Given that the emerging shared mobile precooling device (SMPD) is applied to post-harvest precooling of fresh agricultural products, this study addresses a novel SMPD scheduling problem in which a rental enterprise with limited SMPDs selectively provides precooling services to meet spatiotemporally distributed time-varying demands. We develop a mixed-integer linear programming model that is decomposed using the Dantzig–Wolfe decomposition method. Then, a branch-and-price algorithm is designed to solve the model, in which the pricing subproblems are solved using exact and heuristic label-setting algorithms. Numerical experiments suggest that our algorithm can efficiently obtain optimal solutions for instances of varying sizes. A case study demonstrates the applicability of our model and algorithm, and considering the time-varying characteristics of customers’ precooling needs can significantly improve the enterprise’s profit and better meet customer demands. Sensitivity analysis reveals that increasing the proportion of important customers will decrease the enterprise’s profit and fulfillment rate. While prioritizing precooling needs of important customers can improve the service rate for this segment, it presents a potential tradeoff against overall profitability. The greater the difference in time-varying precooling needs, the lower the enterprise’s profit.

The limitations of traditional indifference-zone (IZ) ranking and selection procedures stem from their reliance on an IZ parameter, which is often unknown in practice and can lead to inefficiencies. Additionally, since exact evaluation of the probability of correct selection (PCS) is generally unattainable, these approaches commonly rely on a Bonferroni-Inequality-type lower bound to ensure the worst-case PCS, which introduces further conservativeness. To overcome these challenges, we propose fully sequential IZ-free procedures that eliminate the need for an IZ parameter and reduce the conservativeness associated with Bonferroni Inequality approaches. We demonstrate that, in the context of IZ-free procedures, the screening statistic corresponds to the Euclidean distance of a multi-dimensional standard Brownian motion, which behaves as a Bessel process. This connection simplifies the derivation of statistical guarantees for the proposed procedures and facilitates the analytical expressions for key parameters. Our procedures accommodate both known equal variances and unknown unequal variances, with statistical guarantees established in the known equal variance case. Numerical experiments confirm that the proposed procedures outperform existing methods in reducing PCS conservativeness and improving sample size efficiency. These advancements enhance the robustness and practicality of IZ-free procedures for diverse applications.

Manufacturing systems often face unexpected disruptions such as machine failures or material shortages, which can severely impact the production performance. Traditional methods for addressing these disruptions tend to be time-consuming and resource-intensive, which cannot effectively maintain the resilience of manufacturing systems. Despite recent advances in digital twins (DTs) and artificial intelligence (AI), very little has been done to mitigate high computational demands and generate resilient designs of manufacturing system under uncertainty. Therefore, this paper presents a new Generative Adversarial Network (GAN) approach for the on-the-fly design of manufacturing systems in response to production disruptions. First, we propose a novel Generative Adversarial Network (D-GAN) to generate diverse, adaptive system designs that align production performance with target key performance indicators (KPIs). Second, DT models are coupled with statistical metamodeling to sequentially generate large amounts of training data samples under different scenarios. Experimental results show the high potential of the proposed D-GAN approaches to generate cost-effective system designs and enhance manufacturing resilience.

Large-scale multivariate profile data have been ubiquitous in many real-world applications with rapid advances in sensing technologies, the joint modeling of which, however, remains notoriously challenging due to high data dimension and complicated between-profile correlation. When a large number of profiles exhibit distinct patterns and turn to be heterogeneous, their correlation becomes more sophisticated—one profile could have its personalized patterns and meanwhile interact with other profiles via some common patterns, and these common patterns could link all profiles together or only a subset of them selectively. In light of this, we propose a novel partially linked multi-matrix decomposition (PLM2D) approach for multivariate heterogeneous profile data. Specifically, to adaptively characterize the intricate interplay between different profiles, we decompose all profiles jointly into three kinds of components: (1) globally common components that are shared by all profiles, (2) partially common components shared by some but not all profiles, and (3) individual components attached only to one specific profile. Then, a two-stage iterative learning strategy is designed. It includes a structure learning stage with a permutation-enabled hierarchical clustering to quickly determine the between-profile common components, and a parameter learning stage where an alternating direction method of multipliers algorithm is developed to efficiently learn model parameters in an aggregating and distributing fashion. Real case studies on passenger flow profile data from many metro stations and extensive numerical simulations validate the superiority of our PLM2D to its benchmarks in multivariate profile modeling and monitoring.

In response to furious competition from emerging online retailers, traditional supermarkets begin leveraging their brick-and-mortar store chains by setting up small in-store front warehouses to facilitate order picking processes for online orders. In this article, we investigate how to select products and allocate space for the in-store front warehouse under dynamic and static strategies to minimize the total order-picking workload. For the dynamic allocation problem, we formulate it as a mixed-integer programming (MIP) model, which can be decomposed into a master multichoice knapsack problem (MCKP) and an order picking problem (OPP) as subproblems. An exact algorithm and a heuristic that is 1/2-bounded under certain routing policies are proposed. For the static allocation problem, we construct a stochastic gradient estimator of the expected travel distance by infinitesimal perturbation analysis (IPA) and introduce a discrete Frank-Wolfe algorithm to solve it. Numerical results reveal that the logic of dynamic allocation for the in-store front warehouse is different from that for regular warehouses, and the storage location assignment for the shelf area in supermarkets greatly impact the performance of these allocation strategies. The experimental results using data from a real retailer validate the effectiveness of proposed algorithms.

The paper investigates the optimal purchasing and production control policy within a circular production system that processes two returned material types through a shared manufacturing center. Materials are purchased or rejected upon arrival, with differences in procurement cost. At each decision epoch, the system must decide whether to purchase incoming materials and which type to prioritize for production. The problem is formulated as a Markov decision process, and a linear programming approach is developed to compute the optimal purchasing and production policy for the system under both non-preemption and preemption rules. Theoretically, the structure of the optimal policy is analytically characterized under the preemption rule in the case of equal production rates. Numerical experiments reveal that this policy remains optimal or near-optimal for the system under non-preemption and preemption rules with varying production rates. Additionally, comparisons of the optimal policies between non-preemption and preemption rules demonstrate that the system under the preemption rule yields higher average reward than that under the non-preemption rule, to varying degrees depending on system parameters, such as production rate, holding cost, and arrival rate.

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. This article studies how to solve the large-scale guaranteed service model (GSM) to handle this problem. It presents a novel approach, called the iterative decomposition (ID) approach, to efficiently solve the large-scale GSM. The ID approach decomposes the problem iteratively by analyzing local solutions obtained from a non-convex optimization method (sequential linear programming) and selecting the appropriate algorithm to solve each subproblem based on its specific network structure. Our approach is evaluated against several methods from the literature using generated inventory networks. Our numerical experiments show that the ID approach performs better when the network size is large and the network structure is complex. We provide a Python library called InvNet, which provides the implementation for all methods considered in this article.

Online customer reviews provide valuable information about product quality, and recently some control chart-based schemes have been proposed to detect product performance anomalies from reviews. As a review outcome depends not only on inherent product quality but also on customer’s rating bias and latent preference, ignoring customer’s latent factors may lead to misjudgements about online product performance. Therefore, this study proposes a risk-adjusted control chart for monitoring the decrease of rating scores in reviews, by separating the personal risk factors of individual customers from the assignable causes with respect to online product performance. The proposed risk-adjustment model is fitted by a united latent factor model that learns user rating bias and preference factors by combining both review texts and corresponding ratings. According to the experimental results of a real-world case study on beer review corpus and extensive simulation studies, the proposed method shows superior performance in review shift detection, also with good diagnostic performance for explaining the reasons behind anomalies.

The Job Shop Scheduling Problem (JSP) is a classical combinatorial optimization problem that involves scheduling a set of jobs on a set of machines to minimize the makespan. Because of its importance, the JSP has been extensively studied in the fields of operations research and manufacturing. This study presents a highly effective hybrid approach that integrates tabu search with frequent pattern mining within a population-based search framework for addressing the JSP. Of particular interest of the tabu search is its novel tabu strategy, which employs hash techniques to accurately mark visited solutions and rapidly determine the tabu status of neighboring solutions in the local search. Such a tabu strategy enables a more efficient exploration of the solution space. Furthermore, the hybrid approach fully embraces the potential of frequent pattern mining by incorporating a frequent job order recognition method to identify promising solution structures. These structures are not only used to generate new promising solutions but also to guide the local search process. Extensive experiments on the benchmark instances widely used in the literature show that the hybrid algorithm performs remarkably well, by improving 13 best-known results (new upper bounds). Additional experiments are presented to gain insight into the role of the key elements of the algorithm.

This study investigates the inventory routing problem (IRP) under ambiguous demand distribution. Compared with existing research that assumes the demand distribution is either predefined or lies in an ambiguity set, we adopt a globalized distributionally robust approach. We construct both inner and outer ambiguity sets to describe distributional ambiguity, regardless of whether retailer demand follows a discrete distribution or an unspecified type. This approach ensures performance is maintained even when the demand distribution extends beyond the inner ambiguity set. Additionally, we introduce more flexible soft constraints, coupled with a budget violation metric, to manage the gaps between actual holding and penalty costs and the budgets at retailers, particularly when demand distribution lies outside the inner ambiguity set. Analysis of our model demonstrates its ability to generalize distributionally robust, robust, or stochastic IRP models through appropriate input settings, and thereby we establish the bounds for the objective values of these models. Numerical studies, based on a real-case scenario, show that the delivery solutions from our proposed models are more effective in mitigating the impact of retailer demand uncertainty and achieve superior out-of-sample performance compared to other benchmark models.