基于LS-SVM算法的加速车内噪声品质评价模型
A LS-SVM-based Model for Sound Quality Evaluation of Vehicle Interior Noise under Accelerating Condition

以汽车加速车内噪声为研究对象，采用最小二乘支持向量机（LS-SVM）方法建立了声品质评价模型。分别以车辆噪声的客观评价结果和主观烦躁度作为模型的输入和输出，在相关分析和显著性检验的基础上，以响度、尖锐度、粗糙度、清晰度指数和A声级为变量建立了LS-SVM声品质评价模型。对未知噪声样本的预测检验表明：预测结果与主观烦躁度具有很高的相关性，预测精度高于多元线性回归方法。所建立的模型具有良好的泛化能力，可用于加速车内噪声品质的预测。
Based on the LS-SVM algorithm, a sound quality evaluation (SQE) model of vehicle interior noise during acceleration is presented in this paper. The objective psychoacoustic parameters and subjective annoyance results are used as the input and output of the model, respectively. Based on the correlation analysis and the significance test, some parameters, such as loudness, sharpness, roughness, AI index and A-weighted sound pressure level, are selected as variables for LS-SVM SQE modeling. The prediction results of unknown samples are high correlated with the subjective annoyance results, which are better than those from the multiple linear regression method. The proposed LS-SVM SQE model, which has performed good generalization ability, can be applied in the sound quality prediction of vehicle interior noise under accelerating condition