融合最优邻域扰动和反向学习策略的蝴蝶优化算法

Butterfly optimization algorithm combining optimal neighborhood perturbation and reverse learning strategy

为解决蝴蝶优化算法(butterfly optimization algorithm,BOA)易出现收敛精度差、易陷入局部极值的问题,提出了一种融合最优邻域扰动和反向学习策略的蝴蝶优化算法(butterfly optimization algorithm combining optimal neighborhood perturbation and reverse learning strategy,ORBOA)。首先,在算法初期引入改进Tent混沌映射,均匀蝴蝶初始位置。其次,在全局搜索阶段引入最优邻域扰动和透镜成像反向学习策略,在保持种群多样性的同时提高算法的收敛速度,并通过贪婪机制择选出最优的蝴蝶个体位置;在局部搜索阶段引入随机权重,增强跳出陷入局优的能力;采用6个基准函数来测试5种算法在30、50、100维度下的搜索性能,结果表明ORBOA具有更高的收敛速度、收敛精度和跳出局优的能力。最后,采用ORBOA对桁架结构进行优化,结果表明桁架结构的总重量可被有效降低。

In order to solve the problem that butterfly optimization algorithm(BOA)is prone to low convergence accuracy and local extremum,a butterfly optimization algorithm combining optimal neighborhood perturbation and reverse learning strategy(ORBOA)is proposed.Firstly,at the beginning of the algorithm,an improved Tent chaotic map was introduced to uniform the initial position of the butterfly.Secondly,in the global search stage,the optimal domain perturbation and lens imaging reverse learning strategies were introduced to maintain the population diversity and improve the convergence rate of the algorithm,and the optimal individual butterfly location was selected through the greedy algorithm.In the local search stage,the random weight was introduced to enhance the ability of jumping out of the local optimization.To compare the search performance of five algorithms in 30,50 and 100 dimensions used six benchmark functions,and the results showed that ORBOA had higher convergence speed and convergence accuracy and the ability of jumping out of local optimization.Finally,this algorithm was used to optimize the section of the truss structure,and the results indicated that the weight of the truss structure were significantly reduced.