基于压缩感知的电力变压器声纹信号特征提取

Feature Extraction of Power Transformer Voiceprint Signal Based on Compressed Sensing

面向在运变压器运行状态在线监测应用,基于压缩感知理论和小波包分析技术,提出一种基于压缩感知和小波信息熵的变压器声纹特征提取方法,用于变压器声振异常检测和故障诊断。采用本文方法提取变压器铁心故障仿真数据声纹信号特征,应用PSO-SVM分类完成故障诊断仿真。实验结果表明,本文方法能够在较高压缩率条件下,获取较高的故障识别精度。

For the online monitoring application of transformer operation status in service,based on the compression sensing theory and wavelet packet analysis technology,a method for extracting the voiceprint feature of transformer based on compression sensing and wavelet information entropy is proposed,which is used for the detection and fault diagnosis of transformer acoustic and vibration anomalies.The method is used to extract the voice signal feature of transformer core fault simulation data,and PSO-SVM classification is used to complete fault diagnosis simulation.The experimental results indicate that the method in this paper can obtain high fault diagnosis accuracy under high compression rate.