Modeling and Analysis of Capacitated Transfer Lines with Unreliable Machines and Deterministic Procesing Times

ＭｏｄｅｌｉｎｇａｎｄＡｎａｌｙｓｉｓｏｆＣａｐａｃｉｔａｔｅｄＴｒａｎｓｆｅｒＬｉｎｅｓｗｉｔｈＵｎｒｅｌｉａｂｌｅＭａｃｈｉｎｅｓａｎｄＤｅｔｅｒｍｉｎｉｓｔｉｃＰｒｏｃｅｓｉｎｇＴｉｍｅｓＳＨＵＳｏｎｇｇｕｉＪＩＡＮＧＣｈａｎｇｊｕｎＩｎｓｔ...     

Technical Note ：SINGLE-PERIOD TWO-PRODUCT INVENTORY MODEL WITH SUBSTITUTION： SOLUTION AND PROPERTIES

In this paper, we study a single-period two-product inventory model with stochastic demands and downward substitution. The optimal order quantities are presented and some properties are provided. Comparing with newsboy model, we prove that both the profit and the fill rate can be improved by using the substitution policy. This work was supported partly by NSFC/RGC Joint Research Program under grant 79910161987 and the National Science Foundation of China (79825102, 70231010, 70321001). Lianqiao CAI is lecturer of School of International Business, Beijing Foreign Studies University. He received his B.S. and M.S. from School of Economics and Management, Tsinghua University, in 1995 and 1998, and Ph.D of Management Science from Tsinghua University in 2002. His research areas include supply chain management, optimization techniques. Jian Chen is Professor and Chairman of the Management Science Department and co-director of the Research Center for Contemporary Management, Tsinghua University. He has over 80 papers published in some leading international journals and first class Chinese journals, and has been a principal investigator for over 20 grants or research contracts with National Science Foundation of China, governmental organizations, and companies. His main research interests include supply chain management, E-commerce, modeling and control for complex systems, decision support systems and information systems, and forecast and optimization techniques. He serves as associate editor of the “IEEE Transactions on Systems, Man and Cybernetics: Part A” and “IEEE Transactions on Systems, Man and Cybernetics: Part C”, and on the Editorial Board of “The International Journal of Electronic Business” and “System Research and Behavioral Science”. He is the recipient of the Outstanding Contribution Award of IEEE Systems, Man, and Cybernetics Society in 1996, and the Young Scientist Award of China in 1992. He was Secretary General of the 1996 IEEE International Conference on Systems, Man, and Cybernetics in Beijing, Co-chair of the IPC of 1998 International Conference on Systems Science and Systems Engineering, Chair of the First Asian eBiz Workshop in 2001, Co-chair of the Asian eBiz Workshop in 2002 and 2003, and Co-chair of the international conference on Global Supply Chain Management in 2002. Houmin Yan received his B.S. and M.S. from the Department of Automation, Tsinghua University, in 1982 and 1985, respectively, and his Ph.D. from the Faculty of Management, University of Toronto in 1993. From Jan. 1994, he is with the Department of Systems Engineering and Engineering Management, the Chinese University of Hong Kong as an Assistance, Associate and Full Professor. He was a tenured Associate Professor at School of Management, the Univ. of Texas at Dallas. His main research areas are operations management, stochastic models, simulations, and supply chain management. He has published in journals such as Operations Research, Manufacturing and Service Operations Management, Production and Operations Management, Journal of Optimization: Theory and Applications, IIE Transactions and IEEE Transactions. He consults several high-tech companies, such as Motorola, C&K Systems, and Oriental Power, on issues spanning from production planning and scheduling, supply chain management to business process re-engineering. He is a member of INFORMS.     

ON IMPROVING AN INTEGRATED INVENTORY MODEL FOR A SINGLE VENDOR AND MULTIPLE BUYERS WITH A RELAXED MATERIAL ORDERING CYCLE POLICY

1. Introduction With the development of information technology and the increasing market competition, simultaneously reducing inventory costs of raw materials, work-in-process, and finished items in different stages has become a major focus for supply chain management. Industrial practice and academic research have shown that VMI (Vendor Managed Inventory) can improve supply chain performance by decreasing inventory levels and costs. VMI is a collaborative business initiative where the vend…     

基于FCN的眼底图像中央凹自动检测算法

针对传统算法很难识别彩色眼底图像中央凹的问题, 提出一种基于全卷积网络（fully convolutional networks, FCN）的眼底图像中央凹自动检测算法. 首先通过彩色眼底图像的局部上下文环境挖掘全局上下文信息, 构建实现局部像素级分类的FCN模型, 然后将局部像素级特征推广到全局金字塔池化模块中, 使空间统计数据为全局语境理解提供了更好地描述与表达, 从而有效获得了极具区分度的全局上下文信息, 最后将全局与局部特征相融合, 实现对中央凹的精准检测. 实验结果表明, 该算法提高了眼底暗病变检测的特异性, 并为眼底严重病变的发现提供了有效证据.     

基于一维堆叠池化融合卷积自编码器的HRRP目标识别方法

针对雷达高分辨距离像(high resolution range profile, HRRP)目标识别中特征提取及识别问题, 提出了一种基于一维堆叠池化融合卷积自编码器(one-dimensional stacked pooling fusion convolutional autoencoder, 1D SPF-CAE)的识别方法。首先构造一维池化融合卷积自编码器(one-dimensional pooling fusion convolutional auto-encoder, 1D PF-CAE), 在编码阶段, 采用最大池化和平均池化同时提取不同的编码特征并进行融合来提取HRRP的结构特征; 然后堆叠多个1D PF-CAE形成1D SPF-CAE; 最后使用标签数据对网络进行微调, 实现HRRP目标识别。并使用AdaBound算法优化网络训练来提高识别性能。基于弹道中段目标仿真数据的实验结果表明, 该方法具有较强的特征提取能力, 对于HRRP目标识别准确率高、鲁棒性强。     

不确定条件下VMI供应链协调的期权合约

通过引入期权合约,研究在产品市场需求不确定条件下供应商管理库存(VMI)供应链的协调。研究表明,首先,在VMI供应链协调中,市场不确定程度越高,期权合约越重要,供应商销售或者零售商订购的期权越多;其次,对整个VMI供应链而言,期权合约使得供应商获得更大比例的利润;最后,零售商订购期权更多是出于规避缺货风险的动机,而非单纯地增加利润。     

带学习效应的易变质供应链库存模型研究

根据订购固定成本带学习效应的易变质供应链库存模型,在有限期内,当需求为固定常数时,分析了两种情形下的最优订购策略。一类是允许发生短缺且缺货需要回补时,证明了最优解存在的唯一性,设计了求解最优订购策略的算法;另一类是不允许发生短缺时,将订购次数作为自变量,证明了系统总成本存在最优解的唯一性,给出了求得最优订购次数的计算方法。最后通过实例进行了验证。     

考虑库存能力约束的批量问题与定价的联合决策

针对多周期离散时间情形,研究了库存能力约束下如何有效地协调定价决策与批量决策,使得企业利润最大化。建立了非线性混合整数规划模型。分析了最优解的特征,详细讨论了子问题的求解方法。提出了基于动态规划的精确算法,可以在多项式时间内求解出联合决策问题的最优定价与最优生产计划。实验结果分析了库存能力与最优利润、最优生产计划之间的关系。     

橡胶阻尼材料的生命周期清单分析

着重介绍了橡胶阻尼材料的生命周期清单分析方法,并以某国产橡胶阻尼卷材为实例,探讨了国内橡胶阻尼材料在能源消耗和环境排放方面的特点：每功能单位该卷材生命周期的总能耗约为6763.81 MJ,其中石油占79.03%,煤占10.29%,天然气占6.77%,电力占3.91%;共排放大气污染物375.79 kg,其中原材料获取阶段占92.68%,产品生产阶段占5.91%,包装材料生产和用后处理阶段仅占0.63%和0.78%。同时为类似产品的生命周期清单分析提供方法上的借鉴。     

改进的多任务学习方法的眼底视盘分割与定位

提出了一种改进的多任务学习方法,网络的主结构由特征提取网络和分别进行视盘分割与视盘定位的双路径网络组成,通过端到端的训练与测试可以实现眼底图像视盘自动分割与定位相结合的多任务目的。在特征提取网络的编码阶段利用密集连接提取眼底图像视盘的上下文特征。视盘分割任务是依靠解码阶段逐步恢复原来的图像分辨率并获取整个视盘轮廓,视盘中心定位任务由空洞空间金字塔模块和金字塔池化模块来进一步提取视盘抽象特征,得到精准的视盘中心坐标。对350幅眼底图像进行了视盘分割和中心定位,实验结果表明:该方法自动分割的视盘结果与手动标注视盘区域的Dice系数为0.965,自动定位的视盘中心坐标与手动标记的视盘中心的平均绝对距离为0.191 mm(34.7像素)。