基于自适应S变换与截断紧致奇异值分解的局部放电源复杂染噪特征提取方法

Feature Extraction of Partial Discharge Source with Complex Noise Based on Adaptive S-Transform and Truncated Compact Singular Value Decomposition

针对局部放电(PD)源信号复杂染噪,导致PD源特征提取较为困难的问题,提出一种基于自适应S变换与截断紧致奇异值分解(TCSVD)的PD源复杂染噪特征提取方法。首先,对S变换进行了优化改进,应用于PD源获取时频域矩阵,自适应的滤除窄带干扰信号,提取局部放电有用时频信号;其次,利用紧致奇异值分解对提取的时频矩阵进行分解;然后,提出拟合求导法寻找到奇异值阈值参数并对奇异值进行截断,从而滤除PD源中的白噪声信号;最后,通过理论仿真与现场测试对该文所提出的PD源特征提取方法进行了验证分析。实验结果表明,该特征提取方法对复杂染噪的PD信号有很好的特征提取能力,能够有效地提取局部放电信号的有用信息。

To solve the problem that the features of the partial discharge(PD) source are difficult to extract because the PD source signal is polluted by the complex noise, a PD source complex noise feature extraction method is proposed based on the adaptive S-transform and the truncated compact singular value decomposition(TCSVD). First, the S-transform is optimized and improved, and then applied for the PD source to obtain the time-frequency domain matrix. The narrow-band interference signal is filtered adaptively, and the useful time-frequency signal of partial discharge is extracted. Second,compact singular value decomposition is utilized to decompose the extracted time-frequency matrix.Then, the fitting derivative method is proposed to find the singular value threshold parameters and truncate the singular value, the white noise signal in PD source is filtered. Finally, the proposed PD source feature extraction method is verified and analyzed by theoretical simulation and field test. The results indicate that the feature extraction method has well feature extraction ability for the PD signal with complex noise, and can be utilized to effectively extract useful information of PD signal.