室内声场重现中的时域去卷积方法  

Time domain deconvolution for reproducing sound field in an ordinary room

作　　者：彭博[1] 郑四发[1] 廖祥凝 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：《振动与冲击》2016年第4期115-120,共6页Journal of Vibration and Shock

基　　金：国家自然科学资金资助项目(51275262);清华大学自主科研项目(2011Z02139)

摘　　要：为了在带反射边界的普通房间内重现声场,需要对原始声场信号进行去卷积处理,以获得扬声器驱动信号。建立了房间内声场重现的时域模型,在模型中采用长时长的去卷积网络,以获得高精度同时抑制了时域混叠效应,再针对卷积滤波的海量计算问题使用了快速滤波算法,提高了计算效率。汽车车内加速噪声的重现实验结果表明:所提出的时域模型可以高效准确地完成声压信号的去卷积化,实现普通房间内声场的定量重现。In order to reproduce a desired sound field in an ordinary room with a reflection boundary,deconvolution of the original sound field signals is neccessary to get loudspeaker driving signals. Here,a time-domain model used for sound field reproduction in a room was established. In the model,a long-time deconvolution network was used to acquire a high accuracy and suppress the time-domain aliasing effect at the same time. Aiming at the massive computations of a convolution filtering process, the fast filtering method was applied to improve the computational efficiency. The reproduction test of the interior noise in an accelerating vehicle indicated that the proposed time-domain model is able to complete the deconvolution of sound pressure signals precisely and efficiently,and then realize the quantificational reproduction of the desired sound field in an ordinary room.

关 键 词：声场重现 非自由场 去卷积 滤波器设计 快速卷积 

分 类 号：TB535.3[理学—物理] TN912.27[理学—声学]