基于小波能谱矩阵相似度的电力暂态信号识别

暂态保护技术正呈现加速发展态势,但如何正确识别各类故障和非故障暂态,是暂态保护实用化必须解决的一个最大难题.信号经过小波分析之后得到的小波能谱能很好的反映信号不同时间的能量分布.电力系统中,不同电力暂态信号其时频特性必然不同,其在多个尺度下的小波能谱的分布必然存在差异,因此,在构造小波能谱矩阵的基础上,借鉴数字图像处理中的相似度的思想,提出基于小波能谱矩阵相似度的分类方法.建立各种电力暂态的标准能谱矩阵,计算测试样本的能谱矩阵与各种标准能谱矩阵的相似度,按照相似度最大原则实现分类.对电力系统单相接地故障暂态、线路正常开关操作暂态、电容投切操作暂态和雷电扰动暂态的仿真分析表明,该方法对电力暂态的识别率高.

Power transients recognition based on wavelet energy matrixes similarity

Transient protection is under fast development,but how to recognize those fault and non-fault transients is the most difficult issue of it.Wavelet energy which is gotten from transient signal by wavelet transform can reflect energy distribution of different instants.In power system,different transient signal has different time-frequency characters.Their wavelet energy distribution of multi-scales must be different.Based on this idea,and combined wavelet energy matrixes and similarity-based classify method in image process,this paper gives a new classifying method based on wavelet energy matrixes similarity.First,the paper constructs standard wavelet energy matrixes of each transient,then calculates similarity of test samples and standard matrixes,and realizes recognition by the similarity values.After simulating four kinds of transients in power system-single phase short,breaker operation,capacitance switching and lighting disturbance,the result shows this method is effective in transient recognition.