基于CEEMD和GG聚类的电能质量扰动识别

提出一种基于完备总体经验模态分解(complete ensemble empirical mode decomposition,CEEMD)和GG(gath-geva)聚类的电能质量扰动识别方法。CEEMD是一种对EEMD(ensemble empirical mode decomposition)的改进算法,其特点是向原始信号中以正负成对的形式加入白噪声,有利于减少重构信号中残余的辅助噪声;且在分解的每一个阶段都加入特殊噪声,计算一个唯一残差以得到每个IMF,因此分解的结果是完整的,优于EEMD。CEEMD不仅有效解决了EEMD的模态混叠的问题,同时也保留了EEMD处理非平稳信号的优势,再将CEEMD 分解的IMF分量的互近似熵值作为特征向量输入到GG模糊分类器中进行电能扰动的分类识别。为了验证该方法的有效性,进行了仿真和实测实验,结果表明,该方法有较好的频谱分离效果,且仅需要较少的迭代次数,减轻了计算成本。

Power Quality Disturbance Identification Based on CEEMD and GG Clustering

A method of power quality disturbance identification based on CEEMD and GG clustering is proposed. CEEMD is a kind of CEMD improved algorithm, its characteristic is putting positive and negative pairs of white noise into the original signal, helps to reduce the residual noise in the auxiliary signal reconstruction; and adding special noise in every stage of decomposition, calculating a unique residual to get each IMF, the decomposition result is complete. Superior to EEMD, CEEMD not only effectively solve the problem of EMD mode mixing, but also retains the advantages of EMD processing non-stationary signals. The CEEMD decomposition of the IMF component of the cross approximate entropy as feature vector is inputted into the GG fuzzy classifier to classify the electric disturbance, The simulational experimental results show that this method has better spectrum separation effect, and needs less iteration times, reduce the computational cost.