6047);2018年度河南省高校创新项目(J18B0413)。

摘　　要：为了准确、有效的预测水电站大坝大体积混凝土浇筑的冷水管内部污垢的累积规律,设计了一种基于K均值算法和Chebyshev神经网络相结合的污垢预测模型。利用改进的神经网络预测污垢系数时,具有算法简单、收敛速度快的特点。改进的Chebyshev神经网络模型提供了一个预测污垢系数的有效方法,且具有较好的预测能力;在相同的精度下,该方法的收敛速度优于一般的神经网络。In order to accurately and effectively predict the accumulation of fouling in the cold water pipes of large-volume concrete pouring of hydropower dams,a fouling prediction model based on the combination of K-means algorithm and Chebyshev neural network is designed.When the improved neural network is used to predict the fouling coefficient,it has the characteristics of simple algorithm and fast convergence.The improved Chebyshev neural network model provides an effective method for predicting the fouling coefficient,and has good predictive ability.Under the same accuracy,the convergence speed of this method is better than the general neural network.

关 键 词：混凝土 冷水管 CHEBYSHEV神经网络 K均值算法 污垢系数 

分 类 号：TV642[水利工程—水利水电工程]