【作者】 何川；

【导师】 杨春节；

【作者基本信息】 浙江大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 在进入21世纪后,所有钢铁企业都面临着以市场为导向、以客户为中心的竞争环境。这就要求企业在市场需求发生变化的时候快速反应,制定相应的采购、生产、配送等计划。而我国钢铁企业多采用事业部制分销模式,不能建立统一的信息共享平台和敏捷的市场反应机制,快速响应市场需求的变化。同时,钢产量的快速增长带来了废钢废铁的快速增加,如何高效地将这些废钢废铁回收再制造也是钢铁企业向绿色企业发展的一个难题。而对于供应链决策者来说,解决以上两方面的问题,并不容易。已经有学者利用控制理论和优化算法来解决以上两个问题,其中模型预测控制在供应链运营优化决策中应用效果良好,但是也存在着一些问题。本文针对前人在供应链运营优化决策研究中所暴露出的一些不足,分别从正向和回收再制造供应链的运营决策上做了深入的研究,主要研究工作包括:1)由于传统的模型预测控制不能应用于非线性模型,从而导致供应链决策模型没有考虑到采购量、生产量和配送量对于产品单位成本的影响,忽略了规模效应在降低单位成本的重要作用。为此,本文提出了基于蚁群算法的滚动优化决策模型和算法,将蚁群算法和模型预测控制相结合,用以求解考虑规模效应的非线性供应链运营决策模型。同时在算法中,还考虑到了企业对自身定位的不同,通过考虑包括客户满意度在内的总成本和采购、生产、分销等成本之间的权衡来求得优化决策。通过Matlab仿真发现,决策方法在求解非线性供应链运营决策模型上效果良好,能够在尽可能低地降低成本的同时,保证一定的客户满足率以及较为平稳的决策。2)逆向回收再制造供应链相比于正向供应链具有很多不同的特点,特别是不确定因素较多。其中回收产品质量、数量的不确定性是影响供应链运营优化决策的重要因素。本文通过对于回收再制造供应链的特点的研究,提出了一种基于预测控制基本原理的合理的优化控制方案,以降低在不确定环境下回收再制造供应链的成本。并用仿真验证了回收再制造供应链决策模型和控制方案的合理性,以及分析了一些重要因素对回收供应链成本的影响。更多还原

【Abstract】 All the iron and steel corporations have to be in a competitive environment which is directed by the market and customer’s demands in 20^th century. And then rapid responds are required to get corresponding plans of operations, e.g. purchase, production, logistics and so on, when market demands change. However, in our country, most iron and steel corporations can not set up an information-share platform and a rapidly responding system to market demand changes because of the decentralized management. At the same time, the rapid growth of steel production brings the rapid growth of scrap iron and scrap steel. It is really a big problem that how to remanufacture the scrap iron and scrap steel effectively when the iron and steel corporations are developing themselves into green corporations. In fact, it is really not easy for the supply chain managers to settle down the two problems mentioned.Of course, there are many scholars have been studying for years to solve the problems with control theories and optimal algorithms. And Model Predictive Control （MPC） performed well in the supply chain optimal operations, but a few problems still exist. In this thesis, the optimal operations of both forward and reverse supply chain were studied. The main contents are included below:1) In traditional optimizations of supply chain operations, because MPC cannot be used on the non-linear models, the supply chain decision-making model didn’t consider the impact of purchase quantity, production quantity and logistics quantity on the unit cost of product. Then the important function of scale-effect to reduce the operation costs was ignored. Therefore, a receding-horizon algorithm integrated with Ant Colony Optimization （ACO） and MPC is presented to get optimal decisions of the supply chain with non-linear model. And the corporation’s own orientation is also considered to get optimal operations with the weights on total cost, purchase cost, production cost, logistics cost and customer’s satisfaction. Through simulation study with Matlab tools, the algorithm performs well on the non-linear model of supply chain operation. The simulation work showed that the total cost could be reduced with a certain degree of customer’s satisfaction and smooth operations.2) There are differences between forward supply chain and reverse supply chain, especially more uncertainties in reverse supply chain. And in the reverse supply chain, the quality and quantity uncertainty of the reverse product play the most important roles in impacting on the optimal decision-making of supply chain operations. In this thesis, an optimal control solution is presented based on basic MPC principles to reduce the operation cost of reverse/remanufacturing supply chain. And some important factors which may have impacts on supply chain cost were studied in the simulation.更多还原