中文摘要：

花朵授粉算法是最近出现的一种新型元启发式群智能算法，已在无线传感网等应用领域取得了较好的成效，但也存在易陷入局部极值且演化后期收敛速度慢等缺陷，使其应用范围受到限制．针对该算法存在的不足，提出一种把复合形法的思想融入到花朵授粉算法中的混合算法．该算法首先计算当前种群的形心，然后依据形心将进化中最差的个体进行反射，把差的个体变成较好的个体，从而引导当前个体不断靠近最优解．通过10个标准测试函数的仿真实验，仿真结果表明，改进算法解的质量及收敛速度比基本的花朵授粉算法、蝙蝠算法及粒子群算法有较大幅度的提高．同时采用改进算法对非线性方程组问题进行求解，通过2个算例仿真实验，验证了改进算法的有效性，扩展了花朵授粉算法的应用领域．

英文摘要：

Flower Pollinate Algorithm{ FPA） is the recent emergence of a new metaheuristic of swarm intelligence algorithm. The FPA has achieved good results in wireless sensor network applications;but it is easy to relapse into local extremum, and having the low speed of convergence and so on in the late, which tend to limit the use of the FPA. In order to solve the problem of the FPA, a hybrid algorithm of complex method into the FPA was presented in the paper. The centroid of current population is firstly calculated in the FPA, and then the evolution worst individual is reflected with the centroid, the difference of individual into a better individual, so as to guide the individuals continue to approach the optimal solution. The FPA was put through ten standard test functions, the simulation re- suits show that the improved algorithm has the advantages of the convergence rate, solution quality, which were significantly better than those of the basic Flower Pollination algorithm ,bat algorithm, particle swarm algorithm. At the same time, by using the improved algo- rithm for solving nonlinear equation group, through the two examples simulation, the experimental results verified the validity of the improved algorithm, which extended the application field of the flower pollination algorithm.