基于波束形成和双目视觉的行驶汽车噪声源识别  

Noise Sources Measurement and Identification for Moving Automobiles

作　　者：李兵[1] 杨殿阁[1] 邵林[1] 连小珉[1] 

机构地区：[1]清华大学,汽车安全与节能国家重点实验室,北京100084

出　　处：《汽车工程》2008年第10期889-892,共4页Automotive Engineering

基　　金：国家自然科学基金项目(50605038)资助

摘　　要：提出一种基于波束形成原理,声阵列与双目视觉相结合的行驶汽车噪声源识别方法。通过对两台摄像机所拍摄图像的分析,计算出车辆的实时位置;将运动过程划分为若干时段进行声场重建,得出汽车行驶过程中的动态声场视频。试验结果表明,应用该方法能够准确识别运动中汽车的主要噪声源,且可更加直观地看到汽车噪声的动态变化。In this paper, a testing scheme is proposed, which combines the techniques of sound array and binocular vision, based on beamforming theory. By analyzing the pictures taken by 2 video cameras, the real-time location of vehicle is calculated; the vehicle moving process is divided into several time segments to conduct sound field reconstruction so that the dynamic sound field video in the course of vehicle driving is generated. The test resuits indicate that the main noise sources of moving vehicle can be identified accurately with this method, and the dynamic variation of vehicle noise can also be intuitively seen.

关 键 词：噪声源识别 传声器阵列 波束形成原理 双目视觉 

分 类 号：U467.4[机械工程—车辆工程]