车内语言清晰度分析评价及其主动控制技术研究

发布时间：2018-05-28 22:11

本文选题：语言清晰度 + 分析评价　；参考：《吉林大学》2013年博士论文

【摘要】：车内噪声是影响乘车人舒适性、听觉、语言清晰度以及对车内外各种声音讯号识别能力的重要因素。车内噪声水平已经成为汽车乘坐舒适性的重要性能指标之一，它的优劣直接影响汽车产品的竞争力和消费者的购车取向。现有汽车噪声法规和标准中，，多采用A计权声压级作为汽车噪声水平的评价指标，其强制规定了汽车噪声允许达到的声压级限值，但是未深入考虑车内噪声特性。随着汽车工业的进一步发展，对汽车噪声水平和品质有了更高的要求，单一A计权声压级指标已经无法满足，车内噪声品质评价迅速发展起来，它指出了人对噪声的感受由人生理和心理共同影响，反映了噪声受众对噪声的主观感觉与评判。结合本课题组承担的国家自然基金项目《车内噪声品质分析评价及其自适应主动控制方法研究》（项目编号：50975119），开展了汽车噪声品质的分析评价，以及自适应主动控制方法和技术的研究。首先，针对不同车型在典型工况下的车内噪声，进行语言清晰度测试试验，研究用声学客观评价参量对车内语言清晰度进行建模，建立与试验结果相一致的车内语言清晰度客观评价模型。具体通过组织听音人对车内噪声环境背景下的语言清晰度进行实验室主观听音评价，得到车内噪声环境下语言清晰度测试试验结果。同时计算了车内噪声样本的声学客观评价参量，包括有语言清晰度指数、响度、尖锐度、粗糙度、两种不同计权声压级、PSIL、SIL、SIL(3)。采用回归分析方法，建立了与语言清晰度测试试验结果相一致的，以声学客观评价参量为变量的车内语言清晰度客观评价模型。对比分析了建立的多元线性回归和二次多项式车内语言清晰度模型之间的优劣，并进行了模型检验，最终验证了与语言清晰度指数相关的二次多项式车内语言清晰度客观评价模型，具有好的拟合优度和数据检验显著水平。其次，在分析车内噪声特性基础上，对车内噪声按照ERB进行了特征频带划分，利用参数化滤波技术对不同特征频带噪声声压级进行处理，得到了用于声学客观评价参量特征频带灵敏度分析的噪声样本，并计算了噪声样本的各声学客观评价参量，分析确定了不同特征频带噪声声压级的变化对声学客观评价参量的灵敏度，特别分析了车内语言清晰度特征频带灵敏度。语言清晰度指数敏感频带集中在特征频带5及其附近特征频带，其特征频带灵敏度相对变化率曲线呈“漏斗”状。优先改变这些特征频带噪声声压级可显著调节语言清晰度指数的数值大小与车内语言清晰度状况。心理声学客观评价参量响度、尖锐度、粗糙度、A计权声压级的敏感频带集中在前5个特征频带，前5个特征频带车内噪声声压级变化对上述车内噪声心理声学客观评价参量影响显著。为寻求车内语言清晰度主动控制最优目标，进一步分析了车内噪声中低频特征频带成分特性。设计了6个特征频带5种试验水平的试验，并提出了以语言清晰度指数和声音品质客观评价模型得分作为试验指标考察车内语言清晰度优化问题。对60km/h、100km/h、120km/h匀速行驶时车内噪声样本进行正交试验，运用极差分析方法，计算得到了语言清晰度指数和声音品质客观评价模型得分的各特征频带优水平和最优组合。依据试验指标各特征频带的最优组合，通过实验验证了最优组合控制在改善车内语言清晰度和声音品质方面的能力。然后，提出了以FSLLMS为核心的ANE主动控制算法，该算法以ERB进行特征频带划分，能实现选择性抵消或补充不同特征频带噪声声压级。在Matlab/Simulink仿真环境下构建了以该算法为核心的车内语言清晰度自适应主动控制系统模型。针对100km/h匀速行驶时车内噪声，利用车内语言清晰度主动控制系统进行了改进车内语言清晰度和声音品质的主动控制仿真。改进车内语言清晰度，使车内语言清晰度客观评价模型达到最优的主动控制，能有效降低噪声声压级，语言清晰度指数提高了14.3%；改进车内声音品质，使声音品质客观评价模型得分达到最优的主动控制，优化了声音品质的综合表现，优化率达到8.93%。最后，设计开发了具有自主知识产权的车内语言清晰度自适应主动控制器，编写了相应的控制程序，构建了车内语言清晰度自适应主动控制系统。进行了60km/h、100km/h、120km/h三种工况下的车内噪声实车控制试验，分析了控制前后车内语言清晰度和声音品质的变化情况。改进车内语言清晰度，使车内语言清晰度客观评价模型达到最优的主动控制，能有效地降低车内噪声的声压级和响度，提高车内语言清晰度，响度最大降低量达到7.14sone，语言清晰度指数最大提高了17.05%；改进车内声音品质，使声音品质客观评价模型得分达到最优的主动控制，能够改善车内声音品质，其最大改进率达到14.48%，同时车内语言清晰度指数也有所提高，最大改进率达到14.96%。该试验结果令人满意，证明了提出的主动控制算法能有效实现车内声音品质的改善。
[Abstract]:The vehicle interior noise is an important factor that affects the comfort , hearing , speech intelligibility and the recognition ability of various sound signals inside and outside the car . The noise level in the car has become one of the important performance indexes of car ride comfort .

Firstly , the speech intelligibility test is carried out for vehicle interior noise under typical working conditions for different vehicle models . The objective evaluation model of speech intelligibility in the vehicle is established by using the acoustic objective evaluation parameter to model the speech intelligibility in the vehicle . The objective evaluation model of the speech intelligibility in the vehicle is established by using the regression analysis method .

Secondly , on the basis of analyzing the noise characteristics of the vehicle , the characteristics band division of the noise in the vehicle is carried out according to the ERB , and the noise samples are processed by using the parametric filtering technology . The sensitivity of the noise sound pressure level in the vehicle is analyzed . The sensitivity of the noise sound pressure level in the vehicle is analyzed . The sensitivity of the noise sound pressure level in the vehicle is analyzed . The sensitivity band of the characteristic frequency band is concentrated in the characteristic frequency band 5 and the characteristic frequency band of the vehicle . The sensitivity frequency band of the acoustic pressure level of the characteristic frequency band is concentrated in the first five characteristic frequency bands .

In order to find out the optimal target for the active control of speech intelligibility in vehicles , the characteristics of low frequency characteristic frequency band components in vehicle interior noise are further analyzed . Six characteristic frequency bands of five kinds of test levels are designed , and the optimal combination of each characteristic frequency band in vehicle interior noise is obtained by using the language definition index and the objective evaluation model score of sound quality . Based on the optimal combination of the characteristic frequency bands of the test indexes , the paper proves that the optimal combination control has the capability of improving the speech intelligibility and sound quality in the vehicle .

In this paper , an active control algorithm based on FSLLMS is proposed . The algorithm uses ERB to divide the characteristic frequency band , which can realize the selective cancellation or supplement different characteristic band noise sound pressure level . In the simulation environment of Matlab / Simulink , the active control simulation of the speech intelligibility and sound quality in the vehicle is improved . The speech intelligibility in the vehicle is improved , the objective evaluation model of the speech intelligibility in the vehicle is optimized to achieve the optimal active control , the noise sound pressure level can be effectively reduced , and the speech intelligibility index is improved by 14.3 % ;
improve that sound quality in the vehicle , make the objective evaluation model score of the sound quality reach the optimal active control , optimize the comprehensive performance of the sound quality , and the optimization rate reaches 8.93 % .

In the end , the vehicle interior language definition adaptive active controller with independent intellectual property is designed , the corresponding control program is written , and the vehicle interior speech intelligibility adaptive active control system is constructed . The vehicle interior noise real vehicle control test is carried out under the condition of 60km / h , 100km / h and 120km / h .
improve that sound quality in the vehicle , make the objective evaluation model score of the sound quality achieve the optimal active control , can improve the sound quality in the vehicle , the maximum improvement rate is 14.48 % , the language clarity index of the vehicle is improved , the maximum improvement rate reaches 14.96 % , the result of the test is satisfactory , and the proposed active control algorithm can effectively realize the improvement of sound quality in the vehicle .
【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：U467.493;U461.4

【参考文献】