基于曼哈顿平均距离和余弦相似度的配网单相接地故障定位

Fault location of single-phase grounding fault based on Manhattan average distance and cosine similarity in distribution network

针对有源配电网谐振接地系统发生单相接地故障时现有定位算法灵敏度不足的问题,文中提出一种基于曼哈顿平均距离和余弦相似度的故障定位方法。首先,明确分布式电源(distributed generator,DG)的中性点接地方式,并分析了DG接入的影响;其次,对故障馈线各区段两端的暂态零序电流进行小波包分解,利用曼哈顿平均距离计算其幅值相似度,并基于余弦相似度比较其波形差异,从而构造综合相似度判据,与设定阈值进行比较即可区分出故障区段;最后,利用DIgSILENT PowerFactory搭建仿真模型。仿真结果表明该方法在不同故障位置、不同故障时刻以及不同过渡电阻下均能实现准确定位。

Aiming at the problem of insufficient sensitivity of existing fault location methods when single-phase grounding fault occurs in resonant grounding system of active distribution network,a fault location algorithm based on Manhattan average distance and cosine similarity is proposed.Firstly,the neutral point grounding method of distributed generator(DG)is defined and the influence of DG access is analyzed.Secondly,the transient zero-sequence current at both ends of each section of the fault feeder is decomposed by wavelet packet.Then the amplitude similarity is calculated by Manhattan average distance,and the waveform difference is compared based on cosine similarity.As a result,a comprehensive similarity criterion is constructed,and the fault section can be distinguished by comparing with the set threshold.Finally,the simulation model is built in DIgSILENT PowerFactory.Simulation result shows that the proposed method can locate faults accurately under different fault locations,different fault moments and different transition resistances.