国家自然科学基金(No. 61862032,No.71863018,No.71403112);江西省教育厅科技项目(No. GJJ200424)。
针对人工蜂群算法（artificial bee colony, ABC）存在寻优精度不高、收敛速度较慢、容易被局部极值吸引的不足，提出一种具有Lévy飞行和反向学习策略的增强型人工蜂群算法（enhanced artificial bee colony algorithm with Lévy flight and opposition-based learning strategy, ELOABC）。首先，在雇佣蜂和观察蜂阶段，引入Lévy飞行改进新产生的解，由于Lévy飞行具有随机步长性，因此可以避免算法陷入局部最优；其次，在侦查蜂阶段，变异解由停滞解和当前最优解的位置决定，再结合反向学习（opposition-based learning, OBL）策略生成变异解的反向解，保留两者中更好的解以提高算法解的精度；最后，利用15个基准测试函数对增强型人工蜂群算法的性能进行实验测试。实验结果表明，改进算法性能明显优于其它算法。
Aiming at the demerits of artificial bee colony algorithm (ABC) such as lower optimization precision, slower convergence speed, and easy to be attracted by local optimal solution, an enhanced artificial bee colony algorithm with Lévy flight and opposition-based learning strategy (OBL) which is called ELOABC is proposed. Firstly, in the phase of employed bees and onlooker bees, the Lévy flight strategy is employed to improve the newly generated solution. It is because Lévy flight has random step length that it can improve the global optimization ability of the ABC algorithm. Secondly, in the scout bee’s phase, mutation solutions are generated according to the positions of the stagnation solution and the current optimal solution, and then the better solutions of mutation solutions and their opposite solutions which are generated in combination with the OBL strategy are retained to improve the solution accuracy of the algorithm. Finally, 15 well-known benchmark functions are utilized to validate the performance of the proposed ELOABC algorithm. The experimental results show that the proposed ELOABC algorithm outperforms other competitors evidently on optimization performance.
李星,张少平,邵鹏. 具有Lévy飞行和反向学习的增强型人工蜂群算法[J]. 科学技术与工程, 2021, 21(36): 15537-15545.
Li Xing, Zhang Shaoping, Shao Peng. Enhanced Artificial Bee Colony Algorithm with Lévy Flight and Opposition-based Learning Strategy[J]. Science Technology and Engineering,2021,21(36):15537-15545.