【作者】 曾宇容；

【导师】 张金隆；

【作者基本信息】 华中科技大学 ， 管理科学与工程， 2018， 博士

【摘要】 联合采购通过对多种物品进行分组组合采购,从而达到分担主要订货成本,节约采购总成本的目的。为有效控制全球采购背景下的总成本,联合采购策略的应用越来越普遍。另外供应链管理环境下,将采购、选址、配送进行综合集成优化显然更具有现实意义和应用价值。因此,综合考虑以上策略,构建集成优化模型和设计实施难度低的高性能求解算法,具有重要的理论意义。本文基于联合采购策略,协同考虑订货量分配、配送与选址等精细化管理要求,克服传统方法局限性,分别构建三种拓展的联合采购决策模型。为有效解决这些NP-hard问题,引入果蝇优化算法新技术并融合其它进化技术的优点,设计精度高、通用性强、稳定性好的改进果蝇优化算法。通过典型的标准函数进行测试,验证了该算法良好的综合性能,为求解所建的优化模型提供方法上的支持。主要内容包括以下四个方面:首先,设计了一种嵌入改进信息交换机制的混合果蝇优化算法(Hybrid fruit fly optimization algorithm,HFOA)。基本果蝇优化算法由于视觉嗅觉的缺陷,面临着种群多样性和局部搜索能力薄弱等挑战。为了克服这些局限性,需要更有效地改善种群多样性和平衡算法的全局和局部搜索能力。采用能有效挖掘果蝇群体位置信息的混合交换策略,从而强化局部搜索能力;嗅觉操作采用概率飞行来平衡全局和局部搜索;设计了一种有效的灾变策略,以帮助果蝇跳出局部极值点。并采用18个复杂的连续基准函数用来测试HFOA的性能,数值实验结果表明HFOA优于目前主流的优化算法。其次,研究了联合采购策略下的订货量分配协同优化问题。分析了采用联合采购策略的订货量分配协同决策问题,该模型考虑了供应商选择问题,并采用HFOA算法进行求解。进而在考虑商品异质性带来的分组约束基础上,构建了一种拓展的供应商选择与订货量分配协同决策新模型。对比算例分析证实HFOA在求解此类问题优于模拟退火算法和标准的果蝇优化算法,随机生成的大规模算例进一步验证了改进HFOA可求解此类复杂优化问题。第三,研究联合采购策略下考虑资源约束的配送调度集成优化问题。首先通过求解无/有资源约束的联合采购-配送调度模型来验证HFOA算法的综合性能,通过一系列数值算例分析发现在大多数情况下,HFOA算法性能要优于一种改进的FOA、差分进化算法和遗传算法。进而在总成本结构中,加入了考虑商业信用导致的费用,构建了考虑商业信用和运输资源约束的集成优化模型。通过较大规模的算例结果证明,HFOA的性能仍然要优于差分进化和遗传算法。因此,HFOA的总体性能满足复杂优化问题的实用要求。最后,研究了联合采购策略下考虑资源约束的选址-配送集成优化问题。构建了更具现实意义的允许缺货且考虑运输容量约束的联合采购-选址-配送集成优化新模型,针对性设计了基于HFOA的求解算法;并将算例结果与其它流行的智能优化算法进行比较,结果验证了HFOA的有效性;对模型相关参数进行了敏感性分析,讨论参数变动带来的影响,为企业运营管理合理化决策提供科学依据。更多还原

【Abstract】 Joint replenishment(JR)policy can achieve total cost saving through the sharing of the main ordering cost and by grouping and purchasing multiple items.In order to control the total cost effectively under global purchase background,the application of joint replenishment policy becomes more and more popular.In addition,it is of great practical significance to integrate and optimize the procurement,location,and distribution under the environment of supply chain management.Therefore,the design of practical optimization models based on the above strategies and easy implementation algorithm with high performance has great theoretical significance and high application value.Considering the requirements of precise management of order quantity allocation,distribution and location,three extended joint replenishment models are proposed by overcoming the limitations of traditional methods.In order to solve these NP-hard problems effectively,the improved fruit fly optimization algorithm(FOA)is designed with high accuracy,versatility and good stability by integrating the advantages of other evolutionary technologies.Typical standard function test are carried out to verify the good comprehensive performance of the proposed algorithm,and it provides a methodllogical support for solving the above optimization models.The main work includes the following four aspects:Firsly,a hybrid FOA(HFOA)using improved location information exchange mechanism is proposed.The basic FOA is faced with the challenges of poor diversity of the swarm and weak local search ability because of the improper osphresis operation and vision operation.To overcome these limitations synthetically,it is necessary to improve the swarm diversity in a more efficient way and well balance the global search and local search abilities.The proposed HFOA enables flies to communicate with each other and conduct local search in a swarm based approach.Moreover,osphresis operation is conducted in probability to balance the global search and local search processes.A mutation strategy called cataclysm policy is designed to help the flies jump out of the local extreme points.Eighteen complex continuous benchmark functions are used to test the performance of HFOA.Numerical experiments results indicate that HFOA outperforms main state-of-the-art algorithms.Secondly,the coordinated order quantity allocation problems based on joint replenishment policy are studied and algorithms are provided.The basic model considering suppliers selection and the improved HFOA is proposed to solve this problem.Moreover,a new coordinated order quantity allocation model considering grouping constraint caused by the heterogeneity of items is developed.Results of contrastive numeric examples show that the HFOA outperforms simulated annealing algorithm(SA)and standard FOA in solving this problem and its extension type.The effectiveness of HFOA is further verified by randomly generated large-scale problems.Thirdly,the practical joint replenishment and delivery(JRD)models are studied by considering resources constraints.The classic JRDs with and without resources constraints are used to vefrify the performcance of HFOA,and results show HFOA is better than an improved FOA,differential evolution algorithm(DE)and GA.Then,the cost caused by trade credit is examphsized and a new and practical JRD model by considering trade credit and transportation resource constraint is proposed.Results of randomly generated JRDs indicate that HFOA can always obtain slightly lower total costs than DE and GA under different situations.The comprehensive performance of HFOA is satisfactory for practicability requirements for solving complex optimization problems.Lastly,a new joint replenishment-location-delivery model is studied considering resource constraints.The proposed model with transport capacity constraint,which permits shortage,is more practical.An effective algorithm based on HFOA is designed to solve the proposed model.Compared with other popular algorithms,numerical studies show the effectiveness of the proposed HFOA.Finally,the sensitivity analysis is further conducted to discuss the influence of parameters on the total cost,which can provide useful references for operations managers to make better decisions.更多还原