【作者】 赵娜；

【导师】 孙光圻；

【作者基本信息】 大连海事大学 ， 交通运输规划与管理， 2008， 博士

【摘要】 近年来,供应链管理已成为国内外企业和学者共同面对的重要课题。供应链管理优化模型及其算法虽研究较多,但尚待探索,予以完善。本文主要针对不确定性条件下供应链管理中的供应商选择、生产计划和分销订货决策问题,从战术层面建立优化模型,并寻求有效的求解方法,为供应链管理提供理论依据和参考。首先分析了供应链中不确性的来源及其影响,指出本文建模过程中所考虑的不确定性,并利用蒙特卡洛法对不确定性需求进行预测。以服从正态分布为例,考虑不确定需求独立和相关两种实际情况,分别产生独立随机数和相关随机数对独立需求和相关需求进行预测,统计检验结果表明预测方法的有效性。对需求相关性的考虑将使对应的一类实际问题求解更具合理性。其次考虑需求和采购商主、客观偏好的不确定性,建立供应商选择及定购计划多目标模型。利用不确定的主、客观偏好系数,结合AHP,DEA求得的权重,确定供应商各评价准则的综合权重,将多目标规划模型转化成单目标规划模型,利用优化软件求解。算例的结果既说明算法的有效性,也表明该方法可为采购商选择供应商和确定定购量提供参考。再次考虑需求的不确定性和相关性,以利润最大和机会损失最小为目标建立生产计划的多目标模型。采用相关随机数作为需求预测,将蒙特卡洛法与遗传算法相结合对问题进行求解。算例的对比试验结果既表明了算法的有效性,也表明该方法能够为决策者提供有效的帮助。最后考虑分销商不同的竞争行为,以及订货量对价格的敏感性,以利润最大为目标,建立分销系统订货决策的优化模型,利用博弈论和运筹学相结合的方法对模型求解。数值实例不仅验证了算法的有效性,而且表明该方法可以为生产商制定批发价格和分销商确定订货量、分销价格提供有效的参考和帮助。更多还原

【Abstract】 Recently, supply chain management has become an important issue in both research field and practical world. Purchasing, production, distribution are crucial parts of it. Though many researchers have been exploring supply chain and supply chain management, there are many problems to be studied. It is very important to study supply chain management with operational research and other theory, which is a quite pressing task in front of us.This dissertation mainly addresses supplier selection, production planning and distribution decision in supply chain management under the uncertainties. Based on tactics, the corresponding optimization models are developed, and efficient solutions are found, for providing insights for supply chain management.First, this paper analyzes domestic and foreign research status about modeling of supply chain management, points out the uncertainties and their influence in supply chain, and forecasts the uncertain demand with Monte Carlo Method. Demand is assumed to follow normal distribution. Independent random number and correlated random number are respectively generated to simulate independent demand and correlated demand. Great flexibility is achieved owing to the consideration of the demand correlation.Second, considering uncertainties of demand and subjective preference factor, a multi-objective programming model is developed for vendor selection. A combination of AHP with DEA is proposed to determine the integrated weight for the evaluation criteria of vendors. Based on the integrated weight, the multi-objective programming model is converted into a single objective programming model. Illustrations prove the effectiveness of the approach.Third, considering uncertainty and correlation of demand, a multi-objective programming model of production planning is developed for the objectives that profit should be maximized and opportunity loss minimized. Monte Carlo Method and Genetic Algorithms are used to solve the model. The approach is well tested in the experiments, and the solutions are appropriate alternatives for the decision-makers. Finally, considering retailers’ different competitive behaviors and the sensitivity of retail quantity to the prices, the order decision of a distribution system is researched. The profit models of manufacture and retailers are developed respectively. Operational research and game theory are utilized to solve the models. The experimental results show that the algorithm can wonderfully analyze the distribution system.更多还原