基于多特征融合神经网络的串联电弧故障识别技术

Series Arc Fault Detection Technology Based on Multi-feature Fusion Neural Network

传统低压保护装置如低压断路器、熔断器等无法有效检测出由于接触不良、绝缘失效等导致的串联电弧故障,因此如何准确检测串联电弧故障成为目前研究的热点问题。为此,采用基于电流波形的检测方法展开深入研究,通过搭建电弧故障平台模拟串联电弧故障,获得了不同负载下正常和电弧故障的数据;然后在此基础上建立了多特征融合的神经网络算法,并利用mini-batch梯度下降、指数衰减的学习率和Adam的优化算法对模型进行了优化。研究结果表明:所提算法的查准率及查全率分别能达到98%和99%,相比于支持向量机和BP神经网络算法具有更高的识别率。研究为串联电弧故障识别提供了一种新的算法,对于该方向的研究拓展了新的思路。

Traditional low-voltage protection devices such as low-voltage circuit breakers,fuses and other protection devices cannot effectively detect series arc faults which are caused by poor contact or insulation failure,therefore,how to accurately detect series arc faults has become a hot issue in current research.In this paper,the detection method based on current waveform is used for in-depth research.By building an arc fault platform to simulate series arc faults,the data of normal and arc faults under different loads are obtained.On this basis,a neural network algorithm for multi-feature fusion is established.The model is optimized by using mini-batch gradient descent,exponential decay learning rate,and Adam’s optimization algorithm.The research results show that the accuracy and recall rate of the algorithm in this paper can reach 98%and 99%,respectively,which has a higher recognition rate than the SVM and BP neural network algorithms.The research provides a new algorithm for series arc fault identification,and expands new ideas for the research in this direction.