[1]徐锦,笪跃武,金一,等.BP人工神经网络模型用于水泵特性曲线拟合[J].中国给水排水,2026,42(2):90-94.
　XUJin,DAYuewu,JINYi,et al.Application of BP Artificial Neural Network Model in Fitting Water Pump Characteristic Curve[J].China Water & Wastewater,2026,42(2):90-94.
点击复制

BP人工神经网络模型用于水泵特性曲线拟合

中国给水排水[ISSN:1000-4062/CN:12-1073/TU] 卷: 第42卷 期数: 2026年第2期 页码: 90-94 栏目: 出版日期: 2026-01-17
Title:
Application of BP Artificial Neural Network Model in Fitting Water Pump Characteristic Curve
作者:
徐锦， 笪跃武， 金一， 叶晓健， 胡淑圆， 王瑞， 王珂
（无锡市水务集团有限公司，江苏 无锡 214000）
Author(s):
XU　Jin， DA　Yuewu， JIN　Yi， YE　Xiaojian， HU　Shuyuan， WANG　Rui， WANG　Ke
（Wuxi Water Group Co. Ltd.， Wuxi 214000， China）
关键词:
水泵;  流量-扬程特性曲线;  BP人工神经网络;  拟合方法
Keywords:
water pump;  Q-H characteristic curve;  BP artificial neural network;  fitting method
摘要:
通过对水泵运行数据的研究获得性能参数之间的非线性关系，是泵房评估、泵组节能和协同优化调度问题求解的关键。针对传统绘制方法的不足，提出一种利用反向传播（BP）人工神经网络模型快捷、高效地拟合每台水泵在实际生产环境中的流量-扬程（Q-H）特性曲线，以提高曲线绘制精度、解决数值稀疏等问题的新方法。实例证明，在相同扬程下，该方法的流量预测值与实际工况值误差维持在3%左右，能较准确地反映性能指标之间的关系。
Abstract:
Obtaining nonlinear relationships between parameters of water pumps through the operation data is crucial for evaluating pump rooms, energy-saving pump units, and solving collaborative optimization scheduling problems. Aiming at the shortcomings of traditional characteristic curve drawing methods, a new method is proposed to use back propagation (BP) artificial neural network mo-del to quickly and efficiently fit the quantity-head (Q-H) characteristic curve of each water pump in the actual production environment, in order to improve the accuracy of curve drawing and solve the problem of numerical sparsity. At the same head, the examples prove that the error between the predicted flow value and the actual value remains around 3%, and can accurately reflect the relationship between performance indicators, which is particularly significant for production and practice.
更新日期/Last Update: 2026-01-17