改进的引力搜索算法用于阵列天线方向图综合

Application of the Improved Gravitational Search Algorithm for the Pattern Synthesis of Array Antennas

针对基本引力搜索算法在处理复杂的阵列天线综合问题时,存在早熟收敛和收敛速度慢的缺陷,提出了一种混合引力搜索算法。首先将精英粒子保护算法及后进粒子微扰算法嵌入到基本的引力搜索算法中,延长了粒子的存活时间,扩大了粒子邻域的搜索范围,保护了种群的多样性,较大程度上改善了算法过早收敛的问题;其次重新定义了惯性质量调节系数q,使种群中粒子惯性质量的差距增大,算法能够快速有效地收敛于问题的最优解,从而改善了全局收敛性与局部收敛性的平衡。将该算法用于20元阵列天线方向图综合中,仿真结果表明,与基本的引力搜索算法以及同类智能优化算法相比,改进后的算法在计算精度和收敛速度,及种群多样性方面均有显著改善。

In order to overcome the problems of the premature convergence at; the slow convergence speed caused by the gravitational search algorithm （GSA） in the complex pattern synthesis of array antennas, an improved gravi- tational search algorithm the hybrid gravitational search algorithm （HGSA） is presented. By extending the sur- vival time of the elite particles and the backward particles appropriately, the diversity of the population is protected, so the problem of premature convergence of the GSA is solved; To balance the global and local searching capabilities, make the algorithm converge to the optimal solution quickly and effectively, the size of the inertia mass coefficient q is adjusted. The synthesis of the 20 elements array antenna with two examples based on the HGSA, GSA, GA, PSO are simulated. The experimental results have demonstrated that the HGSA can achieve better accu- racy and faster convergence rate compared with other three algorithms.