支持向量机在汽车加速车内声品质预测中的应用

摘要

Support vector machine ( SVM) method is applied to evaluate vehicle interior sound quality dur-ing acceleration. With the objective evaluation parameters, including loudness, sharpness, roughness and articula-tion index etc. of noise samples as inputs, subjective annoyance as output, a prediction model for the interior sound quality of accelerating vehicle is set up by using support vector machine ( SVM) regression method. The results of comparison show that the SVM-based prediction model for the interior sound quality of accelerating vehicle can more accurately reflect the nonlinear mapping relationship between objective evaluation parameters and subjective annoy-ance, compared with multivariate linear regression model.%应用支持向量机方法对汽车加速时车内声品质进行预测。以噪声样本的响度、尖锐度、粗糙度、AI指数等客观评价参量作为输入因子，主观烦躁度评价结果作为输出因子，利用支持向量机回归方法建立了汽车加速车内声品质的预测模型。对比结果表明，与多元线性回归模型相比，基于支持向量机的汽车加速车内声品质预测模型能够更准确地反映客观评价参量与主观烦躁度之间的非线性映射关系，预测精度更高。