基于EEMD-AESSAICA的电驱动单通道噪声源分离识别  

Separation and Identification of Single-channel Noise Source of Electric Drive Based on EEMD-AESSAICA

作　　者：甄帅 李基芳 刘海[1] 李维肖 杨瑞[1] ZHEN Shuai;LI Jifang;LIU Hai;LI Weixiao;YANG Rui(Tianjin Key Laboratory of Power Transmission and Safety Technology for New Energy Vehicles,Hebei University of Technology,Tianjin 300130,China;CATARC New Energy Vehicle Test Center(Tianjin)Co.,Ltd.,Tianjin 300399,China)

机构地区：[1]河北工业大学天津市新能源汽车动力传动与安全技术重点实验室,天津300130 [2]中汽研新能源汽车检验中心(天津)有限公司,天津300300

出　　处：《噪声与振动控制》2023年第4期157-163,261,共8页Noise and Vibration Control

基　　金：2021年度河北省引进留学人员资助项目(C20210505)。

摘　　要：为准确分离识别电驱动总成的噪声源,提出一种集合经验模态(EEMD)与改进樽海鞘的独立分量分析(AESSAICA)方法。首先针对传统盲源分离方法存在收敛速度慢、分离精度低的问题,提出基于改进樽海鞘算法的盲源分离算法,提出自适应领导者数目的精英方向学习策略,其能够平衡全局探索和局部开发矛盾、加快收敛速度。其次通过仿真实验验证该方法比传统独立分量算法在分离效果上提升4.38%,能够提高分离效率,提升分离结果质量;然后联合EEMD和AESSAICA算法提出的单通道盲源分离方法,同时验证其相似系数在0.96以上;最后采用该方法分离识别电驱动主要噪声分量。结果表明上述方法能够有效识别电驱动各独立噪声源,通过减速器噪声实验验证最大分离误差为1.1%,分离结果的准确性得到证明。In order to accurately separate and identify the noise source of the electric drive assembly,an ensemble empirical mode(EEMD)and improved independent component analysis of salps(AESSA-ICA)hybrid method is proposed.Firstly,in view of the problem of slow convergence speed and low separation accuracy of traditional blind source separation methods,a blind source separation algorithm based on the improved salps algorithm and an elite direction learning strategy based on the number of adaptive leaders are proposed,which can balance the contradiction between the global exploration and local development,and speed up the convergence rate.Secondly,through simulation experiments,it is verified that the separation effect of this method is 4.38%higher than that of the traditional independent component algorithm.So,this method can improve the separation efficiency and the quality of the separation results.Thirdly,the single-channel blind source separation method proposed by the EEMD and AESSAICA algorithms is applied,and its similarity is verified at the same time.The similarity coefficient is above 0.96.Finally,this method is used to separate and identify the main noise components of the electric drive.The results show that this method can effectively identify the independent noise sources of the electric drive,and the maximum separation error of 1.1%is verified by the noise experiment of the reducer,which proves the accuracy of the separation results.

关 键 词：声学 电驱动动力总成 噪声源识别 樽海鞘算法 盲源分离 

分 类 号：TB533.2[理学—物理]