基于LS-SVR的高速列车车内声品质主观评价

Study on subjective evaluation interior sound quality of high-speed train based on LS-SVR

针对传统主观评价方法用于高速列车车内噪声信号评价存在效率低、评价时间长、可重复性差等问题,本文运用心理声学参数表征主观评价结果,从而建立评价模型,以探究两者之间的复杂非线性关系,用最小二乘法对原始信号进行点拟合,反向筛选短时噪声信号,对筛选出的声音信号分别计算声压级、响度、粗糙度、尖锐度、抖动度和AI指数6种心理声学参数的算数平均值。以“烦躁度”作为主观评价指标,用单位间隔的语义细分法作为主观评价方法来主观评价实验得到评价结果,以样本集中样本的心理声学参数作为输入,主观评价结果作为输出,建立高速列车车内声品质的最小二乘法—支持向量机回归(LS-SVR)预测模型,同时设置测试集对LS-SVR模型进行预测精度检验,并与线性回归模型进行比较。检验结果发现,在测试集中模型预测误差率较低,比多元线性回归模型预测精度高,说明LS-SVR模型针对高速列车车厢内声品质评价预测是有效适用的。

Aimed at the problems of low efficiency,long evaluation time and poor repeatability when the traditional subjective evaluation method is used to evaluate the noise signal in high-speed train,psychoacoustic parameters are used to characterize the subjective evaluation results to establish an evaluation model to explore the complex nonlinear relationship.Point fitting of original signal is conducted with the least square method to reverse screen the short-term noise signal,and the average sound pressure level,loudness,roughness,sharpness of six psychoacoustic parameters of jitter and AI index are calculated.Taking irritability as the subjective evaluation index and the semantic segmentation method of unit interval as the subjective evaluation method,the subjective evaluation experiment is carried out to obtain the evaluation results.Taking the psychoacoustic parameters of the samples in the sample set as the input and the subjective evaluation results as the output,the LS-SVR prediction model of high-speed train interior sound quality is established.At the same time,the test set is used to test the prediction accuracy of LS-SVR model.Compared with the linear regression model,the test results show that the prediction error rate of the model in the test set is lower and the prediction accuracy is higher than that of the multiple linear regression model.It shows that LS-SVR model is effective and applicable for the evaluation and prediction of sound quality in high-speed train compartment.