基于VMD与改进DAE人车地震动包络信号识别算法  

Recognition Algorithm of Man Vehicle Ground Motion Envelope Signal Based on VMD and Improved DAE

作　　者：刘文杰[1] 邹瑛珂 张珊 贾云飞[2] LIU Wen-jie;ZOU Ying-ke;ZHANG Shan;JIA Yun-fei(Southeast University-Monash University Joint Graduate School(Suzhou),Suzhou 215127,China;Nanjing University of Technology,College of Mechanical Engineering,Nanjing 210094,China)

机构地区：[1]东南大学苏州联合研究生院,苏州215127 [2]南京理工大学机械工程学院,南京210094

出　　处：《科学技术与工程》2022年第24期10379-10387,共9页Science Technology and Engineering

基　　金：江苏省国家电网公司2020年科技项目(J2020061)。

摘　　要：为解决在野外环境中使用传统模式识别方法对低信噪比(signal-noise rate, SNR)的人车地震动信号进行分类时应用不便,效果不佳的问题,提出了通过基于包络检波、变分模态分解(variational mode decomposition, VMD)和改进的深度自编码器(deep auto-encoder, DAE)的特征提取算法。首先对目标的地震动信号进行希尔伯特变换以获取信号的平滑包络线,然后进行变分模态分解得到本征模函数(intrinsic mode function, IMF)信号,并利用皮尔森相关系数对分解得到的IMF信号进行筛选,之后将相关度较高的分量加权为高信噪比的中间信号,再使用改进的深度自编码器对其进行特征提取,最后使用泛化性能好的随机森林算法充当分类器,从而实现对人车目标的识别和分类。结果表明:所提算法有效缓解了其他传统算法的部分缺陷,综合识别正确率有所提高,且更加方便应用。In order to solve the problems of inconvenient application and not ideal effect when using traditional pattern recognition methods to classify low signal-noise rate(SNR) ground motion signals in field environment, a feature extraction algorithm based on variational mode decomposition(VMD), and improved deep auto-encoder(DAE) were proposed. First of all, Hilbert transform was used in getting smooth envelope of the signal from the target of the ground motion signal. And then using the envelope variational mode decomposition to get IMF signal. Next setup, using the Pearson correlation coefficient to screen decomposed intrinsic mode function(IMF) signal. Then weighting the signal which had higher correlation to get high SNR signal. Besides, the improved depth self encoder was used to extract its features. Finally, the random forest algorithm with good generalization performance was used as the classifier to realize the recognition and classification of human-vehicle targets. The results show that the proposed algorithm effectively alleviates some defects of other traditional algorithms. And the accuracy of the proposed algorithm is higher than other traditional algorithms and more convenient to use.

关 键 词：变分模态分解(VMD) 深度自编码器(DAE) 相关系数 随机森林 人车地震动信号 

分 类 号：O235[理学—运筹学与控制论]