用微信扫码二维码

分享至好友和朋友圈

INTELLIGENT OPTIMZATION

2023

第一轮通知

智能优化

应用实践

研讨会

4/7-4/9

由湖北省系统工程学会主办，华中科技大学管理学院承办，数据魔术师， Robust Team，《Frontiers of Engineering Management》期刊协办的2023年智能优化应用实践研讨会拟于2023年4月7日至9日在湖北武汉举行。

本次研讨会主要聚焦如何应用各类智能优化技术去解决生产制造和物流供应链领域不同场景中遇到的决策优化问题，涉及的关键词有“运筹优化”，“组合优化问题”，“调度”，“整数规划”，“库存优化”，“鲁棒优化”，“网络设计”，“算法”等等，旨在打造一个业界和学界沟通的平台和桥梁，深度交流智能优化技术，探讨推动智能优化算法在企业管理中的应用和落地，让智能优化技术真正的为社会和企业创造价值。

此次研讨会所有报告线下举行

不设线上直播、不录像

研讨会的日程如下：

拟3月18日-3月30日期间开放注册，注册方式请静候通知，欲参加此次研讨会的学者请通过下方微信二维码联系会务组，然后由会务组邀请至会议微信群。由于场地限制，本次研讨会参会人数计划不超过300人。

会务组微信

有任何疑问或者想咨询的，请通过下方微信二维码联系秦虎教授

主旨报告嘉宾

-按报告顺序排列-

徐宙，香港理工大学工商管理学院物流及航运学系教授。清华大学计算机系本科，新加坡国立大学计算机系硕士，香港科技大学工业工程及工程管理系博士。主要研究运筹学理论和智能优化及其在物流、交通、航运和供应链管理等方面的应用。在《Operations Research》、《INFORMS Journal on Computing》、《Production and Operations Management》、《Transportation Science》、《Transportation Research Part B》等学术期刊上发表论文40余篇。

报告题目：求解连续时间运输服务网络设计的新算法

报告摘要：The continuous-time service network design problem (CTSNDP) occurs widely in practice. It aims to minimize the total operational cost by optimizing the schedules of transportation services and the routes of shipments for dispatching, which can occur at any time point along a continuous planning horizon. In order to be cost effective, shipments often wait to be consolidated, which incurs a holding cost. Despite its importance, the holding cost has not been taken into account in existing exact solution methods for the CTSNDP, since introducing it significantly complicates the problem and makes the solution development very challenging. To tackle this challenge, we develop a new dynamic discretization discovery algorithm, which can solve the CTSNDP with holding cost to exactly optimum. The algorithm is based on a novel relaxation model and several new optimization techniques. Results from extensive computational experiments validate the efficiency and effectiveness of the new algorithm, and also demonstrate the benefits that can be gained by taking into account holding costs in solving the CTSNDP. In particular, we show that the significance of the benefits depends on the connectivity of the underlying physical network, and on the flexibility of the shipments' time requirements.

彭一杰，北京大学光华管理学院副教授，博士生导师。北京大学人工智能研究院、国家健康医疗大数据研究院兼职研究员。本科毕业于武汉大学数学与统计学院，从复旦大学管理学院获博士学位。在美国马里兰大学和乔治梅森大学分别从事过博士后与助理教授工作。主要研究方向包括仿真建模与优化、人工智能、金融工程与风险管理、健康医疗等。主持多项科研基金项目，包括国家自然科学基金委优秀青年科学基金，原创探索项目等。在《Operations Research》，《INFORMS Journal on Computing》和《IEEE Transactions on Automatic Control》等高质量期刊上发表学术论文30余篇。曾获得2019年INFORMS Outstanding Simulation Publication Award。目前担任Asia-Pacific Journal of Operational Research副主编、《系统管理学报》领域主编，北京运筹学会副秘书长，中国运筹学会金融工程与金融风险管理分会常务理事，管理科学与工程协会理事。

报告题目：智慧运营管理——仿真优化与人工智能方法及其在企业运营中的应用案例

报告摘要：本报告介绍仿真优化与人工智能方法及其在企业运营中的应用案例。案例一为国际知名高端装备制造与服务商。针对高端装备制造工艺复杂、生产制造周期长和订单高度定制化等特点，采用自上而下的顶层设计和自下而上建设的路径达到降本增效：针对不同车间的生产特点设计模型和算法实时决策，提升车间产能利用率并有效应对突发情况；实现多车间的高效协同打通车间壁垒，从全局视角减少订单生产周期；融合销售、采购和生产的数据，通过仿真和优化技术代替经验式评估，以统一的平台对各类情境进行预测分析避免部门各自为战，提升订单交付能力与客户满意度。案例二为本土大型电商平台。应用深度强化学习解决多级供应链库存订购的策略优化问题，并进一步基于多智能体放松需求平稳、供应链主体共享全部信息的要求，降低供应链总库存成本与牛鞭效应。最后，针对复杂供应链的仓网结构、仓库容量约束、海量库存单位（SKU）等现实困难，提出供应链库存管理仿真建模、优化与应用的一体化智能解决方案。

杨超林，上海财经大学信息管理与工程学院常任教授，博士生导师，数据、算法与工程系系主任，校创新团队首席专家。长期从事供应链管理、库存管理、数据驱动的运营管理等方面的研究。代表性研究成果发表在MS，OR，POM等国际顶尖期刊。主持国家自科优青、面上、青年基金，以及上海市人才发展资金。研究成果曾获2015年度POMS-HK国际会议最佳学生论文奖二等奖和2015年度Service Systems and Service Management国际会议（ICSSSM15）最佳论文奖。担任杉数科技科学家顾问，为国内多家零售、物流、制造企业（包括京东、顺丰科技、华为等）提供库存管理与优化咨询服务。指导学生获得首届阿里云基础设施供应链大赛亚军。

报告题目：大规模库存网络的安全库存优化：一种基于网络分解的方法

报告摘要：Optimizing safety stock placement on a large-scale inventory network is challenging since it may involve a massive number of nodes and many shared materials. In this paper, we study how to solve the large-scale guaranteed service model (GSM) to handle this problem. This paper presents a novel approach, called the iterative decomposition (ID) approach, to efficiently solve the large-scale guaranteed service model (GSM). The ID approach decomposes the problem iteratively by using a non-convex optimization method (sequential linear programming) to select the appropriate algorithm to solve each sub-problem based on its specific network structure. Our approach is evaluated against several methods from the literature using real and generated large-scale inventory networks. Our numerical experiments show that the ID approach performs best, especially when the network size is large and the network structure is complex. We provide a Python library for our solution approach, called InvNet that can be found at

https://github.com/durianh96/InvNet-InventoryNetworkOptimizationLab.

罗志兴副教授于2010年在华南理工大学获得学士学位，于2014年在香港城市大学获得博士学位，现为南京大学工程管理学院副教授，主要研究的领域是运筹优化算法设计、智慧物流、智能制造等。他主持国家自然科学基金青年项目、面上项目和优秀青年项目各一项，在国际知名期刊Manufacturing & Service Operations Management、INFORMS Journal on Computing、Transportation Science、Transportation Research Part-B: Methodological以及Naval Research Logistics发表论文十多篇。他2018年参加京东物流举办的“全球运筹优化挑战赛”，在城市物流运输车辆智能调度赛题获得冠军，2020年入选中国科协青年人才托举工程。

报告题目：On-Demand Delivery from Stores: Dynamic Dispatching and Routing with Random Demand

报告摘要：On-demand delivery has become increasingly popular around the world. Motivated by a large grocery chain store who offers fast on-demand delivery services, we model and solve a stochastic dynamic driver dispatching and routing problem for last-mile delivery systems where on-time performance is the main target. We propose a novel structured approximation framework to approximate the value function via a parametrized dispatching and routing policy. We analyze the structural properties of the approximation framework and establish its performance guarantee under large-demand scenarios. We then develop efficient exact algorithms for the approximation problem based on Benders decomposition and column generation, which deliver verifiably optimal solutions within minutes. The evaluation results on a real-world data set show that our framework outperforms the current policy of the company by 36.53% on average in terms of delivery time.

魏丽军，广东工业大学机电学院教授、博导。香港城市大学博士、新加坡国立大学博士后。长期致力于离散制造过程中优化问题求解算法的研究，提出了通用算法求解框架与系列领域依赖的专用智能算法，涵盖精确求解算法和智能优化算法两大体系，研究问题包括装箱配载、排样下料、车辆路径优化和调度排产。累计发表SCI论文30余篇，主持国家重点研发计划课题1项、国家自科面上项目2项、青年项目1项。申请发明专利30余项，相关技术成果已在尚品宅配、兴森快捷等行业龙头企业应用，经济效应明显。

报告题目：考虑实际布线集成电路布局规划方法

报告摘要：布局规划是集成电路设计中关键的一环，传统布局规划方法由于未考虑实际布线，导致后续布线频繁失效，从而严重影响集成电路设计效率。本研究将布线因素加入布局过程的评价策略中，提出了求解考虑实际布线集成电路布局快速启发式算法，并引入布线代理模型快速评价布局方案的优劣，通过机器学习方法和二分类策略提高算法的搜索效率，从而能够在短时间内得到合理的布局方案，提高了布线成功率，提升了布线性能。

张真真，同济大学经济与管理学院副教授、博士生导师。入选上海市高层次人才计划（2019）。曾担任新加坡国立大学工业工程系助理教授，于香港城市大学管理科学系获得博士学位，厦门大学计算机科学系获得硕士和学士学位。长期致力于大规模整数规划和鲁棒优化的理论研究与算法设计，及在物流与运输规划、智能制造等方面的应用。目前已发表SCI/SSCI期刊论文20多篇，包括Operations Research（1篇）、Transportation Science（3篇）、Transportation Research Part B（3篇）等。现任世界交通大会（WTC）货运规划与物流管理技术委员会委员、管理科学与工程学会交通运输分会执行秘书长，并长期担任Operations Research，Transportation Science等30多个国际知名期刊的审稿人。

报告题目：机场行李运输中的复杂车辆路径问题研究

报告摘要：In this study, we focus on the routing and scheduling of multi-carriage transit trains for airport baggage transit. It is modeled as a vehicle routing problem with cross-route dependencies caused by side constraints, including split demand, multiple trips per vehicle, and simultaneous pickup and delivery. Besides, some other practical constraints such as time windows, baggage release time, baggage waiting time, and priority of unloading are taken into account in the implementation. The joint consideration of these characteristics brings a unique challenge to determine the start time of each service for aircraft due to the interdependencies across vehicle routes. Thus, we adopt topological sort to construct the directed acyclic graph and then derive the flight service time. Based on that, we develop an Adaptive Large Neighborhood Search (ALNS) algorithm. Moreover, in order to examine the moves quickly, a two-stage solution evaluation method is proposed, where the single tour is checked based on the segment-based evaluation method in the first stage, and the complete solution is checked using the topological sort in the second stage. The results demonstrate the superiority of our algorithm in computational time and solution quality. In addition, some insightful conclusions are drawn through detailed analyses.

寻其锋，三一集团AI算法团队负责人，算法科学家。中国科学院数学与系统科学研究院博士。主要研究方向为智能优化算法、智能应用落地场景。发表SCI、EI论文4篇。

报告题目：智能应用落地场景及方法

报告摘要： 制造业当前快速发展，对无人化、无纸化、自动化、降本增效的需求越来越迫切，而智能优化算法、机器学习预测等在制业的落地场景丰富，具体而言，首先介绍一些智能应用场景及落地方法。其次，讨论低空城市物流的无人机、物流配送车的调度系统，最后，给出智能应用场景的一些可行研究方向。

邵赛俊，逗号科技联合创始人。2013年于浙江大学取得学士学位、2017年于香港大学取得博士学位，曾获香港政府全额奖学金。主要研究方向为城市物流、运筹优化算法。在TRB，TRE，IISE Transactions (获年度最优论文奖)，IJPE等知名国际顶级期刊上发表10余篇高水平研究论文，主持国家自然科学青年基金项目。邵赛俊已主导为多家国内外知名企业打造供应链算法产品与服务，典型案例包括：顺丰DHL、联想、国药、上药、华润三九、中远海运、中烟、海尔等，涉及快消、零售、汽车、生鲜、医药、冷链、航运、制造等行业，拥有丰富的供应链系统咨询和算法综合解决方案能力。

报告题目：“用算法对抗复杂性”——后疫情时代下的国际多式联运

报告摘要：三年疫情期间，国际物流的计费结构、运力构成、服务模式都发生了巨大而深刻的变化。特殊的时代逐渐褪去，全球多式联运的复杂性却被保留了下来。本次报告将分享一个立足于中国的制造业巨头的案例，回顾我们为其打造的算法引擎如何应对国际物流遇到的前所未有的挑战：海运船期极其不稳定；空运运力碎片化严重、原本一家货代就能满足的需求，如今需要拼凑八家、组合优化才能满足货运需求；而持续至今的地缘冲突甚至直接切断了中欧班列的铁路运输...本次报告还将展望在疫情时代练就的动态路由规划能力，在未来又将如何持续助力其精细化运作，成为国际端到端的物流运输降本增效的核心大脑。

徐亮，西南财经大学大数据研究院教授、博导。北京师范大学本科，中山大学硕士、香港理工大博士。现任交易软件公司乾隆科技首席科学家，四川省首批天府万人计划入选者。主要从事物流、供应链、组合投资等运营管理相关研究。主要研究兴趣包括车辆路径和鲁棒优化，在车辆路径领域主要提出了多个车辆路径公开问题的近似算法，在鲁棒优化领域主要提出了“非参鲁棒优化”。主持国家自然科学基金项目3项。在Manufacturing &Service Operations Management, INFORMS Journal on Computing和Transportation Research: Part B等期刊发表第一作者或通讯作者文章。获银保监会科技进步二等奖、中国期货业协会联合研究项目二类优秀成果奖。受邀担任Financial Innovation期刊专刊首席客座主编，并担任编委。任系统工程学会理事、天府对冲基金学会金融产品创新委员会主任委员。

报告题目：A robust approach to airport gate assignment with a solution-dependent uncertainty budget

报告摘要：Airport gate assignment (AGA) is a critical issue for airport operations management. It aims to assign flights to gates according to their arrival and departure times. To tackle flight delays in airports, we propose a robust airport gate assignment (RAGA) to minimize the (1-α) -quantile of the total real-time overlap between consecutive flights at the same gate, namely, the total gate blockage time, so that the realized total gate blockage time is worse than its quantile with a probability, at most α. Given any constant, we develop an asymptotically tight upper bound for the violation probability that total gate blockage time is worse than the constant. Based on the upper bound, a solution-dependent uncertainty budget is introduced to develop a robust counterpart (RCP) for the RAGA. We further de- velop a solution technique for the RCP by transforming the problem into a finite number of tractable binary programmings. An empirical study of the Shuangliu International Airport (CTU) indicates that our proposed robust approach for AGA outperforms existing methods.

夏俊，上海交通大学中美物流研究院副研究员。香港理工大学物流与航运系博士。主要研究大规模整数规划求解技术及其在物流与交通运输管理决策中的应用。《Transportation Science》、《Transportation Research Part B》、《Naval Research Logistics》等学术期刊上发表论文9篇。

报告题目：物流与供应链数智化实践分享

报告摘要：随着互联网快速发展，新的商业模式和市场环境给物流和供应链发展带来了机遇和挑战。本次报告首先介绍当前物流与供应链领域的发展特点和数智化变革需求；其次介绍美的安得智联的“1+3”的全链路物流服务模式，及其与运筹优化相关的应用场景；最后从产学研合作的视角探讨理论与实践之间的链接。

章宇，西南财经大学大数据研究院教授、博导。东北大学本科、直博，新加坡国立大学联合培养博士。曾赴新加坡国立大学任研究员，并多次受邀访问。主要从事物流、供应链、交通、医疗运营管理中的鲁棒优化与决策研究。主持和参与国家自然科学基金项目3项。在Operations Research，Mathematical Programming，Production and Operations Management, INFORMS Journal on Computing等期刊发表学术论文10余篇。获中国管理科学与工程学会优秀博士学位论文奖、Omega期刊最佳论文奖，单篇论文入选ESI高被引论文。受邀担任Operations Research，INFORMS Journal on Computing，Transportation Science等期刊审稿人，任中国运筹学会决策科学分会理事。

报告题目：基于患者特征信息的手术室调度鲁棒优化方法

报告摘要：Patient features such as gender, age, and underlying disease are crucial to improving the model fidelity of surgery duration. In this paper, we study a robust surgery scheduling problem augmented by patient feature segmentation. We focus on the surgery-to-operating room allocations for elective patients and future emergencies. Using feature data, we classify patients into different types using machine learning methods and characterize the uncertain surgery duration via a feature-based cluster-wise ambiguity set. We propose a feature-driven adaptive robust optimization model that minimizes an overtime riskiness index, which helps mitigate both the magnitude and probability of working overtime. The model can be reformulated as a second-order conic programming problem. From the reformulation, we find that minimizing the overtime riskiness index is equivalent to minimizing a Fano factor. This makes our robust optimization model easily interpretable to healthcare practitioners. To efficiently solve the problem, we develop a branch-and-cut algorithm and introduce symmetry-breaking constraints. Numerical experiments demonstrate that our model outperforms benchmark models in a variety of performance metrics.

Dr. Shuai Jia （贾帅）is an assistant professor with the Intelligent Transportation Thrust, The Hong Kong University of Science and Technology (Guangzhou). He obtained a PhD in Transport and Logistics Management from The Hong Kong Polytechnic University in 2019, and was a research fellow with the Institute of Operations Research and Analytics and the Department of Civil and Environmental Engineering, The National University of Singapore during 2020 and 2021. His research interest lies in the application of operations research and artificial intelligence methods to solve problems arising from transportation and logistics management. His has published papers in leading journals such as Transportation Science and Transportation Research Part B.

报告题目：Robust Yard Crane Deployment and Truck Service Management for Gate Congestion Mitigation in a Container Terminal

报告摘要：Container terminals are critical hubs in the global supply chain, as they provide essential facilities and services for cargo transshipment between land transportation and marine transportation. In the event of global supply chain disruptions, the container transportation demand of road trucks in a port area can be highly uncertain, causing severe truck congestion at container terminal gates. When congestion occurs at a terminal gate, the long waiting line of trucks can block the traffic, hindering the terminal operations and lowering the container handling efficiency. An important approach to gate congestion mitigation is to proactively organize the utilization of yard resources to effectively serve the container handling requests of trucks. In this paper, we propose a two-stage robust optimization method for truck service management to enable terminal operators to hedge against uncertain number of truck arrivals and mitigate gate congestion. In our two-stage robust optimization model, the first-stage deploys yard cranes among yard blocks over a planning horizon, while the second-stage assigns trucks to yard blocks when the number of truck arrivals becomes available. The goal of the model is to minimize the worst-case average truck queue length at the gate, subject to various operational constraints such as compatibility between trucks and yard blocks, limited crane service capacity, and prioritized vessel service requests. We generate an optimal solution of the robust optimization model using a tailored column-row generation method. The proposed model and solution method are evaluated on problem instances generated from real operational data of a container terminal in Shanghai.

陈奇，武汉云筹优化科技有限公司产品经理，从事国产自主APS高级计划与排程软件研发10多年，熟悉国内外主流的APS软件产品，参与几十家制造企业的高级计划排程系统项目选型与实施交付工作，项目实战经验丰富。武汉云筹优化科技有限公司是国内领先的供应链高级计划与排产解决整体方案服务商，专注计划优化领域10余年，自主研发了国产APS高级计划与排程工业软件，为制造企业提供全面的预测计划，采购计划，产能规划，人力计划，MPS/MRP，主生产计划，工序计划，装车计划，配送计划等软件模块，覆盖中长期计划与短周期排产等供应链计划业务场景，助力企业的数字化智能车间改善与产业转型升级，云筹APS软件在中国中车，上海电气，中国烟草机械，沈阳鼓风机厂，汽车零配件等数十个行业成功应用，帮助制造企业解决生产计划难题。

报告题目：大型装备制造企业的生产计划调度优化问题应用探讨

报告摘要：结合APS高级计划与排程软件在大型装备制造企业的项目实战经验，分享运筹优化算法在大型装备制造企业中的应用流程，应用难点，以及应用效果等。

兰鹏有十多年制造、零售消费品行业物流信息化经验，服务过迈瑞医疗、广州宝洁、 阿克苏诺贝尔、梅塞尔、延锋江森、香港联合书刊、 哈药集团、ABB、云南建投、屈臣氏等行业知名企业的仓储、运输及订单管理项目，以及沃尔玛、家家悦的运输优化项目。上海科箭软件科技有限公司成立于2003年，作为一家供应链云服务提供商，致力于帮助企业构建更敏捷、更高效、更智慧的数字化供应链网络，实现供应链全流程可视化。科箭供应链管理云平台-Power SCM Cloud，是一个整合订单管理（OMS云）、运输管理（TMS云）、仓储管理（WMS云）、预约管理（AMS云）、供应链控制塔（SCCT）的云解决方案。

报告题目：科箭智能优化方案应用实践

报告摘要：科箭不断探索并实践数据及AI在供应链执行层面的落地与应用。2019年起，科箭与华中科技大学成立人工智能实验室，研发落地运输优化-TOS云、装载优化-LOS云等AI产品，助力客户智能升级。

张子臻，中山大学计算机学院副教授。香港城市大学博士，香港科技大学、美国新泽西理工大学访问学者。主要研究方向为智能优化算法、深度强化学习、系统决策与组合优化、智能交通与物流。主持了国家与省部级自然基金项目5项以及多项企业委托课题，在国际知名期刊与会议发表论文60余篇，其中IEEE Trans.论文20余篇。担任中山大学ACM-ICPC国际大学生程序设计竞赛总教练，华为云算法创新Lab赛事合作老师。

报告题目：深度强化学习在路径优化中的应用

报告摘要：深度强化学习是近年来较热的技术，本次报告将介绍深度强化学习的一些基本概念，并探讨如何将其用于求解一些组合优化问题，特别是路径优化问题。具体而言，首先介绍深度强化学习的一些经典方法。其次，讨论深度强化学习求解路径优化问题的一些通用范式，并介绍深度强化学习在一类动态路径优化问题的应用。最后，给出深度强化学习求解路径优化问题可行的一些研究方向。

李纪柳，2017年7月-2018年7月任香港理工大学商学院的助理研究员；2020年8月-2022年12月为华中科技大学管理学院博士后；2021年2月-2022年2月任香港理工大学工业工程系助理研究员；2023年1月 - 至今任西北工业大学管理学院教授。研究兴趣为包括运筹优化理论及其相关应用：基于APS（高级-排程）系统的排产调度，车辆路径优化，电网电力巡查系统优化，基础设施投资与优化，资源调度与分配；组合优化算法设计：针对不同组合优化问题的各类精确算法和启发式算法的设计，包括Benders分解、Branch-and-Price算法、Branch-and-Cut算法、禁忌搜索、大领域搜索等。参与多项科研项目，在INFORMS Journal On Computing、Transportation Science、Transportation Research Part B: Methodological等著名期刊上发表过多篇论文。

报告题目：Large-Scale Facility Inspection Problem with UAV and Autonomous Battery Swap Station

报告摘要：In this paper, we address a novel variant of the location-routing problem that arises in the context of unmanned aerial vehicle (UAV)/drone inspections of large-scale mega-facilities (LRP-D).The problem involves locating multiple automatic battery swapping stations (BSSs) at candidate locations, assigning facilities from different locations to the BSSs while considering duration constraints, and designing UAV routes to inspect facilities while taking power constraints into account.The objective is to minimize the total cost, including fixed BSS costs and UAV travel costs.To solve this problem, we propose a mixed-integer linear programming (MILP) formulation that can be solved directly using CPLEX. Additionally, we develop a novel logic-based Benders decomposition algorithm (LBBD) for practical-sized instances. The LBBD splits the problem into a master problem (MP) that determines BSS locations and facility assignments to BSSs and a set of independent multiple travel repairman subproblems (MTRPs). The master problem is solved using a branch-and-cut procedure operating on a single search tree. Once an incumbent solution is found, the subproblems are solved to generate cuts dynamically added to the master problem.Due to the NP-hardness of MTRP, we develop a Branch-and-Price method (B&P) for the subproblems. We propose a combinatorial Benders cut as a feasibility cut and two optimality cuts based on the subproblems’ optimality condition. In the computational study section, we analyze the efficiency of our formulation and LBBD and perform sensitivity analysis on critical parameters.The numerical results on five scales of randomly generated instances show that LBBD performs better than the MILP formulation.

特别声明：以上内容(如有图片或视频亦包括在内)为自媒体平台“网易号”用户上传并发布，本平台仅提供信息存储服务。

Notice: The content above (including the pictures and videos if any) is uploaded and posted by a user of NetEase Hao, which is a social media platform and only provides information storage services.