基于竞争机制的自适应人工蜂群算法

Self-adaptive artificial bee colony algorithm based on competition mechanism

从基本人工蜂群算法机制出发,针对原始算法容易陷入早熟和局部最优等问题,提出一种改进的人工蜂群算法(competitive self-adaptive artificial bee colony algorithm,CSABC)。利用竞争机制初始化种群,在保证种群多样性的前提下,提高初始解的质量;设计一种自适应局部搜索策略提高算法的邻域搜索能力;在选择操作上,引入自适应比例选择策略,避免算法加速收敛,陷入局部最优;对于即将抛弃的蜜源,采用最优值和最劣值指导重置个体,提高算法的计算精度。对8个典型测试函数的求解结果表明,改进算法在求解精度及可靠性方面有显著提高。

By analyzing the optimization scheme of artificial bee colony algorithm（ABC）,an improved version of algorithm named CSABC was proposed to overcome the shortcomings such that ABC traps into local optima and pre-matures easily.The competition mechanism was used to initialize the population and improve the quality of the initial solution on the premise of keeping population＇s diversity.An adaptive local search strategy was designed to enhance the ability of neighborhoods search.In the selecting operation,an adaptive proportional selection strategy was introduced to avoid the algorithm join convergence which might lead to a local optimum.The best and the worst value were used to guide to reset the unit for the nectar source that about to be abandoned,thereby improving the accuracy of the solution.The result of solving eight typical test functions indicate the solving accuracy and credibility are improved significantly using the proposed algorithm.