Abstract:

For the problem that the Flower Pollination Algorithm（FPA） has low precision and low stability, a Flower Pollination Algorithm combining Sine Cosine Algorithm and Elite operator（SCA-EFPA） is proposed. Aiming at the local pollination process of flower pollination algorithm, the problem of small pollination range and easy to fall into local optimum value, this paper uses the “local development”and “global search” characteristics of sine cosine algorithm, and introduces it after simplifying and improving. For the global pollination process, the search range is large and the optimization accuracy is low. The elite pollination operator is introduced to improve the precision of optimization and to carry out mutation and cross operation to maintain the diversity of the population. Therefore, achieving the entire improved algorithm has the purpose of improving the precision of optimization. The results show that compared with the flower pollination algorithm, particle swarm optimization and differential mutation algorithm, the flower pollination algorithm based on sine cosine algorithm and elite operator has higher precision, better stability and convergence.

Key words: flower pollination algorithm, sine cosine algorithm, elite operator, optimization accuracy

摘要：

针对花授粉算法（FPA）具有寻优精度较低，稳定性不高的问题，提出了一种融合正弦余弦算法和精英算子的花授粉算法（SCA-EFPA）。针对花授粉算法的局部授粉过程，授粉范围小且易陷入局部最优值的问题，利用正弦余弦算法的“局部开发”和“全局搜索”特性，并作简化改进后引入；针对其全局授粉过程，搜索范围较大且寻优精度低的问题，引入精英花粉算子以提高寻优精度并且进行变异和交叉操作以保持种群多样性。达到整个改进后的算法具有提高寻优精度的目的。选取多组标准测试函数来测试改进算法的各项性能。结果表明，与基本花授粉算法、粒子群算法和差分变异算法等相比，融合正弦余弦算法和精英算子的花授粉算法具有更高的寻优精度，更好的稳定性和收敛性。

关键词: 花授粉算法, 正弦余弦算法, 精英算子, 寻优精度