摘要
为了克服现有低压交流串联故障电弧检测算法中,需要线下训练样本仅适用于特定条件下负载的缺点,找到一种多负载适用、运行可靠、便于嵌入式系统执行的检测算法,提出了基于特征模态分量半周期能量差异的检测算法。用EMD分析提取正常电弧和故障电弧电流特征分量,经Hilbert变换得到特征分量的瞬时幅值分布,通过特征IMF的半周期能量与参考值的相对大小差异,构造故障识别标志量,实现故障电弧的判断。采集不同类型负载电流数据进行仿真,结果说明了方法的有效性、通用性。
Most of the existing low voltage arc fauh diagnosis method have disadvantages such as needing offline training of samples, available for only conditioned loads. To overcome these defects and find a multi-load available,response reliable and embedded system friendly detection method, this paper proposed an identification algorithm based on the half-cycle energy of feature mode function from empirical mode decomposition （EMD）. Hilbert transform was applied to work out the instantaneous amplitude distribution of the feature component of arcing current and that of normal. Then compute feature component＇ s half-cycle energy and regard the relative magnitude with a predefined threshold as the criterion. Laboratory tests prove the effectiveness and universality of the method with different kinds of loads.
出处
《电器与能效管理技术》
2015年第21期1-7,17,共8页
Electrical & Energy Management Technology
