科技创新支持向量机在汽车自动变速器故障识别中的应用.pdfVIP

学兔兔 汽 车 工 程 2012年(第34卷)第3期 Automotive Engineering 2o12(Vo1．34)No．3 2012048 支持向量机在汽车自动变速器故障识别中的应用术 杜灿谊 ，杨翠丽 ，潘 威 (1．广东技术师范学院机电学院，广州 510635；2．华南理工大学机械与汽车工程学院，广州 510640) [摘要] 本文中提出基于支持向量机的汽车自动变速器故障识别方法。首先利用设计的虚拟测试系统采集 自动变速器3种状态下的运行数据，再对数据进行整理和筛选，提取合适的数据作为训练样本，然后设计基于支持 向量机的多值分类器进行故障识别，最后与基于BP人工神经网络的诊断方法进行对比。结果表明，基于支持向量 机的故障识别方法具有更快的收敛速度和更强的分类能力，适用于汽车自动变速器实时故障识别和诊断。 关键词：自动变速器；故障识别；支持向量机；神经网络 The Application of Support Vector Machine to the Fault Identificati0n of Vehicle Automatic Transmission Du Canyi ，-．Yang Cuili Pan Wei 1．School ofElectromechanicalEngineering，GuangdongPolytechnicNormal University，Guangzhou 510635； 2．School ofMechanical and Automotive Engineering，South China University ofTechnology，Gaangzhou 510640 [Abstract] A fault identification technique for vehicle automatic transmission based on support vector ma- chine(SVM)is presented in this paper．Firstly，a virtual testing system designed is used to collect，process and screen the operation data of an automatic transmission under three statuses with appropriate data extracted as training samples．Then，a SVM—based multi—class classifier is designed to identify faults．Finally，the outcomes are COB- pared with that using BP neural-network diagnostics．The results show that the fault identification based on SVM converges faster with stronger classification ability，suitable for real—time fault identification of vehicle automatic transmission． Keywords：automatic transmission；fault identification；support vector machine；neural network 术来诊断汽车变速器故障进