摘要
利用极限梯度提升(XGBoost)算法建立特种车车内声品质预测模型。首先,进行车内噪声的采集试验并进行试验数据挑选,将其处理成68个可以进行主观评价试验的有效声音样本,然后计算声音样本的客观参数,并分析各参数随工况的变化趋势。以客观参数作为声品质预测模型输入,主观评价预测值为输出,得出预测值与实际主观评价值的平均相对误差为2.43%,相关性系数为0.943,表明根据XGBoost预测模型所得结果与主观评价一致。最后通过分析声品质客观参数的特点得到客观参数对主观分数的影响权重。
Sound quality prediction model of the special vehicles is established based on an extreme gradient boosting(XGBoost)algorithm.The inside-vehicle noise is collected by test,and the test data are selected and divided into 68 valid sound samples that can be applied for subjective evaluation tests.Then,the objective parameters of the sound samples are calculated,and the variation trend of each parameter with the working conditions is analyzed.With the objective parameters as the input of the sound quality prediction model,and the subjective evaluation prediction value as the output,it is obtained that the average relative error between the prediction value and the actual subjective evaluation value is 2.43%,and the correlation coefficient is 0.943,indicating that the XGBoost prediction model is consistent with the subjective evaluation.Finally,according to the characteristics of the objective parameters of sound quality,the influence weight of the objective parameters on the subjective score is obtained.
作者
张勇
彭沸潭
杨鄂川
任克琳
欧健
ZHANG Yong;PENG Feitan;YANG Echuan;REN Kelin;OU Jian(School of Vehicle Engineering,Chongqing University of Technology,Chongqing 400054,China)
出处
《噪声与振动控制》
CSCD
北大核心
2023年第3期161-166,211,共7页
Noise and Vibration Control
基金
重庆市巴南区科技计划资助项目(2019TJ09)。
关键词
声学
车内噪声
心理声学
声品质预测
XGBoost算法
权重
acoustics
vehicle interior noise
psychoacoustics
sound quality prediction
XGBoost algorithm
weight
