供需匹配视角下基于语义相似聚类的技术需求识别模型

在技术供需文本匹配视角下，提出了一种基于语义相似聚类的技术需求识别模型.首先，采集网络中技术需求文本提取关键短语；然后，建立领域专利技术转让索引库，基于需求关键短语检索出高相关专利，构建专利技术供给背景库，并对背景库中专利标题与摘要进行分词；第三，提出基于词向量的供需文本语义匹配度算法，筛选有效技术需求并进行语义相似聚类；最后，考虑技术需求对应的需求量和专利技术转让量，对聚类结果进行二维分类.以新能源领域为例进行实证，识别出有效技术需求195个，基于语义相似聚成12类，结合需求量与专利转让量，将12类技术需求分为“高需求、高转让”、“高需求、低转让”、“低需求、高转让”、“低需求、低转让”四大类.该研究为网络技术需求挖掘及供需匹配提供一种新思路.     

幂函数需求模式下具有公平偏好的供应链协调

本文基于文献提出的协调框架，研究了幂函数需求形式下具有公平偏好的供应链的协调问题.我们的研究发现当只有零售商具有公平偏好时，供应商通过设计简单的批发价格合同就可以协调供应链.与线性函数需求模式下的研究以及指数函数需求模式下的研究相比较，我们的研究发现在幂函数需求模式下，具有公平偏好的供应链更容易达到协调.更进一步，我们获得协调的显式解.     

Coordination of a fashion and textile supply chain with demand variations

Owing to the changing fashion trends and a volatile market situation,demand in fashion and textile (FT) industry is unpredictable and could vary and change completely in a short time,which makes it more difficult to coordinate a FT supply chain.A change in product preference due to fashion trends is the main reason why the demand of FT industry shows more variations than other industries.In this paper,we use a well known contract,the all-unit quantity discount policy(AQDP),to coordinate a FT supply chain with certain demand,and we further consider it under the demand variations scenario to investigate whether it can still coordinate the supply chain.In detail,before the selling season,an AQDP is provided by the manufacturer to the retailer,and under which the FT supply chain coordination achieved with a certain demand.During the selling season,demand variation is realized after an abrupt changing of fashion trends,therefore,the manufacturer may need to revise the original AQDP to insure the supply chain is still coordinated.Utilizing the mechanism design theory,we prove that:(i) while the traditional AQDP can coordinate the supply chain when no demand variations,it cannot always coordinate the supply chain after the demand variations;(ii) when the AQDP fails,we can use the proposed capacitated linear pricing policy(CLPP) to achieve a new coordination;(iii) a more dominant decision maker,who can set a higher profit goal,is favorable to stabilization of the supply chain system under demand variations.Numerical examples are proposed also to show our results.     

A DYNAMIC ALLOCATION MODEL FOR MEDICAL RESOURCES IN THE CONTROL OF INFLUENZA DIFFUSION

In this paper,we develop a unique time-varying forecasting model for dynamic demand of medical resources based on a susceptible-exposed-infected-recovered(SEIR) influenza diffusion model.In this forecasting mechanism,medical resources allocated in the early period will take effect in subduing the spread of influenza and thus impact the demand in the later period.We adopt a discrete time-space network to describe the medical resources allocation process following a hypothetical influenza outbreak in a region.The entire medical resources allocation process is constructed as a multi-stage integer programming problem.At each stage,we solve a cost minimization sub-problem subject to the time-varying demand.The corresponding optimal allocation result is then used as an input to the control process of influenza spread,which in turn determines the demand for the next stage.In addition,we present a comparison between the proposed model and an empirical model.Our results could help decision makers prepare for a pandemic,including how to allocate limited resources dynamically.     

Stochastic comparisons in a price-quantity setting firm with uncertain demand and emergency procurement

We consider a newsvendor problem with price-dependent demand, in either additive or multiplicative format. The newsvendor has two modes of purchasing： regular ordering at the beginning of the selling season and emergency ordering （if the realized demand exceeds the initial order quantity） at the end of the selling season. By stochastic comparisons, we systematically investigate the effects of demand magnitude and demand randomness on pricing and ordering quantity decisions as well as expected profit of the newsvendor, under both usual stochastic order （first order stochastic dominance） and convex order （less variable）. Our key findings include： （i） in contrary to the case where price is exogenous, a stochastically larger demand shock may even lead to a lower order quantity; （ii） a stochastically larger demand shock leads to a higher price for the additive demand case, but may lead to a lower price for the multiplicative demand case; （iii） a stochastically larger demand shock leads to a higher expected profit for both demand models; （iv） in general, a less variable demand leads to a higher expected profit for both demand models; and furthermore, a less variable demand shock has no effect on price for the additive demand model, but leads to a higher price for the multiplicative demand model. The implications of all these findings for pricing and order quantity are discussed in detail.     

双渠道订货条件下基于预测更新过程的报童模型研究

零售商可以在整个计划时段内持续不断的收集市场需求信息,对市场需求进行预测.离销售季节开始越近,市场需求预测越准确,但采购成本会较高.零售商有两次订货机会.在计划时段开始通过正常渠道订货,单位采购成本最低.在计划时段内零售商还有一次紧急渠道订货机会,可以动态地根据市场需求预测更新结果决策是否订货.当市场需求预测更新过程为MMFE (martingale model of forecast evolution),且在各时点存在不同固定订货成本条件下,证明正常渠道和紧急渠道最优订货策略分别为状态相关的basestock和(s,S)策略,最优订货点与当前时点市场需求预测值线性相关.建立了最优订货策略的求解算法.分析了正常渠道和紧急渠道订货对库存管理的不同意义,结合敏感性分析说明零售商如何在正常渠道和紧急渠道,以及市场需求信息和采购成本间进行权衡.通过大量数值算例发现双渠道订货充分地进行了权衡.     

供应链系统中零售商横向转载的随机规划模型及算法

为了研究随机需求下供应链系统中多零售商的横向转载问题, 提出了一种基于随机规划方法的多零售商横向转载模型, 并设计了一种启发式算法用于模型求解.考察由一个外部供应商和多个零售商组成的供应链系统, 供应商和零售商均采用order-up-to方法控制库存, 零售商面临随机性需求且相互之间存在横向转载.在横向转载经典假设条件下, 构建了以供应链总成本为目标的混合0-1规划模型.考虑到该模型中存在随机变量, 将模型转化为随机期望值规划模型进行求解, 结合随机拟梯度算法及遗传算法设计了一类启发式算法来求解模型, 给出了详细的求解过程.利用数值算例的仿真结果验证了模型及算法的有效性, 还就需求相关程度的影响及转载假设条件的影响进行了详细分析.研究结果表明:模型及算法对存在横向转载的供应链系统确定最优库存水平是相当有效的.     

交通需求非对称价格效应对比分析

交通运输对于石油产品的消耗造成的空气污染已成为我国城市及区域雾霾成分的重要来源.价格是调整供需的主要手段,且具有不对称性,研究石油产品价格变动对交通需求的不对称影响效应具有实际意义.基于此,本文通过价格分解模型,将中国、日本、美国的汽油和柴油价格分解为三部分,探讨其对各国客运需求和货运需求的非对称性影响效应.实证分析结果表明:较高的油价到目前为止并未使中国的交通需求减少,而美国和日本的交通需求与价格均为负相关关系.日本交通需求受油价变化的影响并不显著.中国和美国的客运需求对于油价变化的非对称性效应均表现为价格恢复阶段的影响最大,同时,两国的货运需求对于油价变化较客运需求都更加敏感,但交通需求对油价变化的反应与历史价格相关,美国消费者对于价格上涨的预期与中国消费者有很大的区别.     

Outsourcing decision and order policy with forecast updating under capacity reservation contract

This paper investigates an optimal decision problem in a single-period, two-stage supply chain with capacity reservation contract. At the beginning of the planning horizon, the retailer, who faces stochastic demand, reserves future capacity according to his forecasting of the demand. The supplier then constructs capacity. At the beginning of the selling season, the retailer updates the demand forecasting and places an order. When the retailer＇s demand is greater than the supplier＇s capacity, the supplier can meet the retailer＇s demand by outsourcing. We analyze the optimal decision of each player in both centralized and decentralized systems. Furthermore, under the case in which demand follows a uniform distribution, we obtain the closed-form optimal strategies of each player for both centralized and decentralized systems and conduct numerical studies to reveal additional conclusions. The numerical studies show that the optimal reservation capacity for the retailer and the optimal constructing capacity for the supplier in the decentralized system are both less than the optimal constructing capacity in the centralized system. Furthermore, we also find that the profit loss due to decentralization always exists and increases in indeterminacy.     

我国航空客运量需求预测模型:基于随机前沿预测模型和模型平均

针对多数研究以产品或服务的历史消费量来代替不可观测的需求量而导致的需求预测出现实质性偏差的问题，本文将包含技术无效率项的随机前沿预测模型应用于航空客运量需求的预测，从而有效解决实质性偏差的问题.同时我们在此基础上引入一种模型平均权重确定方法，即通过最小化M折交叉验证准则（CV_M）确定候选模型权重.本文证明了该方法在理论上的最优性.由于模型中技术无效项的存在，我们可以同时预测航空客运量的实际产生量和需求量，实证研究也表明，相比其他常用的预测方法，该方法在预测航空客运量中长期的实际产生量上更具优势.     

带有约束的运输问题及其推广应用

运输问题所讨论的是同种物资的供需调拨问题 ,本文将其推广到供点对需点的供给量或需点对供点的需求量受到一定限制的运输问题 ,并将其经过技术处理后使用表上作业法求解 ,从而避免了使用比较复杂的单纯型法 ,同时将这类问题推广应用到不同物资的运输调拨问题.     

THE DEMAND DISRUPTION MANAGEMENT PROBLEM FOR A SUPPLY CHAIN SYSTEM WITH NONLINEAR DEMAND FUNCTIONS

This paper addresses the problem of handling the uncertainty of demand in aone-supplier-one-retailer supply chain system. Demand variation often makes the real productiondifferent from what is originally planned, causing a deviation cost from the production plan. Assumethe market demand is sensitive to the retail price in a nonlinear form, we show how to effectivelyhandle the demand uncertainty in a supply chain, both for the case of centralized-decision-makingsystem and the case of decentralized-decision-making system with perfect coordination.     

基于Matlab支持向量回归机的能源需求预测模型

分析了支持向量回归机在能源需求预测中的优势, 确定了输入向量集合和输出向量集合, 建立了基于Matlab技术的SVR能源需求预测模型. 对我国1985-2008年能源需求相关数据进行模拟与仿真, 并对中国2010年和2020年能源需求量进行预测. 研究结果表明: 一是中国未来对能源的需求量逐渐增加, 从2010年的330400万吨标准煤上升到2020年418320万吨标准煤, 年均增长率为2.39%; 二是在解决我国能源系统小样本. 非线性及高维模式识别问题中SVR比BP神经网络等方法有更高的预测精度.     

基于两个不确定参数乘积的鲁棒设施选址模型

传统设施选址往往被看作为确定问题,但实际存在需求、成本、风险等不确定因素,这些不确定因素增加了决策的困难.本文在考虑设施选址中单一不确定因素基础上,同时考虑需求和运输成本两个独立参数的不确定性,且在模型中两者为乘积形式,引入两个budget不确定集合刻画不确定性,建立一个新颖的鲁棒设施选址模型,并将非线性问题转化为易求解的鲁棒等价模型,然后通过CPLEX和MATLAB编程求解.最后,以四川西北部的汶川等13个县市的应急物资临时供应点的选址为例,确定最优的选址分配布局.结果表明,较之运输成本的不确定性,需求的不确定性影响更显著,且需求扰动对选址总成本和选址分配方案有明显的影响.决策者可根据其风险偏好程度,选择恰当的不确定水平参数组合,以获得最优的总成本和选址分配方案.     

Centralized Decisions for a Two-Stage Supply Chain with Price-Discount Dependent Demand

In this paper,we study a centralized supply chain for a two-stage with selling price discount.This supply chain consists of a supplier and a retailer. Based on the feature that the product's selling season is short and the supply chain faces great demand uncertainty. We consider a two-stage scenario where,at the beginning of stage 1,the supplier reserves production capacity based on historic data in advance,stage 2 comes to us after some leadtime,both the supplier and the retailer update the demand information,the retailer then places an order not exceeding the reserved capacity based on the selling-pricing discount dependent demand. We make optimal decisions on the reserved capacity in stage 1,selling price discount and order quantity in stage 2. In this supply chain,the pattern in stage2 is figured out first,and then stage 1 is cleared as well. Then we present a numerical example to give some insights. Finally we get some conclusions.     

需求不确定状态下的工厂选址和规模决策的综合优化方法

提出一种整合的工厂选址研究方法,在工厂选址的同时确定所选取工厂的生产规模.改变过去单一不变的成本项,该研究假设随着工厂生产规模扩大,固定建设成本投入增加,单位生产成本减少.建立了消费者需求不确定状态下的两阶段随机规划模型.为了求解这一NP困难问题,设计了Benders分解算法并且用Matlab 7.0编程对所提出的算法进行检验.用随机产生的一组数据进行实验,计算结果显示所提供的算法是有效的.     

中国能源需求的结构突变研究

中国的能源需求是理论界研究的一个热点。本文基于结构突变理论，研究了数据生成过程中的几种不同结构突变模式，并对结构突变的单位根过程和结构突变的趋势稳定过程给出了不同结构突变模式的检验方法。文章还对中国能源需求的数据生成过程进行了实证研究。我们的研究表明，中国能源需求的数据生成过程是带有结构突变的趋势稳定过程，这说明中国能源需求将会沿着确定的均衡增长路径平稳增长。     

供需不确定情形下基于期权的血液供应链优化

在供需不确定决策环境下,构建了基于期权的血液供应链决策模型,通过对模型的优化分析,以期揭示期权合约和供需不确定对血液供应链的决策影响.首先,在供需不确定和期权合约下,分别得出了医院的最优血液产品的采购策略(包含血液产品的最优常规采购量和最优期权采购量),以及血液中心的最优血液采集和生产策略.其次,将血液供应链在分散和集中决策下的策略进行比较,得出了实现血液供应链协调的合作条件.该协调条件不包含血液产品的采购价,只依赖于血液产品供应特征的影响.最后,论文证明了在血液供应链协调时,存在一种可以实现帕累托最优的供应链合作方案.     

基于动态预测企业生产计划模型

在静态的环境下 ,生产计划的决策主要来源于市场的需求预测 ,但是 ,如果在动态环境里 ,由于预测信息本身在不断地变化 ,预测值在不断地改变 ,如何将动态预测信息 (预测修正量 )转化成生产计划 ?研究需求信息向生产计划的动态转变与及时更新 ,具有很强的理论价值和重要的实际意义     

两企业序贯进入市场时的最优进入时间及定价分析

在竞争环境下,当企业向未知市场推出新产品时,由于市场需求的不确定性,企业需要平衡好其较早进入市场带来的风险和较晚进入市场所丢失的机会。建立2个企业分别以先行者和后行者身份向同一市场推出可替代新产品的双寡头竞争模型,研究了后行者的最佳进入时间,以及双方的最优定价问题。使用微分对策理论分析发现,在两种不同的市场需求函数形式(指数形式和线性形式)下,当考虑成本学习效应时,双方的最优价格均是随着已进入市场的时间递减的,并给出了不同条件下,后行者的最优进入时间。数值分析进一步展示了一些常见运作参数对后行者进入时间及双方定价的影响规律。