Computers & Operations Research最新文献

{"title":"A fully adaptive framework for continuous-state stochastic dynamic programming","authors":"Huiyuan Fan ,&nbsp;Prashant K. Tarun ,&nbsp;Amith Viswanatha ,&nbsp;Victoria C.P. Chen","doi":"10.1016/j.cor.2025.107160","DOIUrl":"10.1016/j.cor.2025.107160","url":null,"abstract":"<div><div>Approximate dynamic programming (ADP) carries out approximation of the future value function (FVF) to enable numerical solutions to dynamic programming (DP). Recent ADP methodologies often employ the design and analysis of computer experiment (DACE) techniques for the FVF approximation. Use of DACE-based ADP approach, however, creates a “chicken and egg” situation where we cannot collect the data for statistical modeling until we know the state space region, but we do not know the state space region until we collect the data. To overcome this dilemma, this paper introduces a sequential state space exploration (SSSE) approach to adaptively identify the state space region for the experimental design while also sampling useful data for the statistical model. In the proposed methodology, the SSSE approach works in tandem with an adaptive value function approximation (AVFA) algorithm that gradually grows the complexity of the statistical model as more data are observed. This novel SSSE-AVFA approach features a “<em>fully adaptive dynamic programming</em>” algorithm, which can automatically and appropriately identify the three critical components (<em>state space region</em>, <em>sample size of the data</em>, and <em>statistical model structure</em>) for FVF approximation, thereby eliminating the need for time-consuming trial-and-error computational runs that were previously required. The SSSE-AVFA approach is examined with a nine-dimensional inventory forecasting problem and is compared with fixed structure runs in which the state space region, sample size of the data, and statistical model structure are assumed in advance. Our proposed methodology ensured either that the established solutions could be more reasonable or that the modeling process could effectively save the computational effort. With its full adaptiveness in determining those critical components, the SSSE-AVFA approach has the potential to be more effective and efficient than the traditional methods in handling a wide range of real-world continuous-state DP problems.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107160"},"PeriodicalIF":4.1,"publicationDate":"2025-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144279803","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Graph neural network and expert-guided deep reinforcement learning for solving flexible job-shop scheduling problem","authors":"Wenqiang Zhang ,&nbsp;Xuan Bao ,&nbsp;Huili Geng ,&nbsp;Guohui Zhang ,&nbsp;Mitsuo Gen","doi":"10.1016/j.cor.2025.107155","DOIUrl":"10.1016/j.cor.2025.107155","url":null,"abstract":"<div><div>Flexible Job-shop Scheduling Problem (FJSP) is crucial for efficient and flexible automated production. Recent advancements in Reinforcement Learning (RL) have shown promise in solving sequential decision problems. However, most Deep Reinforcement Learning (DRL) algorithms rely on Priority Dispatching Rules (PDRs), which limits scheduling efficiency. This paper proposed a novel Graph Neural Network and Expert-Guided Deep Reinforcement Learning (GNN-EGDRL) framework that employs Graph Neural Network (GNN) to extract features from both machines and operations, thereby integrating operation selection and machine assignment into a unified composite decision-making process. Expert solutions generated by PDRs guide the initial stages of training, allowing agents to gradually transition to self-directed action selection, thus balancing exploration and exploitation. The expert guidance strategy emphasizes the relationships between operations and machines, enhancing the model’s sensitivity and providing granular guidance for feature extraction, leading to optimized solutions. Extensive experiments demonstrate that the proposed GNN-EGDRL method consistently outperforms traditional PDRs and other end-to-end DRL over all problem instances. Notably, this superiority has been validated in a wide range of scenarios, including larger-scale examples, underscoring the method’s scalability and robustness. Furthermore, the method exhibits strong performance in scenarios not encountered during training, highlighting its effectiveness and adaptability in diverse production environments.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107155"},"PeriodicalIF":4.1,"publicationDate":"2025-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144291019","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Random-Key Optimizer with reinforcement learning for the Capacitated Multi-period Cutting Stock Problem with setup cost","authors":"Eduardo M. Silva ,&nbsp;Antônio A. Chaves ,&nbsp;Silvio A. de Araujo ,&nbsp;Raf Jans","doi":"10.1016/j.cor.2025.107159","DOIUrl":"10.1016/j.cor.2025.107159","url":null,"abstract":"<div><div>This paper introduces a Random-Key Optimizer (<em>RKO</em>) procedure incorporating reinforcement learning to solve the One-Dimensional Multi-Period Cutting Stock Problem (<em>MPCSP</em>) with setup costs and capacity constraints. The <em>MPCSP</em> involves determining cutting plans for each period to meet customer demands, where inventory variables link consecutive periods. The <em>RKO</em> represents solutions as random-key vectors, which are decoded into feasible solutions for the <em>MPCSP</em> through a decoder process. During the optimization process, the <em>RKO</em> dynamically adapts its parameters using reinforcement learning. This framework integrates Biased Random-Key Genetic Algorithm (<em>BRKGA</em>), Particle Swarm Optimization (<em>PSO</em>), and Simulated Annealing (<em>SA</em>), all utilizing a unified decoder function. A novel penalization mechanism is also introduced within the decoder to handle infeasibilities effectively. The proposed <em>RKO</em> is evaluated on benchmark instances from the literature and compared against state-of-the-art methods, including a hybrid column generation heuristic and a dynamic programming-based heuristic. In addition, a new set of large-scale instances is introduced for further evaluation. Computational experiments reveal that the <em>RKO</em> employed by <em>BRKGA</em> consistently outperforms other solution methods in benchmark instances, delivering superior average solution quality. A sensitivity analysis is also conducted, examining the impact of setup costs and production capacity. Moreover, the study includes a comparative analysis of the <em>RKO</em> framework with and without reinforcement learning.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107159"},"PeriodicalIF":4.1,"publicationDate":"2025-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144291017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On the multiple optimal solutions and patterns of the pallet loading problem","authors":"Junmin Yi ,&nbsp;Mingming Li ,&nbsp;Yiping Lu","doi":"10.1016/j.cor.2025.107182","DOIUrl":"10.1016/j.cor.2025.107182","url":null,"abstract":"<div><div>The Pallet Loading Problem is a common optimization issue in the fields of manufacturing, transportation, and logistics. It involves placing as many identical rectangular boxes as possible onto a rectangular pallet. As methods for solving this problem continue to advance and gain widespread adoption, it has become common to find multiple optimal solutions for various pallet-loading scenarios. However, this phenomenon remains underexplored in literature. This study identifies and analyzes the characteristics of multiple optimal solutions for the Pallet Loading Problem, offering insights into their patterns, transformations, and practical applications, thus bridging the gap in the current literature. We present a substantial number of instances, solutions, and practical examples by analyzing the results in relation to the identified pattern features. We also identify the conditions for achieving single optimal patterns and perfect patterns while exploring the concepts and applications of perfect partitions, perfect contours, and perfect patterns. This study aimed to enhance the understanding of the solutions and loading patterns of the problem, ultimately facilitating the generation and implementation of perfect or near-perfect pallet loading patterns for real-world logistics applications.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107182"},"PeriodicalIF":4.1,"publicationDate":"2025-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144298471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A tabu search algorithm for Unspecified Parallel Machine scheduling with shift consideration","authors":"Ponpot Jartnillaphand ,&nbsp;Elham Mardaneh ,&nbsp;Hoa T. Bui","doi":"10.1016/j.cor.2025.107151","DOIUrl":"10.1016/j.cor.2025.107151","url":null,"abstract":"<div><div>This paper addresses the Unspecified Parallel Machine Flexible Resource Scheduling (UPMFRS) problem with shift consideration, focusing on assigning jobs to parallel machines while accounting for shifts and worker breaks, a practical aspect often overlooked in the literature. In this problem, teams of workers are treated as machines, and the duration of each job depends on the number of workers assigned to the team. We propose a two-stage algorithm combining bin-packing with a job scheduling heuristic to generate initial solutions. In the first stage, jobs and resources are allocated to active teams, while in the second stage, jobs are scheduled for each team. Then, the initial solutions are refined using a novel tabu search algorithm designed to handle the complexities of the problem. Our tabu search integrates neighborhood exploration techniques and strategic move selection to avoid local optima. The proposed algorithm’s performance is compared with the exact methods, the branch and cut in CPLEX, and the state-of-the-art bilinear branch and check (BBCh) algorithm. Numerical experiments indicate that our tabu search algorithm generates high-quality solutions. When these solutions are used as a warm start for BBCh (hybrid BBCh), BBCh’s performance is significantly enhanced. While BBCh alone and CPLEX can solve instances with up to 35 jobs, tabu search and the hybrid BBCh successfully handle problems with up to 100 jobs. These results confirm that the hybrid approach, with the high-quality solutions provided by our tabu search algorithm, is highly effective, practical, and reliable for large-scale scenarios while maintaining reasonable computational times.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107151"},"PeriodicalIF":4.1,"publicationDate":"2025-06-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144272572","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Two-stage distributionally robust optimization with a finite support","authors":"Agostinho Agra","doi":"10.1016/j.cor.2025.107142","DOIUrl":"10.1016/j.cor.2025.107142","url":null,"abstract":"<div><div>We consider two-stage distributionally robust mixed-integer problems where the uncertain parameters have discrete support. We propose an ambiguity set based on the feasible set of a transportation problem with a single knapsack constraint, extending the well-known Kantarovich ambiguity set in order to model a wider set of practical situations. The properties of this set are analysed. Based on different approaches to model the second-stage decisions and to impose the worst-case expected cost, three solution approaches are discussed: the Benders-like method proposed by Bansal, Huang, and Mehrotra (2018), a single-stage model obtained from the dualization of the transportation problem, and an epigraph formulation that enforces the expected cost through a series of optimality cuts which are generated dynamically. To evaluate the approaches a location-transportation problem is considered. Computational tests based on the three proposed approaches show that the best approach depends on the characteristics of the ambiguity set considered.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107142"},"PeriodicalIF":4.1,"publicationDate":"2025-06-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144264036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Decomposition heuristics for the Hybrid Flexible Flowshop with transportation times","authors":"Michele Garraffa ,&nbsp;Helmut Simonis ,&nbsp;Barry O’Sullivan ,&nbsp;Eddie Armstrong","doi":"10.1016/j.cor.2025.107145","DOIUrl":"10.1016/j.cor.2025.107145","url":null,"abstract":"<div><div>This paper proposes efficient heuristic approaches for the Hybrid Flexible Flowshop with Transportation Times (HFFTT), an extension of both the Hybrid Flowshop (HFP) and Hybrid Flexible Flowshop (HFF) problems. Two classes of heuristics are introduced: Constraint Programming (CP)-based heuristics and decomposition heuristics. While the CP-based heuristics can be applied to any instance of the HFFTT, the decomposition heuristics are specifically designed for “rectangular” instances, where the number of machines is the same at each stage. Both approaches are compared against two iterated greedy algorithms adapted from the state-of-the-art, one of which is tailored exclusively for rectangular instances. The results show that the CP-based heuristics achieve the best performance for non-rectangular instances, while the decomposition heuristics strongly dominate all other approaches for rectangular instances, as soon as the size of the instances considered is large enough. We show that most of the results obtained can be generalized to the case without transportation times, where the HFFTT problem reduces to the HFF.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107145"},"PeriodicalIF":4.1,"publicationDate":"2025-06-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144470827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A unified robust optimization approach for problems with costly simulation-based objectives and constraints","authors":"Liang Zheng ,&nbsp;Yanzhan Chen ,&nbsp;Guangwu Liu ,&nbsp;Ji Bao","doi":"10.1016/j.cor.2025.107179","DOIUrl":"10.1016/j.cor.2025.107179","url":null,"abstract":"<div><div>This study proposes a unified robust optimization approach to address min–max problems involving expensive simulation-based objectives and constraints impacted by implementation errors and parameter perturbations. This approach optimizes the worst-case scenarios of stochastic simulation responses across multiple evaluation criteria to achieve robust efficient solutions. It integrates multiple objectives and constraints into a cohesive framework, featuring a novel performance metric designed to rigorously assess solution quality. This metric can simplify the inner constrained multi-objective maximization problem into an unconstrained, stochastic, and single-objective minimization problem, based on which a softened condition is provided to identify robust efficient solutions. Then, these neighborhood exploration and robust local move mechanisms leverage infeasible neighbors’ information to guide the iterative solution process towards a globally robust efficient point. To mitigate computational costs, surrogate models of simulation-based objectives and constraints are utilized to guide the initial exploration of worst-case neighbors. The proposed approach’s effectiveness and superior performance are demonstrated through test results on four synthetic multi-objective robust optimization problems with constraints. Furthermore, the approach is utilized to design robust traffic signal timing plans under cyber-attacks and uncertain traffic volumes, yielding satisfactory results within limited simulation budgets. This approach presents a promising tool for addressing constrained multi-objective simulation-based optimization problems under uncertainty.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107179"},"PeriodicalIF":4.1,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144272573","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bilevel competitive facility location and design under a nested logit model","authors":"Dongdong He ,&nbsp;Qingyun Tian ,&nbsp;Yun Hui Lin ,&nbsp;Yitong Yu","doi":"10.1016/j.cor.2025.107146","DOIUrl":"10.1016/j.cor.2025.107146","url":null,"abstract":"<div><div>This paper investigates a competitive facility location problem involving two companies entering a market. Both aim to locate and design new facilities to maximize revenue. The decision-making process involves one company acting as the “leader” making the initial decision, and the other as the “follower”, observing the leader’s decision before making its own. Customers choose between the two companies’ facilities based on a nested logit model (NLM). In our model, facilities are grouped into two categories to reflect similarities within each company’s facilities, resulting in a two-nest NLM. We aim to determine the optimal decision for the leader, considering the follower’s potential responses and customer preferences as dictated by the NLM. To solve this problem, we develop a nonlinear 0–1 bilevel program and propose an exact solution algorithm with two bounding problems and specialized branch-and-cut subroutines. The algorithm is guaranteed to converge to an optimal pessimistic bilevel solution in a finite number of iterations. Extensive computational experiments using a common testbed from existing literature validate our algorithm’s efficiency. Additionally, we conduct sensitivity analysis to examine the impact of NLM parameters on strategic location decisions and compare our NLM-based model with a multinomial logit model (MNL)-based model. Our findings show that using NLM over MNL can prevent potential revenue losses. Specifically, if the actual decision context is better represented by NLM but MNL is used by the leader, then the leader could face significant revenue decreases, up to 30.44%.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107146"},"PeriodicalIF":4.1,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144255084","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimal departure-time advice in road networks with stochastic disruptions","authors":"Rens Kamphuis ,&nbsp;Nikki Levering ,&nbsp;Michel Mandjes","doi":"10.1016/j.cor.2025.107148","DOIUrl":"10.1016/j.cor.2025.107148","url":null,"abstract":"<div><div>Due to recurrent (e.g. daily or weekly) patterns and non-recurrent disruptions (e.g. caused by incidents), travel times in road networks are time-dependent and inherently random. This is challenging for travelers planning a future trip, aiming to ensure on-time arrival at the destination, while also trying to limit the total travel-time budget spent. The focus of this paper lies on determining their <em>optimal departure time</em>: the latest time of departure for which a chosen on-time arrival probability can be guaranteed. To model the uncertainties in the network, a Markovian background process is used, tracking events affecting the driveable vehicle speeds on the links, thus enabling us to incorporate both recurrent and non-recurrent effects. It allows the evaluation of the travel-time distribution, given the state of this process at departure, on each single link. Then, a computationally efficient algorithm is devised that uses these individual link travel-time distributions to obtain the optimal departure time for a given path or origin–destination pair. Since the conditions in the road network, and thus the state of the background process, may change between the time of request and the advised time of departure, we consider an online version of this procedure as well, in which the traveler receives departure time updates while still at the origin. Numerical experiments exemplify a selection of properties of the optimal departure time and, moreover, quantify the performance of the presented algorithms in an existing road network – the Dutch highway network.</div></div>","PeriodicalId":10542,"journal":{"name":"Computers & Operations Research","volume":"183 ","pages":"Article 107148"},"PeriodicalIF":4.1,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144222104","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}