【作者】 张晓庆；

【导师】 高尚；

【作者基本信息】 江苏科技大学 ， 计算机应用技术， 2017， 硕士

【摘要】 当今社会,从生产调度到经济分析再到生物医药,这些领域无不存在大量需要解决的优化问题。尤其是电子商务行业的兴起,现代物流业迎来了发展的高峰期,而物流配送作为其中最关键的一环,它的速度快慢直接影响客户对物流公司的服务评价。车辆路径优化问题是物流运输这一环节极其重要的内容,目前主要是通过群智能优化算法对其进行优化。人工蜂群算法作为一种极具发展前景的仿生算法,具有控制参数少、鲁棒性强和易于实现等优点,已被成功应用于组合优化、无线传感网路和图像处理等领域。但是,它仍存在局部开发能力较差、收敛速度较慢和寻优精度不高的缺点。为了进一步提高人工蜂群算法的性能,本文针对算法中存在的上述问题,分别从算法改进和应用两个方面进行了研究分析。在算法改进方面,我们把差分进化和高斯变异的思想引入到搜索策略中,设计了一种基于当前最优解的混合变异的人工蜂群算法。该算法中跟随蜂采用差分进化的搜索策略,通过引入差分变异因子实现在迭代初期进行全局搜索,以增加解的多样性,在迭代后期进行局部搜索,来加快算法的收敛速度;侦查蜂采用高斯变异的侦查策略,借助高斯变异算子对局部最优解进行扰动,使侦查蜂跳出局部最优值,然后在当前最优解附近生成一个新解,通过变异来丰富种群的多样性,并在当前最优解地引导下有效的提高了算法的收敛速度,避免其陷入局部最优值。基于6种典型测试函数的仿真实验结果表明,基于混合变异的人工蜂群算法在收敛速度和寻优精度方面都优于其他的人工蜂群算法。在算法应用方面,本文将改进后的人工蜂群算法应用到了物流配送车辆的路径优化问题中。我们首先对算法进行了离散化处理,然后利用贪心策略生成问题的初始解,再设计了3种候选解的生成策略并加以组合,扩大了邻域的搜索范围,同时,将遗传算法中交叉思想引入到了侦查策略中,通过2种交叉方式改善了侦查蜂的侦查性能,最后用两个不同规模的实例对改进后的算法做了性能验证,实验结果表明,与其他进化算法相比,本文的算法在求解车辆路径优化问题上较为有效。更多还原

【Abstract】 Nowadays,there are a lot of optimization problems which need to solve,such as production scheduling,economic analysis,biological medicine,etc.Especially,with the rise of e-commerce industry,modern logistics industry has ushered in the peak period of development.However,as one of the most critical part in logistics industry,the speed of logistics distribution has directly impact on customers’ service evaluation of logistics companies.Vehicle routing optimization problem is an important topic in logistics transportation,which is mainly optimized by swarm intelligence optimization algorithms.Artificial Bee Colony algorithm is a very promising bionic algorithm.It has the advantages of few control parameters,strong robustness and implementation simplicity and has been successfully applied to combinatorial optimization,wireless sensor network and image processing,etc.However,Artificial Bee Colony algorithm still owns some disadvantages,such as weak local development ability,slow convergence speed,and low optimization precision.To further improve the performance of Artificial Bee Colony algorithm,this paper analyzes the existing problems,and focuses on the improvements and applications of Artificial Bee Colony algorithm.In the aspect of algorithm improvement,the ideas of differential evolution and Gauss mutation are introduced into the search strategy and we have designed an artificial bee colony algorithm based on the mixed mutation of current optimal solution.The differential evolution search strategy is adopted by followers.In early iterations,differential mutation factor is introduced to increase the diversity of solutions with the global search,and in later iterations,it can accelerate the convergence speed of the algorithm with the local search.The scouts adopt Gauss mutation detection strategy to perturb the local optimal solution with Gauss mutation operator,and then jump out of local optimal value.A new solution can be generated near the current optimal solution.The proposed algorithm enriches the diversity of population and effectively improves the convergence speed of algorithm with the guidance of the current optimal solution.The experimental results on 6 typical test functions show that the proposed artificial bee colony algorithm based on hybrid mutation is superior to other artificial bee colony algorithms in the convergence speed and optimization precision.In the aspect of algorithm applications,the proposed artificial bee colony algorithm has been applied to the path optimization problem in logistics distribution vehicles.Firstly,the algorithm is discretized,and then the greedy strategy is used to generate the initial solution of problem.We design 3 generation strategies of candidate solution and combine them to expand the neighborhood search range.Meanwhile,the crossover idea of genetic algorithm is introduced into the detection strategy,and the detection performance of scouts is improved through 2 kinds of crossover methods.Finally,the experimental results on 2 different scale examples show that,compared with other evolutionary algorithms,the proposed algorithm is very effective for solving the vehicle routing problems.更多还原