基于遗传算法优化BP神经网络的管网漏失定位模型研究

Study on Leakage Location Model of Water Supply Network Based on BP Neural Network Optimized by Genetic Algorithm

针对BP神经网络用于管网漏失定位时易出现收敛速度慢及陷入局部极小值的问题,利用遗传算法对BP神经网络的权值和阈值进行优化,并以A市供水管网为例,选取各个分区的供水管段进行不同程度的漏失模拟,将模拟数据作为训练样本训练遗传算法优化的BP神经网络,得到管网漏失时压力监测点的压力变化率和漏点位置之间的非线性关系,构建基于遗传算法优化BP神经网络的管网漏失定位模型。实例应用结果表明,基于遗传算法优化BP神经网络的管网漏失定位模型的收敛速度和预测精度均优于传统BP神经网络模型,可应用于实际工程。

When the back propagation(BP)neural network is used for the leakage location of water supply network,it was easy to slow down the convergence and fall into the local minimum.To solve this problem,genetic algorithm was used to optimize the weight and threshold of the BP neural network.The water supply pipelines of each district in city A was selected to perform leakage simulations with different degrees.The BP neural network optimized by genetic algorithm was trained with the simulation data.Then the nonlinear relationship between the pressure change rate of pressure monitoring points and the location of leakage points was established,and a leakage location model of water supply network based on BP neural network optimized by genetic algorithm was constructed.The example results show that the optimized leakage location model is superior to the traditional BP neural network model in convergence rate and prediction accuracy,and it is better to use in water supply network.