基于ITD和K均值聚类的电能质量扰动分析与识别

Analysis and Identification for Power Quality Disturbance Signals Based on ITD and K-mean Clustering Algorithm

将固有时间尺度分解ITD(intrinsic time-scale decomposition)与K均值聚类算法相结合,提出一种新的电能质量扰动分析与识别方法。首先,对电能质量扰动信号作ITD分解,得到各内禀尺度分量ISC(intrinsic scale component),然后提取能量最大的前3个ISC的能量比重值及ISC能量熵值作为扰动信号的特征向量,最后基于特征向量采用K均值聚类算法对扰动信号进行分类。对常见电能质量扰动信号的仿真分析结果表明,ITD能够有效分析和提取扰动信号特征,并具有较强的抗噪能力,K均值聚类算法分类正确率高,该方法具有良好的工程应用前景。

Combined intrinsic time-scale decomposition（ITD）algorithm with K-mean clustering algorithm,a new power quality disturbance（PQD）analysis and identification method is proposed in this paper. Firstly,the PQD signals are decomposed by ITD algorithm to obtain the intrinsic scale component（ISC）,and then the biggest three ISCs of maximum energy are extracted and the ISC energy entropy is calculated as the feature vector of PQD signal. Finally,based on the feature vector,PQD signals are classified using K-mean clustering algorithm. The simulation results of common PQD signal analysis show that ITD algorithm can effectively extract PQD characteristics with strong anti-noise ability,and K means clustering algorithm can classify PQD signals with high correct ratio. The proposed method has a good prospect of engineering application.