混合噪声总烦恼度客观评价及实验研究

Evaluating Experimentally Accumulation of Combined Noise Annoyance

借助标准音效资源,依据所提出的混合噪声组合方式,人工合成3种单一类型车辆噪声"同时作用"下混合而成的道路交通噪声样本。在此基础上,设计完成混合噪声总烦恼度主观评价实验,获得全部单一与混合噪声样本的烦恼度评价值。研究发现:(1)从模型计算值与主观测量值的相关性角度考察,矢量叠加模型最好,主导源模型次之,线性回归模型较差;(2)从2类误差指标角度衡量3种模型的评价性能可知:主导源模型的预测误差最小,线性回归模型较大,矢量叠加模型预测误差最大,(3)矢量叠加模型的总烦恼度计算值与主观测量值相关性较高,但预测误差偏大。

Aim. To our knowledge, there does not exist any paper in the open literature on experimental evaluation. Subsection 1.1 of the full paper briefs three classical models： （1） strongest component, （2） vector summation, （3） linear regression. Subsection 1.2 takes from literature eqs. （5） and （6） as criteria for evaluating respectively model prediction performances. Subsection 2.1 designs our experiments. Section 3 gives Table 6 to show that, judging from the available correlation coefficient criterion, the vector summation model has the best performance. It also gives Table 7 to show that, judging from either of the two criteria （eqs. 5 and 6）; the strongest component model is the best.