随机多级库存成本管理的风险建模与优化方法及其应用

The background of this project is the operation and management of complex logistics of large-scale iron and steel raw materials procurement processes and multi-echelon inventory cost risk control. Facing the random uncertain transportation time, the demand for raw materials and other factors, we will investigate the problem of random multi-echelon inventory cost control and collaborative distribution scheduling, by considering decision-makers' risk preferences and risk measurement theory. For the ocean-going procurement process，applying prospect theory and the risk measurement theory, we will establish the prospect value optimization model and the conditional value at risk optimization model of random multi-echelon inventory total cost respectively, each subjected to the resource capacity constraints and time constraints of each inventory node. By analyzing the mathematical properties of these two types of model, a variety of heuristic intelligent algorithms based on model decomposition and simplification, stochastic simulation techniques will be designed for solving these models, the safety inventory control strategy of every inventory node are then given. For the ocean-going unloading and load shedding process and the domestic distribution process, the prospect value optimization model and the conditional value at risk optimization model of random multi-echelon inventory total cost will be further formulated respectively, each subjected to the storage capacity constraints, resource balance constraints, security inventory constraints and time constraints of each inventory node. Based on the decomposition-coordination techniques, a variety of hybrid intelligent optimization algorithm for solving large-scale stochastic optimization model will be investigated; collaborative distribution scheduling strategies between the various inventory nodes are then given. The results of this project to be implemented can be applied and verified in the Baosteel raw materials procurement logistics cost risk operation and management processes.

本项目背景为大型钢铁原料采购复杂物流过程的多级库存成本风险运作管理.在原料运输时间、原料需求等多种随机因素并存的环境下，考虑决策者的风险偏好和风险度量理论，研究随机多级库存的成本控制及其协同配送优化调度问题。针对远洋采购过程，利用前景理论和风险度量理论，基于库存各节点的资源容量约束和时间约束，分别建立随机多级库存总成本前景值优化模型和条件风险值优化模型，分析这些模型的数学性质，利用模型分解与化简、随机模拟等技术，设计多种启发式智能算法对模型求解，给出各库存节点的安全库存控制策略。针对远洋卸载-减载过程和国内配送过程，基于各节点库存能力约束、资源平衡约束、安全库存约束和时间约束，分别建立随机多级库存总成本的前景值和条件风险值优化模型，设计大规模随机优化模型基于分解协调技术的多种混合智能优化算法，给出各库存节点之间的协同配送调度策略。项目成果将在宝钢原料采购物流成本风险运作管理中实施应用验证。

针对大型钢铁原料采购复杂物流过程，研究了包含随机需求、随机供给和随机运输时间等不确定性因素的多级库存管理及协同配送问题，建立了基于前景理论和风险度量理论的随机多级库存成本控制及协同配送优化模型，分析模型的数学性质，利用模型转化、模型分解协调、拉格朗日分解协调、随机模拟、改进次梯度优化等方法，设计了多种混合智能求解算法求解模型，给出各库存节点的安全库存控制策略与协同配送策略。将前景理论、风险价值、条件风险价值和鲁棒条件风险价值等理论引入随机多级库存管理，有效刻画了主观风险偏好对决策的影响；利用博弈论、灵活合同及多目标优化等方法，降低了原料、运输不确定性因素对库存配送成本控制的不利影响。研究成果应用于宝钢原料物流采购系统的远洋采购、远洋卸载-减载和国内配送等业务流程。应用验证结果表明课题组提出的模型与方法，可以有效降低各类不确定性因素带来的运作风险，优化各品种和各级库存节点的铁矿石库存量，有效降低了各库存点的库存成本及运输费、中转费、滞期费等配送成本。受本项目资助，共发表学术论文53篇（其中SCI检索的期刊论文35篇），翻译学术专著《非线性规划》，获教育部高等学校自然科学二等奖1项。