高速汽车车内声品质声—振混合主动控制方法研究

The noises generated by wind, tyres and engine are main sound sources of a vehicle running at high speeds. How to reduce and/or eliminate these low-frequency noises for maintaining good interior acoustical environment is an international challenge. Aimming at psyacoustical indices, sunch as loudness, sharpness, roughness and integrated annoyance, etc., this project studies an active vibro-acoustic method for sound quality control of the interior noise of a high-speed vehicle. Based on the interior noise signals measured under different working conditions, the rules of frequency characteristics and annoyance indice of the interior noises changing with vehicle running speed are investigated. Based on the image processing technology, the moving objective recognition method for the ear space coordinates of the passengers is presented. Based on the multi-sensor data fusion method, a transfer function matrix between the measured signals of some assumed noise sensitive points and the sound pressures in the ear area of the passengers will be established for noise signal reconstruction and intelligent sound quality eveluation. Utilzing the inverse effects of piezoelectric materials and the principle of active phase deniosing, an adaptive filtering least mean square algorithm in time and frequency domains, so-called TFD-FXLMS, is developed, which is used for vibro-acoustic combined active sound-quality control (ASQC) at ear positions of the passengers. According to the above theoretical achievements, an ASQC system for vehicle interior noises is developed and further validated by tests. The works done in this project can be extended to other types of high-speed vehicle, and may provide both theoretical and technical supports for effectively interior noise control and acoustical design of vehicles.

汽车高速运行时，风噪、胎噪和发动机噪声成为最主要噪声源，如何对其减小和消除以保持良好的车内声学环境是一个国际难题。本项目针对响度、尖锐度、粗糙度、综合烦躁度等心理声学评价指标，研究车内声品质声—振混合主动控制方法。基于不同工况的噪声信号，研究车内噪声时频特征、综合烦躁度指标随车速的变化规律；以图像处理技术为基础，研究乘员耳侧空间位置坐标动态识别方法；采用多传感器数据融合方法，建立极限车速范围内噪声敏感点信号与乘员耳侧区域声压的传递函数矩阵，实现耳侧区域噪声的时域重构和声品质智能评价；利用压电材料逆效应和有源反相消噪原理，研究时频域最小均方误差自适应滤波算法（TFD-FXLMS），实现乘员耳侧声品质的声—振两级混合主动控制；基于上述理论，研制车内噪声主动控制系统并进行试验验证。项目研究成果可推广到其他类型的高速车辆，为有效控制车辆噪声和车内声学设计提供理论和技术支持。

汽车高速运行时，风噪、胎噪和发动机噪声成为最主要噪声源，如何对其减小和消除以保持良好的车内声学环境是一个国际难题。本项目针对响度、尖锐度、粗糙度、综合烦躁度等心理声学评价指标，提出了一套高速汽车车内声品质声—振混合主动控制方法。主要工作如下：（1）分别采集不同工况的发动机转速、车身关键点振动、发动机舱内噪声、轮胎噪声、高速风噪以及乘员耳侧噪声信号，通过相干分析、去冗余处理和时频分析等方法，获取了乘员耳侧噪声的传递函数矩阵、车内噪声时频特征以及综合烦躁度随车速的变化规律，揭示了车内噪声的产生机理。（2）建立了乘员耳侧空间坐标动态识别数学模型及双目视觉系统。基于RAC两次标定重建了耳侧三维坐标，建立了人耳肤色特征直方图，采用自适应阈值分割方法提取肤色区域完成耳侧区域检测，基于SIFT特征对乘员耳侧进行特征提取与立体匹配，最终采用卡尔曼滤波方法获取了乘员耳侧区域空间的动态坐标。（3）利用前面获取的试验数据和人耳坐标，建立了基于多传感器数据融合的乘员耳侧噪声信号的时域重构模型，提出了BP神经网络（BP-NN）、信号分解优化网络（DBENR）和信号压缩优化网络（CDBENR）3种时域噪声信号智能重构算法，解决了模型输入变量的“维度灾难”问题，实现了乘员耳侧噪声的时域重构和声品质评价。（4）利用压电材料逆效应和有源反相消噪原理，分别提出了针对车身结构振动的离散小波变换（DWT）自适应FxLMS算法（DWT-FxLMS），以及针对车内噪声有源控制的改进变步长FxLMS算法（VS-FxLMS）、归一化频域块FxLMS算法（NFB-FxLMS）和时频域主动噪声均衡算法（TFD-ANE）；综合声振混合主动控制，提出了串联延时间隔LMS算法（HVA-DeLMS）、并联混合控制算法（HVA-FUXLMS）、EMD频域块ANE算法（HVA-EBANE）和FELMS算法（HVA-EBFELMS），实现了乘员耳侧声品质的声—振两级混合主动控制。基于上述理论，研制车内噪声主动控制系统并进行试验验证。项目研究成果可推广到其他类型的高速车辆，为有效控制车辆噪声和车内声学设计提供理论和技术支持。