基于小波-隐马尔可夫的波形异常扰动类型识别研究

Research on Recognition of Waveform Abnormal Disturbance Types Using Wavelet-Hidden Markov Models

针对波形异常扰动的识别问题,提出一种基于小波-隐马尔可夫的分类方法,利用小波变换提取波形异常扰动特征。波形异常扰动利用小波变换被分解成多分辨率小波域,其小波系数由HMM构建模型。基于此模型,结合最大似然实现波形异常扰动的分类识别,并在7200 V配电线上的实际波形异常扰动数据进行分类,并通过后处理对结果进行调整。实验结果表明,该方法在电力工业的507个真实波形异常扰动事件分类的正确率为99%。

The Wavelet-based HMM is designed to recognize waveform abnormal disturbances,where the wavelet transform is adopted to extract waveform abnormal disturbance features.The waveform abnormal disturbance is decomposed into multi-resolution wavelet domains,and the wavelet coefficients are modeled by the HMM.Based on this modeling,the maximum likelihood classification is applied to classify actual waveform abnormal disturbance data recorded on a 7200 V distribution line,and the result is tuned by post-processing.The experimental results show that the method has a correct rate of 99%for the classification of 507 real waveform abnormal disturbance in the power industry.