基于多测度的闪络故障行波与干扰杂波辨识

发布时间：2018-03-13 09:41

本文选题：故障行波　切入点：闪络故障行波　出处：《电力系统自动化》2017年19期 　论文类型：期刊论文

【摘要】：灵敏启动并经高速采录的宽频暂态数据富含干扰杂波,从海量暂态数据中自动筛选闪络故障行波是个急需解决的命题。采用小波能量熵、小波能量均值和小波能量方差3个测度对闪络故障行波和干扰杂波数据进行表征,形成行波数据的特征矩阵,并采用主成分分析对其特征矩阵进行降维处理,降维后的特征矩阵最大限度地保留了原有特征矩阵的主要信息。将降维后的特征矩阵作为样本,分别采用马氏距离和混合高斯模型聚类算法进行闪络故障行波和干扰杂波的辨识,实现对闪络故障行波的计算机筛选。大量现场实测数据的测试表明,采用马氏距离和混合高斯模型聚类算法均能够实现闪络故障行波的可靠筛选,并且混合高斯模型的辨识效果更好,正确率达97%以上。
[Abstract]:The broadband transient data, which is sensitive to start and recorded at high speed, is rich in interference clutter. It is an urgent proposition to automatically screen flashover fault traveling waves from massive transient data. Wavelet energy entropy is used. The three measures of wavelet energy mean and wavelet energy variance are used to characterize the flashover fault traveling wave and interference clutter data to form the characteristic matrix of traveling wave data, and the principal component analysis is used to reduce the dimension of the characteristic matrix. The reduced dimension matrix retains the main information of the original feature matrix to the maximum extent. Using Markov distance and mixed Gao Si model clustering algorithm to identify flashover fault traveling wave and interference clutter, computer screening of flashover fault traveling wave is realized. Both the Markov distance and the hybrid Gao Si model clustering algorithm can be used for reliable screening of flashover fault traveling waves, and the identification effect of the mixed Gao Si model is better, and the accuracy is over 97%.
【作者单位】：昆明理工大学电力工程学院;深圳供电局有限公司;
【基金】：国家自然科学基金资助项目(51667010) 云南省人培项目(KKSY20160428)~~
【分类号】：TM75

【相似文献】