针对MUSIC归一化算法在低信噪比时波达方位( DOA)估计性能下降的问题,提出了基于支持向量机( SVM)可控功率响应的MUSIC-DOA算法.该算法首先对宽带信号进行快速傅里叶变换,然后用MUSIC算法进行DOA估计,最后通过SVM对每个子带信号的DOA估计结果进行分类,选择分类后DOA估计结果较为准确的子带信号进行融合,得到宽带信号的DOA估计.该文方法有效解决了MUSIC归一化算法在低信噪比时定位精度不高的问题,提高了定位系统的稳定性. strong reverberation,the steered response power( SRP) MUSIC algorithm based on the support vector machine(SVM) is proposed here. Firstly,the proposed algorithm performs fast Fourier transform on the wideband signals,and then uses the MUSIC algorithm to estimate the DOA. Finally,results of the DOA estimation of the each sub-band signal are classified by using the SVM,and the more accurate sub-band signals are selected after the classification to make the fusion and to get the broadband signal DOA estimation. The method in this paper solves the problem that the MUSIC normalization algorithm has low precision under the low signal to noise ratio effectively,and improves the stability of the algorithm.