MUSIC spectrum estimation method based on multi-mode PSO searching

The invention discloses an MUSIC spectrum estimation method based on multi-mode PSO searching, wherein the method belongs to the field of spectrum estimation. The method comprises the steps of constructing an autocorrelation matrix according to observing data; performing characteristic value decomposition on the autocorrelation matrix, and forming a noise subspace and a signal subspace; afterwards, constructing a fitness function by means of the noise subspace and a signal frequency vector; and finally, performing multi-spectrum-peak simultaneous searching in a frequency feasible domain by means of a multi-mode bare bones particle swarm optimization algorithm, and realizing signal frequency estimation. According to the MUSIC spectrum estimation method, fixed-step traversal searching is replaced by intelligent cluster searching. Not only is frequency estimation precision improved, but also computing cost is greatly reduced. 本发明公开了属于频谱估计领域的种基于多模态PSO搜索的MUSIC频谱估计方法。首先根据观测数据构建自相关矩阵；然后对自相关矩阵进行特征值分解，并形成噪声子空间和信号子空间；接