基于模态分解与特征匹配的串联故障电弧识别方法研究

串联故障电弧具有隐蔽性强、短时释放热量大等特点,过流型断路器难以及时发现或采取动作,极易引发电气火灾,造成重大损失和人员伤亡,因此实现建筑内串联故障电弧的快速可靠识别与监测具有重大意义;按照线路负载类型对电气线路高频电气参数运行数据进行分析,利用结合串联电弧故障特征的互补集合经验模态分解(CEEMD,complementary ensemble empir-ical mode decomposition)方法,实现对电气线路串联电弧故障的识别;经实验验证,并与灰度梯度共生矩阵与支持向量机(GL-GCO-SVM)、时域可视卷积神经网络(TDV-CNN)等方法识别结果进行对比效果更好,识别准确率达到94.8％及以上.

Research on identification method of series arc faults based on modal decomposition and feature matching

Series fault arc has the characteristics of strong concealment and large heat release in short time. It is difficult for overcurrent circuit breakers to detect or take action in time, which is easy to cause electrical fire and cause significant losses and casualties. Therefore, it is of great significance to realize the rapid and reliable identification and monitoring of series fault arc in buildings. According to the line load type, the operation data of high frequency electrical parameters of electrical lines are analyzed, and the CEEMD combined with the characteristics of series arc fault is used to realize the identification of series arc fault of electrical lines. Compared with the recognition results of GLGCO-SVM and TDV-CNN, the recognition accuracy reaches 94.8 % and above.