基于集合经验模态分解的交流航空故障电弧识别  

Identification of AC aviation arc fault based on ensemble empirical mode decomposition

作　　者：高飞[1] 董伟 桂美景 张俊民[1] GAO Fei;DONG Wei;GUI Mei-jing;ZHANG Jun-min(School of Automation Science and Electrical Engineering,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191

出　　处：《电工电能新技术》2020年第4期73-80,共8页Advanced Technology of Electrical Engineering and Energy

基　　金：国家自然科学基金项目(51877004,51477004)。

摘　　要：故障电弧作为一种破坏性强,识别难度大的电路故障,给飞机的安全带来很大的威胁。然而目前航空业应用的电弧识别方法远不能达到要求。所以本文提出了一种将集合经验模态分解和LM优化的BP神经网络相结合的交流航空故障电弧识别方法。首先建立串联和并联交流电弧实验模拟平台,采集电弧电流波形,分析波形动态特性。将波形进行集合经验模态分解,选取差别明显模态分量作为故障特征分量。计算故障特征分量的能量熵作为交流故障电弧的特征量,输入到LM算法优化的BP神经网络,进行识别。结果表明,识别率达到90%以上,较好地识别出了航空故障电弧。Fault arc is a destructive and difficult to discriminate circuit fault,which poses a great threat to the safety of the aircraft.However,the current arc identification method applied in the aviation industry is far from fulfilling the requirements.Therefore,this paper proposes an AC aviation fault arc identification method that combines Ensemble Empirical Mode Decomposition and LM optimized BP neural network.Firstly,a series and parallel AC arc experimental simulation platform is established to collect the arc current waveform and analyze the dynamic characteristics of the waveform.The waveform is subjected to collective Ensemble Empirical Mode Decomposition,and the distinct modal component is selected as the fault feature component.The energy entropy of the fault feature component is calculated as the feature quantity of the AC fault arc,and is input to the BP neural network optimized by the LM algorithm for identification.The results show that the recognition rate reaches to more than 90%,and the aviation fault arc is well recognized.

关 键 词：交流航空故障电弧 集合经验模态分解 电弧动态特性 能量熵 BP神经网络 

分 类 号：TM247[一般工业技术—材料科学与工程]