摘要
等效源法近场声全息是进行声源识别的重要方法。传统的基于Tikhonov正则化方法局限于相对低的频率,进行高频声源的声场重建时效果较差,而基于最速下降法的宽带声全息(wideband acoustic holography,WBH)方法则在中高频效果较好。为了拓宽声场重建的频率范围并提高声源识别分辨率,提出一种基于增广拉格朗日方法(augmented Lagrangian method,ALM)的等效源法声源识别算法,该方法将L1范数正则化模型转化为增广拉格朗日方程的最小化问题,并应用不动点迭代求解得到声源强度。通过仿真与试验表明,与Tikhonov正则化、WBH和快速迭代收缩阈值算法(fast iterative shrinking threshold algorithm,FISTA)三种方法对比,所提方法适用于更宽的频率范围,且对不同的全息距离和信噪比具有很好的适应性。
Near-field acoustic holography based on the equivalent source method is an important method for sound source identification.The conventional Tikhonov regularization method is limited to apply in the condition of relatively low frequencies,and the effect of the sound field reconstruction of high-frequency sound sources is rather poor,while the wideband acoustic holography(WBH)method based on the steepest descent approach has better effects at middle-to-high frequencies.In order to widen the frequency range of sound field reconstruction and to improve the resolution of sound source identification,an equivalent source method based on the augmented Lagrangian method(ALM)was proposed for sound source identification,which transforms the L1-norm regularized model into the minimization problem of an augmented Lagrangian equation and applies fixed-point iteration to gain the source strength.The simulations and experiments results show that,compared with the three methods of Tikhonov regularization,wideband acoustic holography and fast iterative shrinking threshold algorithm(FISTA),the proposed method is suitable for a wider frequency range and has good adaptability to different holographic distances and signal-to-noise ratios.
作者
贺岩松
贾晨阳
黄琳森
昝鸣
徐中明
HE Yansong;JIA Chenyang;HUANG Linsen;ZAN Ming;XU Zhongming(School of Automotive Engineering,Chongqing University,Chongqing 400030,China;The State Key Laboratory of Mechanical Transmission,Chongqing University,Chongqing 400030,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2022年第14期243-252,共10页
Journal of Vibration and Shock
基金
国家自然科学基金(11874096)。
关键词
等效源法
声源识别
压缩感知
增广拉格朗日方法(ALM)
equivalent source method
sound source identification
compressive sensing
augmented Lagrangian method(ALM)
