基于时频域分析方法与分类器算法相结合的Shunt辨识  

Identification of Shunt Phenomenon in Vehicle Transmission Systems Based on the Combination of Time Frequency Domain Analysis Method and Classifier Algorithms

作　　者：蒋敏凯 吴光强[1,2] 彭尚 陈凯旋 Jiang Minkai;Wu Guangqiang;Peng Shang;Chen Kaixuan(School of Automotive Studies,Tongji University,Shanghai 201804;Institute of Industrial Science,the University of Tokyo,Tokyo 153-8505)

机构地区：[1]同济大学汽车学院,上海201804 [2]东京大学生产技术研究所,东京153-8505

出　　处：《汽车技术》2024年第11期57-62,共6页Automobile Technology

基　　金：国家自然科学基金项目(52075388)。

摘　　要：为了准确辨识汽车抖动和半轴扭矩振荡(Shunt)现象,以传统汽车为载体,结合短时傅里叶变换方法分析补充了Shunt的定义并优化标签数据集,使用决策树、支持向量机和随机森林算法,以发动机转速、变速器输入轴转速等传感器信号作为输入来识别Shunt。结果表明,与传统的模型构建方法相比,该方法降低了模型构建的难度和成本,后续可用于可解释的机器学习来解释模型。In order to accurately identify vehicle vibration and shaft torque oscillation(Shunt)phenomenon,this paper analyzes and supplements the definition of Shunt phenomenon and optimizes the label data set by combining the short-time Fourier transform method with the traditional vehicle as the carrier,and uses the decision tree,support vector machine and random forest algorithm to identify Shunt with the engine speed,transmission input shaft speed and other sensor signals as inputs.The results show that,compared with the traditional model construction method,this method reduces the difficulty and cost of model construction,and can be used for interpretable machine learning to explain the model.

关 键 词：Shunt辨识 短时傅里叶变换 决策树 随机森林 支持向量机 

分 类 号：TB534[理学—物理]