基于参数自优化SVM的供水管道泄漏多特征融合检测方法

Leak Detection Method Based on Multi⁃feature Fusion and SVM with Self⁃optimizing Parameter for Water⁃supply Pipeline

针对供水管道在多工况环境中小样本条件下泄漏状态难以辨识问题,提出了一种基于参数自优化SVM的供水管道泄漏多特征融合辨识方法。该方法通过对不同工况条件下采集的声信号求小波包熵值、峭度值与样本熵值特征,然后将这3种单一特征进行多特征融合后可以获得包含丰富信息的多维特征向量,并将该特征向量作为SVM分类器的输入向量进行分类辨识,通过网格化搜索法对SVM进行参数寻优,进一步提高泄漏辨识准确率。实验结果表明:该方法能够有效实现供水管道泄漏状态辨识以及其他工况状态的分类辨识,分类辨识准确率为95%。

Considering it is difficult to identify the leakage state of water supply pipelines(WSP)in multi working environ⁃ment under the small samples,a multi⁃feature fusion identification method for water supply pipeline leakage based on parameter self⁃optimization SVM was proposed.The wavelet packet entropy value,kurtosis value and sample entropy characteristic value of the detection signals were calculated in different working conditions through the acoustic signals collected.Then,after the multi⁃feature fusion of these three single features,multi⁃dimensional feature vectors containing rich information can be obtained,and the feature vector can be used as the input vector of SVM classifier for classification identification.The mesh search method was used to optimize the parameters of SVM to further improve the accuracy of leakage identification.The experimental results show that this method can effectively realize the identification of water supply pipeline leakage state and other working state classifica⁃tion identification,and the classification identification accuracy is 95%.