基于改进收益分配的人工蜂群算法

摘要

针对人工蜂群算法收敛速度较慢和容易早熟的缺点，提出一种基于改进收益度的人工蜂群算法。采用分段函数的方法计算收益度，加大收益度之间的区别，更容易选中位置更好的蜜蜂进行更新；借鉴粒子群思想，在观察蜂的更新公式中增加全局最优个体的信息反馈，加快人工蜂群算法的收敛速度。在8个测试函数上的仿真和对比实验结果表明，在30维上有7个函数的测试结果优于其它算法，在5个函数上的 T 测试结果有显著提高，在函数维数加到60维时仍然有7个函数测试结果优于其它算法，将函数维数进一步加大到100维函时，该算法依然具有很强的鲁棒性和处理高维复杂函数的能力。%To solve the problem that the basic artificial bee colony algorithm converges slowly and prematurely,an artificial bee colony based on the improved income distribution was proposed.The bee with better position was easier to be selected to update when the difference of income degrees increased by adopting the method of piecewise function to calculate the income.At the same time,based on the particle swarm thought,the global optimal individual was added in observing bees update formula to in-crease the algorithm’s convergence speed.The simulation results of the problem in eight test functions show that,seven function test results are superior to other algorithms on the 30 d and the T-test show that the results on the five functions are significantly improved.Also seven function test results are superior to other algorithms on the 60 d,what’s more,when the function dimension in-crease to 100 d,the algorithm still has strong robustness and the ability to deal with high-dimensional complex functions.