基于最小二乘双SVM的管道泄漏检测  

Pipeline Leak Detection Based on Least Squares Twin SVM

作　　者：郝新新 Xinxin Hao(Unit 61267,Beijing 101100)

机构地区：[1]61267部队,北京101100

出　　处：《新疆钢铁》2025年第1期126-128,共3页Xinjiang Iron and Steel

摘　　要：供水管道是城市的重要基础设施,检测管道是否出现泄漏对于维护管道正常运行具有重要意义。本文提出了一种基于最小二乘双支持向量机(Least squares twin support vector machines,LSTSVM)的管道泄漏检测方法。首先,通过无线传感器采集供水管道的各种泄漏程度的声振动信号并进行特征处理,提取时域特征信息以构成特征数据集。然后,在基本LSTSVM模型的基础上,基于“一对一”策略构建了一种多分类LSTSVM泄漏检测模型,最后,将特征数据集送入所构建的多分类LSTSVM泄漏检测模型完成供水管道不同泄漏程度的检测识别。实验结果表明,本文所提的供水管道泄漏检测方法具有较高的泄漏识别精度,识别准确率达到了94.89%。Water supply pipeline is an important urban infrastructure,and detecting whether the pipeline leakage occurs is of great significance for maintaining the normal operation of the pipeline.This paper proposes a pipeline leak detection method based on Least squares twin support vector machines(LSTSVM).Firstly,acoustic vibration signals of various leakage degrees of water supply pipelines were collected by wireless sensors and processed to extract time domain feature information to form a feature data set.Then,on the basis of the basic LSTSVM model,a multi-classification LSTSVM leakage detection model was constructed based on the"one-to-one"strategy.Finally,the feature data set was sent into the constructed multi-classification LSTSVM leakage detection model to complete the detection and identification of different leakage degrees of water supply pipelines.The experimental results show that the proposed leakage detection method of water supply pipeline has high leakage recognition accuracy,and the recognition accuracy rate reaches 94.89%.

关 键 词：供水管道 泄漏检测 特征提取 LSTSVM 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]