基于动态核独立成分统计散度的高阻故障检测

High Impedance Fault Detection Based on Dynamic Kernel Independent Component Statistical Divergence

为提高配电网运行的可靠性,提出了一种基于动态核独立成分统计散度的配电网高阻故障检测方法。将滞后系数和核独立成分分析结合得到动态核独立成分分析算法,将Wasserstein散度引入T^(2)统计量形成动态独立成分散度判别统计量,对动态核多元独立成分间的概率分布差异进行度量,最后将统计量幅值和越限时间构成判别条件用于对统计量度量结果的判别,实现高阻故障检测。通过仿真结果可知所提方法能够准确地检测出不同类型的高阻故障,具有有效性和可行性。

In order to improve the reliability of distribution network operation,a method of detecting high resistance fault based on dynamic kernel independent component statistical divergence was proposed.Firstly,the lag coefficient was combined with the kernel independent component analysis to obtain the dynamic kernel independent component analysis algorithm.Then,Wasserstein divergence was introduced into the T?statistic to form a discriminant statistic of dynamic independent component divergence,thus measuring the difference of probability distribution among dynamic kernel multivariate independent components.Finally,the discriminant condition composed of the statistic amplitude and the exceed time was used to distinguish the results of the statistic metrics,thus achieving high resistance fault detection.Simulation results show that the proposed method can accurately detect different types of high resistance faults,and has higher effectiveness and feasibility.