汽车工程 ›› 2025, Vol. 47 ›› Issue (9): 1742-1751.doi: 10.19562/j.chinasae.qcgc.2025.09.010

• • 上一篇    

基于特征熵奇异值分解的振动传递路径分析方法

孔令琛,袁晓磊(),赵轩,余强,周辰雨,黄榕   

  1. 长安大学汽车学院,西安 710064
  • 收稿日期:2025-01-10 修回日期:2025-02-24 出版日期:2025-09-25 发布日期:2025-09-19
  • 通讯作者: 袁晓磊 E-mail:yuanxiaolei@chd.edu.cn
  • 基金资助:
    国家自然科学基金面上项目(52372375);国家自然科学基金(52202463);国家资助博士后研究人员计划(GZC20241449)

A Vibration Transfer Path Analysis Method Based on Entropy Singular Value Decomposition

Lingchen Kong,Xiaolei Yuan(),Xuan Zhao,Qiang Yu,Chenyu Zhou,Rong Huang   

  1. School of Automobile,Chang’an University,Xi'an 710064
  • Received:2025-01-10 Revised:2025-02-24 Online:2025-09-25 Published:2025-09-19
  • Contact: Xiaolei Yuan E-mail:yuanxiaolei@chd.edu.cn

摘要:

工况传递路径分析(OTPA)是识别复杂机械系统关键振动传递路径的重要手段,但是结构间的信号串扰严重影响了OTPA的分析精度。本文基于特征熵理论改进了传统截断奇异值分解(TSVD)方法,构建了基于特征熵奇异值分解理论的工况传递路径分析(OTPA-ESVD)模型,该模型首先利用奇异值熵对该特征值包含的信息进行量化评价,再遵循高贡献配对原则设计了更为全面的奇异值选取方法,克服了传统TSVD过于依赖工程经验的问题。本文设计了以某款电动汽车为对象的实车试验,并以基于中值奇异值分解的工况传递路径分析模型(OTPA-MSVD)为对比,对OTPA-ESVD模型的有效性开展验证。结果表明,在不同工况下,OTPA-ESVD模型分析效果均优于OTPA-MSVD模型,OTPA-ESVD模型的重构目标信号更加接近真实测量信号,在误差均方根值和关键频率幅值对比方面分别有不低于13.8%和8.1%的优化效果。

关键词: 电动汽车, 工况传递路径分析, 串扰消除, 特征熵

Abstract:

Operational transfer path analysis (OTPA) is an important way to identify the key transfer paths of complex mechanical systems, but the signal crosstalk between structures seriously affects the analysis accuracy of OTPA. In this paper, the traditional truncated singular value decomposition (TSVD) method is improved based on the entropy theory, and the operational transfer path analysis model based on the entropy singular value decomposition method (OTPA-ESVD) is constructed. In the proposed model, the singular value entropy is used to evaluate the information of eigenvalues, and then the principle of high-contribution pairing is used to design a more comprehensive singular value selection method, which overcomes the dependence on engineering experience of conventional TSVD. In this paper, an actual electric vehicle test is designed to verify the OTPA-ESVD model by comparing it with the operational transfer path analysis model based on the median singular value decomposition (OTPA-MSVD). The results show that the performance of the OTPA-ESVD model is better than the OTPA-MSVD model under different driving conditions, and the reconstructed target signal of the OTPA-ESVD model is closer to the actual measured signal, with optimization of no less than 13.8% and 8.1% in the comparison of the error root mean square value and the amplitude of the key frequency, respectively.

Key words: electric vehicle, operational transfer path analysis, crosstalk elimination, entropy