基于高频重构信号与Bayes-XGBoost的低压电弧故障辨识方法研究

针对低压配电系统中单个用电负载支路串联电弧故障辨识困难的问题，提出一种基于高频重构信号和Bayes-XGBoost的低压电弧辨识方法。首先，搭建多支路、多负载类型的低压电弧故障真型实验平台，并采集相关数据。其次，基于故障前后主线路电流高频信号变化规律，提出信号微弱变化叠加法重构故障有效信号。最后，建立适用于单个负载支路电弧故障辨识的XGBoost模型，并采用Bayes算法对模型多个超参数进行优化。实验结果表明，所提方法在多种工况下对单个负载支路电弧故障具有较高的辨识准确率。与6种主流故障分析方法对比，所提方法在精度、训练速度和泛化能力等方面展现出了显著的优越性，有利于实现低压配电系统单个负载支路电弧故障的可靠辨识。

Low voltage arc fault identification method based on high frequencyreconstructed signal and Bayes-XGBoost

There is a problem of difficult identification of series arcing faults in single power-using load branches in low-voltage distribution systems. Thus a low-voltage arc identification method based on high-frequency reconstructed signals and Bayes-XGBoost is proposed. First, a multi-branch, multi-load type LV arc fault true type experimental platform is built and relevant data is collected. Second, based on the main line current high frequency signal change law before and after a fault, the signal weak change superposition method is proposed to reconstruct the effective signal of the fault. Finally, an XGBoost model for single load branch arc fault identification is established, and the Bayes algorithm is used to optimize several hyperparameters of the model. The experimental results show that the proposed method has a high accuracy in identifying arc faults in a single load branch in a variety of operating scenarios. In comparison with the six mainstream fault analysis methods, the proposed method shows significant advantages in terms of accuracy, training speed and generalizability. The proposed method is useful for the reliable identification of arcing faults in single load branches of LV distribution systems.