引入蝙蝠算法的最大似然DOA估计

Introduction of bat algorithm into maximum likelihood DOA estimation

DOA估计理论的传统算法中,最大似然DOA估计方法能准确地估计出目标方向角度,性能优良,并且具有很好的稳定性。与MUSIC及其他的子空间分解类算法相比,在信噪比较低、小快拍信号时,最大似然DOA估计算法优势更为突出。但是由于其自身算法复杂度较高的缺陷而碍于工程上的应用。针对这一问题,将蝙蝠算法与最大似然算法相结合,应用于信号的DOA估计,利用蝙蝠搜索算法搜索路径优、寻优能力强的优点,快速搜索到似然函数的全局最优值,优化多维非线性的估计谱函数。仿真结果表明,蝙蝠搜索算法有效地克服最大似然DOA估计中存在的运算量大,计算复杂度高等问题,通过与其他经典的仿生智能优化算法相比较,该方法体现出更好的收敛性。

The maximum likelihood（ML）direction-of-arrival（DOA）method can estimate the angle of object direction accurately,and has an excellent performance and high stability,which is the better one in the traditional algorithms based on DOA estimation. Compared with MUSIC and other subspace decomposition class methods,the ML DOA estimation algorithm has more outstanding superiority when signal-to-noise ratio（SNR）is lower and snapshot signal is smaller. However,it is blocked in the engineering application due to its high complexity. To reduce the heavy computational burden of ML method and make it more suitable for engineering applications,the bat algorithm and the maximum likelihood algorithm are integrated to estimate signal DOA. The advantages of optimal search path and strong search capability of the bat algorithm are used to search the global optimal value of likelihood function quickly. The simulation results demonstrate that the bat algorithm can overcome the problems existing in the maximum likelihood DOA estimation,such as large amount of calculation and high computation complexity.Compared with other typical bionic intelligent algorithms,this method has better convergence.