基于第二代小波变换和离散隐马尔可夫模型的电能质量扰动分类

Disturbance Classification Based on Second Generation Wavelet Transform and Discrete Hidden Markov Models

提出了一种将第二代小波变换和离散隐马尔可夫模型相结合的电能质量扰动分类方法。首先使用第二代小波变换对电能质量扰动信号进行时频分析,给出了一种基于模极大值的小波变换后处理方法,用以提取分析结果中表征扰动特征的模极大值,将这些模极大值组成扰动特征量组,经矢量量化后得到特征序列,然后将特征序列输入到由离散隐马尔可夫模型构建的分类系统中,实现对电能质量扰动的分类。在此基础上,给出了确定每类扰动定量指标的方法。仿真结果表明,该分类方法在强噪声环境下分类正确率高,且训练易收敛。

A new approach combining second generation wavelet transform （SGWT） and discrete hidden Markov models （DHMM） is proposed to classify power quality disturbances （PQD） in this paper. First, the PQD signals are decomposed with SGWT time-frequency analysis. A post-processing method of wavelet-transform based on module maxima is presented. The method extracted the module maxima which could represent the PQD character from SGWT time-frequency analysis results, and the module maxima are used to make feature vectors. The feature vectors are converted into feature sequences by vector quantization. Then, the feature sequences are input into the classification system constructed by DHMM to identify the PQD types. Based on the classification results, a method which is used to calculate the PQD index is presented. Numerical simulation results show that the proposed classification method has high classification correct ratio in strong noise condition, and the training is easy to converge.