配电变压器环境噪声识别关键技术研究

Research on Key Technologies for Environmental Noise Identification of Distribution Transformer

针对配电变压器环境噪声复杂多变的特点,提出一种基于小波变换、支持向量机(Support Vector Machine,SVM)、噪声特征库的噪声识别方法。该方法首先采用小波变换对噪声信号进行预处理,其次利用SVM提取噪声特征并进行分类识别,最后结合噪声特征库综合判定噪声类型。实验结果表明,该方法的噪声识别精度、召回率、F1值等指标均优于传统方法,为配电变压器精准降噪提供了新思路。

A noise recognition method based on wavelet transform,Support Vector Machine(SVM),and noise feature library is proposed to address the complex and variable environmental noise of distribution transformers.This method first uses wavelet transform to preprocess the noise signal,then uses SVM to extract noise features and classify them,and finally combines the noise feature library to comprehensively determine the type of noise.The experimental results show that the noise recognition accuracy,recall rate,F,value and other indicators of this method are superior to traditional methods,providing a new idea for precise noise reduction of distribution transformers.