一种模拟人耳实现噪声分类的方法  

A human ear simulation model for noise classification

作　　者：郭伟[1] 左曙光[1] 李徐钢[1] 

机构地区：[1]同济大学,上海201804

出　　处：《振动与冲击》2010年第11期152-155,共4页Journal of Vibration and Shock

基　　金：国家863计划电动汽车重大专项(2008AA501000);上海市曙光计划项目(08SG22)及汉高公司资助

摘　　要：声品质主观评价方法中有基于成对比较的参考评分法,该方法在等级评分之前需要首先利用成对比较法对声音样本进行排序,设立参考噪声。该过程比较繁琐,劳动强度较大。提出通过建立人耳模型,对噪声样本按照声品质的不同属性实现粗略的分类排序,省去原方法中排序这一步骤,直接在每一类中设立参考噪声进行等级评分。选取燃料电池汽车8种行驶工况,24个噪声样本为例,介绍说明该方法的原理及有效性。Reference-Scale Estimation method based on paired comparison is one of sound quality subjective evaluation methods.This method needs to use the paired comparison method to sequence all of sound samples and to establish standard noise sample before magnitude estimation.This process is very cumbersome with great labor intensity.In order to eliminate this step,a human era model was proposed to achieve preliminary noise classification and establish reference noise sample automatically.24 noise samples from 8 driving cycles of a fuel cell vehicle (FCV) were chosen as examples to introduce the main theory and prove its effectiveness.

关 键 词：声品质 主观评价 等级评分法 响度 SOM概率神经网络 

分 类 号：TH212[机械工程—机械制造及自动化] TH213.3