车内噪声声品质的支持向量机预测  

Interiror sound quality forecast for veheicles based on support vector machine

作　　者：申秀敏[1] 左曙光[1] 李林[1] 张世炜[1] 

机构地区：[1]同济大学汽车学院,上海201804

出　　处：《振动与冲击》2010年第6期66-68,共3页Journal of Vibration and Shock

基　　金：国家863计划电动汽车重大专项(2005AA501000);上海市曙光计划项目(05SG22)资助

摘　　要：对多元线性回归、神经网络和支持向量机的三个预测模型进行了研究。以车内噪声为例,建立了基于以上三种方法的车内噪声声品质预测模型,并采用留一法交叉检验作比较,所构建的支持向量机模型预测精度高于其他两种方法。实验结果同时也表明,支持向量计算法具有较强的稳健性和良好的泛化能力,能够用于车内噪声声品质的预测。Three forecasting models,namely,multiple linear regression model,neural network forecasting model and support vector machine(SVM) forecasting model were inspected.Taking vehicle interior noise as object,vehicle interior sound quality prediction models based on the above three methods were established.The models' outputs were crossly-validated by the leave-one-out method and compared with each other.The results indicate that the prediction accuracy of SVM is superior to that of the other two methods.The experiment indicates that SVM is of better robustness and generalization capability,and is the best method to predict vehicle interior sound quality.

关 键 词：声品质 多元线性回归 神经网络 支持向量机 

分 类 号：TB535[理学—物理] U467.493[理学—声学]