结合滑动窗奇异值分解的EEMD暂态电能质量扰动检测法

为了解决噪声、模态混叠等原因造成提取电能质量扰动信号的时频特征不清晰的问题,根据电能质量扰动信号具有非平稳、不确定性以及周期性强的特点,应用总体经验模态分解(ensemble empirical model decomposition,EEM D)的方法对电能质量扰动信号进行分解,基于滑动窗奇异值分解(singular value decomposition,SVD)数据压缩方法对EEMD分解得到的一系列固模函数(intrinsic mode function,IMF)分量组成的矩阵进行了重构,并对重构后的IMF分量作Hilbert变换降维,提取了扰动信号时间、频率、幅值上的特征。对比传统的EEMD算法,新方法能更加准确定量地提取各个扰动成分的起始时刻、幅值、频率等扰动特征,同时能够有效抵御噪声的干扰,克服了以往只能通过人为选取IMF分量来提取扰动时频特征过于主观的缺点。算例仿真的结果验证了该方法的有效性。

Ensemble Empirical Mode Decomposition Method Based on Singular Value Decomposition for Transient Power Quality Disturbance Features Extraction

In view of the influence of noise and mod mixing on unclear time frequency domain characteristics from extraction of power quality disturbance signal, ensemble empirical model decomposition (EEMD) method is used to decompose the non stationary, uncertain and strong periodical characteristics of the disturbance signal.An improved(EEMD)method based on singular value decomposition (SVD) is introduced to reconstruct intrinsic mode functions (IMF) component matrix,then a Hilbert transformation is applied to the reconstructed IMF component for dimension reduction to extract the characteristics of time, amplitude, and frequency of the disturbance signal.Compared with the conventional EEMD method, the novel method can extract the instantaneous characteristics of time amplitude and frequency more accurately and quantitatively, it can also effectively resist the interference of noise and surmount the previous defect if the IMF component is chosen manually Simulation result validates the effectiveness of the method.