摘要
提出了采用支持向量机对CVT夹紧力控制阀进行质量分类的方法,并设计了相应的提取夹紧力控制阀性能特征参数的试验方案与夹紧力控制阀SVM多类分类器。利用试验得到的性能参数对SVM分类器进行训练,然后分别使用训练成熟的SVM分类器与RBF神经网络分类器对夹紧力控制阀进行质量分类。结果表明,采用SVM分类器的分类准确率明显高于RBF神经网络分类器。
A method of quality classification on CVT clamping force control valve with support vector machine （SVM） was proposed. Firstly corresponding test plan for feature parameters extraction from clamping force control valve and the clamping force control valve SVM multi-class classifier were designed. Secondly, based on the feature parameters acquired from test, the SVM classifier was trained to maturity. Lastly, clamping force control valves were classified based on quality with SVM classifier and RBF neural network classifier respectively. Results showed that the accuracy of SVM classifier was better than RBF neural network classifier.
出处
《汽车技术》
北大核心
2009年第11期4-7,共4页
Automobile Technology
关键词
CVT
支持向量机
夹紧力控制阀
质量
CVT
Support vector machine
Clamp force control valve
Quality
