基于模态分解的低压串联电弧故障特征提取方法比较

Comparison of Low Voltage Series Arc Fault Feature Extraction Methods Based on Mode Decomposition

低压交流系统串联电弧电流的非线性、非平稳和随机等特点给故障特征提取和检测带来极大困难,同时以包络线分析为基础的模态分解在非平稳信号分析中展现了良好效果。鉴于模态分解方法的优异效果以及串联电弧故障检测的实际困难,首先对目前较为成熟的经验模态分解(empirical mode decomposition,EMD)等6种模态分解方法进行了系统梳理,并深入分析了该系列方法在电弧故障信号分析和特征提取中的适用性和有效性。然后,通过实测电弧电流的分解实验和特征计算实验,从不同角度探讨了模态分解算法在电弧电流特征提取和故障检测中的优势与不足。最后,对未来可能的研究方向做了展望。

The nonlinear,non-stationary,and random characteristics of series arc current in low-voltage AC systems bring great difficulties to fault feature extraction and detection.Meanwhile,mode decomposition methods based on envelope analysis show a good effect in non-stationary signal analysis.Considering the excellent effect of the mode decomposition methods and the practical difficulties of series arc fault detection,six mature mode decomposition methods,such as empirical mode decomposition(EMD),were systematically combed,and the applicability and effectiveness of these methods in arc fault signal analysis and feature extraction were thoroughly analyzed.Then,through the decomposition experiments and feature calculation experiments of measured arc currents,the advantages and disadvantages of mode decomposition methods in arc current feature extraction and fault detection were discussed from different perspectives.Finally,future research direction of mode decomposition were prospected.