Abstract: Due to the microphone array positioning non-convexity problems in the process of nonlinear optimization in beamformer design, local search techniques might not yield the best result. To address this problem, a hybrid descent method is proposed which consists of a genetic algorithm together with a gradient-based method. The gradient-based method can help to locate the optimal solution rapidly around the start point, while the genetic algorithm is used to jump out from local minima. Both of which will promote to find the optimal position for bean shaper design. Experimental results show that the beamformer designed by position determined by proposed hybrid descent method has better performance compared with several frequently-used positioning approaches.

Key words: beamformer design, microphone array, local search techniques, gradient-based method, Genetic Algorithm（GA）, hybrid descent method

摘要： 在波束形成器设计中，由于麦克风阵列定位优化过程中的非凸性问题，传统的局部搜索技术可能不会产生最优的结果。为了解决这一问题，提出了一种联合遗传算法和梯度方法的混合下降法。通过使用梯度方法在启动点附近迅速找到最优解决方案，同时利用遗传算法避免了局部最小化，从而促进寻找更好的波束形成器设计的最优位置。实验结果表明，与其他几种常用的定位方法相比，使用混合下降方法确定的位置所设计出的波束形成器性能更好。

关键词: 波束形成器设计, 麦克风阵列, 局部搜索技术, 梯度方法, 遗传算法, 混合下降方法