基于PSO-SVM模型的供水管网漏损诊断

为了找到更高精确度的供水管网定位方法,基于支持向量机搭建了PSO-SVM给水管网漏失诊断模型,对影响支持向量机(SVM)性能的两个重要参数c和g使用粒子群优化算法(PSO)做了优化处理,使得支持向量机的运算速率和准确度显著提高,利用单漏点供水管网仿真模拟试验平台测得管网各运行工况下的漏点特征数据,测得数据通过后期归一化处理作为PSO-SVM模型的输入样本数据集,样本数据经PSO-SVM模型运行后证明该模型可有效对管网漏失点做出精准定位,并能对各漏失点的漏失量做出精准预测。

Leakage Diagnosis of Water Supply Network by PSO-SVM Model

In this paper, a leakage diagnosis model of PSO-SVM water supply network is built based on support vector machine (SVM). Two important parameters c and g, which affect the performance of SVM, are optimized by particle swarm optimization (PSO), and improve the operation rate and accuracy of SVM significantly. The characteristic data of leakage point under each operating condition of pipe network are measured by using the simulation experiment platform of single leakage point water supply network. The measured data are used as the input sample data set of PSO-SVM model through later normalization treatment. After the sample data is run by PSO-SVM model, it is proved that this model can effectively refine the leakage point of pipe network. The accurate prediction of the leakage of each leakage point can be made.