Epidemic resource-allocation decisions combine rapidly changing demand, incomplete reporting, and conflicting evidence. We formulate an Ambiguous Cognitive Map (ACM) in which each concept has four coupled evidence channels—true, false, partially true, and partially false. The evidence-conditioned operator maps to a capped convex state space, admits a verifiable contraction condition, and generates priorities for a constrained allocation layer. Evaluation combines 30 paired synthetic supply-chain runs, observational NHS England and U.S. Department of Health and Human Services hospital panels, retrospective GetUsPPE assignment records, and PPE-Match request, offer, and distance streams. Hyperparameters for the two allocation replays are selected by seeded tree-structured Parzen-estimator optimization on temporally separated calibration and validation periods, with the final periods held out. ACM attains NDCG@3 values of 0.948 and 0.905 on the NHS holdout and external HHS panels, respectively, while persistence-based comparators rank subsequent stress more accurately. On held-out GetUsPPE assignments, calibrated ACM reaches an average precision of 0.081 and allocation overlap of 0.008, compared with 0.177 and 0.110 for a scalar baseline. On held-out PPE-Match streams, calibration raises the prespecified operational composite from 0.132 to 0.200 and reduces unit-miles from 1999.6 to 24.8; ACM exceeds demand-only allocation (0.161) but remains below equal priority (0.204). Coupled and independent four-channel variants are operationally indistinguishable in both replays. ACM consequently provides a mathematically bounded and auditable ambiguity-aware decision interface, with performance that depends on the operational endpoint and data environment.
Post-consumer waste management has been a growing concern regarding manufacturing, especially with the expansion of the plastic industry. Decision-making in such settings often relies on a range of factors in different aspects, rather than forecasting accuracy and sustainability. Although time series forecasting and Multi-Criteria Decision Making (MCDM) in sustainable manufacturing have been widely studied, their integration is still limited, particularly when uncertainty in forecasting is considered. MCDM methods often do not incorporate forecasting results, limiting their ability to simulate realistic environments, whereas most traditional forecasting models are built under Gaussian assumptions and mainly focus on point predictions, restricting dynamic uncertainty estimation. On the other hand, recent approaches attempt to simulate uncertainty but often come with increased complexity not only in requirements but also in model comprehension, limiting their applicability. This study proposes an integrated framework combining probabilistic forecasting with MCDM to address such issues while maintaining sustainability in manufacturing. Additionally, a bootstrapping technique is applied to construct prediction intervals, discarding any specific distribution assumption, which elevates the potential of classical models. The case study results demonstrate that the proposed hybrid framework outperforms traditional and probabilistic approaches by 6% and 3% in forecasting accuracy, respectively, while simultaneously facilitating more balanced decision-making across environmental, economic, technical, and social dimensions, thereby enhancing the overall sustainability and practicality of manufacturing operations.
Capacitated Facility Location Problems under uncertainty are of great significance in real-world location scenarios. We propose a Capacitated Facility Location Problem under multiple uncertainties (CFLP-MU) that considers demand fluctuations, facility disruption risks, and capacity availability uncertainty. The model incorporates flexible capacity expansion and reactive recovery mechanisms. To solve this problem, we aim to minimize total costs while maximizing the expected Service Satisfaction Ratio (SSR). We develop a DSAA-Hybrid NSGA-II framework that outperforms baseline MOEAs (e.g., NSGA-II, MOEA/D, SPEA2) in both convergence efficiency and Pareto quality under uncertainty. This framework is driven by four innovative mechanisms: a Deterministic Warm Start (DWS), a Dynamic Sample Average Approximation (DSAA), a VIP capacity reservation policy, and a Feedback-driven Smart Repair (FSR) operator. Extensive numerical experiments reveal that capacity expansion with the tiered pricing structure reduces the need for idle backup capacity by buffering demand surges at a lower fixed cost than physical redundancy. Sensitivity analysis of medium scale demonstrates that stochastic planning yields superior value under high-penalty operational constraints, while the marginal utility of perfect information diminishes as network redundancy increases. This research offers a practical tool for balancing economic efficiency and reliability in strategic network design under uncertainty.
Rank aggregation problem is fundamental to multi-criteria decision analysis and model prioritization in logistics planning. In particular, the Kemeny framework achieves a consensus ranking by minimizing the Kemeny distance to the input rankings. Although many heuristics exist for the NP-hard Kemeny problem, these algorithms suffer from scalability issues, especially when ties are allowed. To address these limitations, we propose two complementary heuristics, BridgeSubiteration and GreedyReposition. BridgeSubiteration constructs a consensus ranking by aggregating smaller subsets of objects, while allowing ties via a Bridging mechanism. GreedyReposition makes locally optimal decisions using a proposed repositioning operator. This operator generates a set of neighbor rankings in the tied ranking space. Through real and simulated datasets, we demonstrate that the proposed algorithms achieve lower Kemeny distance in comparable computational time, making large-scale Kemeny Rank Aggregation practically feasible.
This technical note studies a deterministic single-item lot-sizing problem with fixed product lifetime, First-In-First-Out (FIFO) issuing and nondecreasing concave replenishment and inventory costs. The setting is motivated by blood product inventory management in tactical planning contexts where a relevant part of demand is predictable, such as scheduled treatments and elective procedures, and where newly collected units cannot be used immediately because of mandatory testing and processing requirements.The formulation is a fixed-expiration lot-sizing model with time-varying inventory-capacity restrictions induced by the product lifetime. We use this structure to relate the problem to the bounded-inventory lot-sizing framework of Hwang and van den Heuvel and to derive a direct dynamic programming formulation tailored to the no-backlogging, fixed-lifetime FIFO setting. The resulting algorithm has O(n2) worst-case time and space complexity, matching the complexity of the broader bounded-inventory approach while providing a specialized implementation for the blood-inventory interpretation.
Container terminal scheduling requires coordinated berth allocation and quay crane assignment decisions under uncertainty in vessel arrivals and cargo handling times. Deterministic scheduling approaches often produce fragile plans, while stochastic methods can be computationally demanding and rely on distributional assumptions. This paper proposes a Conformal Prediction-Augmented Sequential Scheduling Framework that integrates distribution-free prediction intervals into a sequential mixed-integer optimisation pipeline for uncertainty-aware scheduling. Two Extreme Gradient Boosting models predict arrival delays and handling times, wrapped with split conformal intervals calibrated at 90% coverage. These intervals are propagated through berth allocation and crane assignment models under three operational modes (optimistic, nominal, pessimistic), with a final calibrated makespan estimate. Evaluation on 241 vessel calls at Durban Container Terminal demonstrates 13–31× speedups relative to a 50-scenario stochastic heuristic, statistically significant scheduling improvements over deterministic baselines, and the result that 98% of scheduling gains are attributable to conformal uncertainty quantification rather than prediction accuracy alone. The framework additionally provides infeasibility diagnostics absent from conventional approaches, and is distribution-free and structurally non-invasive.
The Analytic Hierarchy Process (AHP) is widely used to derive priority rankings in complex multicriteria decision analysis (MCDA), yet formal operational guidance on the number of participants required for stable group-level rankings remains limited. We developed a scenario-based framework for assessing sample-size requirements for stable hierarchical AHP rankings under explicit modeling assumptions. The computational component used a stochastic simulation model in which simulated respondents were generated through between-participant heterogeneity and within-participant pairwise-judgment noise. The framework was empirically anchored using a primary healthcare AHP dataset and was further examined using supplementary datasets from transportation and social vulnerability contexts. The computational study showed that required sample size depends strongly on the separability of the underlying priority structure and on respondent variability. More steeply separated priority structures required substantially fewer participants than flatter structures. The empirical analyses were broadly consistent with these simulated stabilization patterns, but the resulting stabilization points are conditional on the adopted simulation assumptions, including the perturbation model, Saaty-scale discretization, aggregation rule, and the characteristics of the empirical panels. AHP sample-size requirements are not absolute, and this study does not provide a universal numerical rule. Instead, it provides a scenario-based assessment framework. Sample-size planning for hierarchical AHP should be tailored to the anticipated mathematical difficulty of the decision problem, the expected variability of the respondent panel, and the assumptions of the chosen aggregation and simulation model.
Highly integrated container terminals are extremely sensitive to fluctuations in vessel arrival times. This systemic vulnerability is further intensified by the extensive adoption of onshore power supply (OPS). To mitigate these issues, this study first develops a nonlinear mixed-integer programming model incorporating OPS capacity constraints. To quantify practical navigation risks, survival analysis is used to characterize vessel arrival uncertainty. Based on this, a chance-constraint programming (CCP) model is constructed and subsequently linearized via a piecewise approach to approximation. Furthermore, a tailored column generation algorithm is designed for efficient global optimization. Numerical results demonstrate that the proposed CCP model accurately captures uncertainties and yields superior objective function values compared to conventional chance-constraint methods.
Scheduling problems are typically studied under the assumption that each resource can process only one job at a time. However, in many real-world settings, resources are capable of processing multiple jobs simultaneously, leading to cumulative scheduling problems, where each resource can process several jobs up to a specified capacity. While cumulative scheduling has been investigated in job shops and parallel-machine contexts, it remains unexplored in flow shop environments. In this work, we introduce the cumulative flow shop scheduling problem as a generalization of the classical flow shop. We propose a mixed-integer programming formulation along with two metaheuristic approaches – an iterated local search algorithm and an iterated greedy algorithm – designed to minimize total flow time. Both methods incorporate problem-specific features, such as an adapted initial solution and a rescheduling mechanism to improve resource utilization. With this, they demonstrate strong performance in a comprehensive computational study. Additionally, we analyze trade-offs between total flow time and resource utilization, and investigate how varying job sizes and resource capacities affect outcomes. Our findings provide actionable managerial insights and open new avenues for research on cumulative scheduling problems.
The superposition of arrival processes is a fundamental yet analytically intractable operation in queueing networks when inputs are general non-renewal streams. Classical methods either reduce merged flows to renewal surrogates, rely on computationally prohibitive Markovian representations, or focus solely on mean-value performance measures.We propose a scalable data-driven superposition operator that maps low-order moments and autocorrelation descriptors of multiple arrival streams to those of their merged process. The operator is a deep learning model trained on synthetically generated Markovian Arrival Processes (MAPs), for which exact superposition is available, and learns a compact representation that accurately reconstructs the first five moments and short-range dependence structure of the aggregate stream. Extensive computational experiments demonstrate uniformly low prediction errors across heterogeneous variability and correlation regimes, substantially outperforming classical renewal-based approximations.When integrated with learning-based modules for departure-process and steady-state analysis, the proposed operator enables decomposition-based evaluation of feed-forward queueing networks with merging flows. The framework provides a scalable alternative to traditional analytical approaches while preserving higher-order variability and dependence information required for accurate distributional performance analysis.Experiments indicate that the neural framework achieves errors on the order of 1%–5% for steady-state predictions across the tested networks, while classical methods often exhibit much larger errors, typically tens of percent and in some regimes exceeding 100%. This gap persists across varying SCV and utilization levels, with the learning-based approach remaining stable even under high variability and strong dependence, where traditional approximations significantly deteriorate.
The Capacitated Family Traveling Salesman Problem (CFTSP) models critical operational decisions in modern warehouse management, particularly in large-scale distribution centers where items are organized into families. This problem requires selecting specific items from each family and assigning them to capacitated pick-up agents while minimizing total travel distance. We propose two novel solution approaches: an exact method based on Constraint Programming (CP) and a matheuristic that hybridizes a bin-packing formulation, which acts as a surrogate model for determining the number of family items assigned to each agent, with a Reactive Greedy Randomized Adaptive Search Procedure incorporating probabilistic stopping criteria. Computational experiments on benchmark instances demonstrate that our CP approach outperforms existing mixed-integer programming formulations, solving 93% of instances compared to 51% for the state-of-the-art method. Our matheuristic achieves superior solution quality, obtaining the best-known solutions for 91% of test instances while maintaining reasonable computational times. These results establish new benchmarks for the CFTSP and provide practical tools for warehouse optimization.
Reconfigurable manufacturing systems with auxiliary modules enhance flexibility but complicate resource allocation. Furthermore, traditional scheduling often ignores how post-processing aging treatments affect final product performance. This paper introduces a novel performance-aware flexible job shop scheduling problem with aging-time windows and machine reconfiguration (PFJSP-AW-MR) to simultaneously minimize total weighted tardiness and total performance deviation. First, a mixed-integer linear programming (MILP) model employing piecewise linear fitting is formulated to capture the non-linear temporal evolution of material properties. To solve this strongly NP-hard problem, an improved NSGA-II with a hybrid neighborhood search (INSGA-II_NS) is proposed. A problem-specific three-layer encoding resolves the spatio-temporal coupling of discrete configurations and continuous aging times. Moreover, a customized two-stage neighborhood search, integrating an improved K-insertion method on an extended disjunctive graph, significantly enhances local exploitation. By incorporating aging arcs into the disjunctive graph, the improved K-insertion method enables more accurate critical-path identification and targeted adjustment of operations related to both tardiness and performance deviation. Extensive benchmark experiments and a real-world industrial case demonstrate that INSGA-II_NS outperforms well-known multi-objective algorithms, effectively balancing manufacturing performance and production efficiency.
The growing diversity of higher education systems worldwide highlights the need for methodologies that classify universities based on their traditional missions, enabling a better understanding of their socio-economic impact. Given their specialization in teaching and research, the identification of homogeneous groups of universities seems vital for more effective benchmarking and policy planning. This paper applies a novel multi-criteria approach to classify universities according to their performance in these core missions. The multiple reference point composite indicator technique is employed to assess overall performance, constructing composite indicators for each mission and period considered. Empirical results demonstrate the importance of establishing comparable groups for university leaders when comparing performance effectively, which promotes informed strategic decision-making and long-term monitoring. As a proof of concept, the proposed methodology is applied to the Spanish public universities, using twenty performance indicators classified into six dimensions and two main missions. The results show that grouping by period gives rise to three homogeneous groups that remain stable over the time horizon, with some improvements among universities with moderate performance. On the other hand, grouping by mission shows that Spanish public universities perform similarly in teaching, forming only two groups. In contrast, the mission of research and industrial income shows greater diversity, with three clearly differentiated groups. Finally, this paper provides useful information for university leaders, administrators and policymakers by suggesting new methodologies to promote and foster the socio-economic development of Spanish public universities in their respective territories.
The objective of this work is the implementation and improvement of classical Geometric Programming methods through Duality Theory. The proposed approach is based on a branch-and-bound algorithm, which provides efficient lower bounds but does not adequately exploit upper bounds. As the main contribution, an extension of this algorithm is proposed through the incorporation of an upper bound obtained via the condensation technique, integrated as a local search strategy within the branch-and-bound framework, allowing the refinement of upper bounds, reduction of the search space, and acceleration of convergence. The choice of the initial point influences the partitioning of bounds in the search tree and may impact algorithm performance, being a relevant criterion in global optimization. To address this issue, a method for generating initial points based on the optimal solution and problem bounds is incorporated. The method constructs a sphere centered at the optimal solution, covering approximately (90%) of the search space, whose interior is explored through random sampling for the selection of initial points. The improved algorithm is validated through numerical experiments on classical benchmark problems, in which optimal solutions are obtained in nearly all tested instances, demonstrating the robustness, efficiency, and potential of the proposed approach.
Machine breakdowns directly affect scheduling performance and production costs in manufacturing systems. Instead of focusing on reactive measures after failures, managing the factors that cause failures beforehand can help reduce them. To address these issues, this paper proposes a flexible job shop model that integrates machine age, workload, and processing speed directly into scheduling decisions, rather than relying on predefined failure thresholds. By linking these operational factors to breakdown occurrence and capacity loss, the model minimizes makespan while maximizing machine lifetime. To deal with the bi-objective function, we propose a novel approach that integrates a Memetic Algorithm (MA) with an augmented ɛ-constraint method. In this approach, the ɛ-constraint component guides the multi-objective search by prioritizing objectives and ensuring feasible exploration, while the Memetic Algorithm performs global search and local refinement to enhance solution quality. Benchmark instances from Brandimarte were used to evaluate the performance of the proposed method. It was compared with the Non-dominated Sorting Genetic Algorithm II (NSGA-II) and classical Memetic Algorithms. In comparison with NSGA-II, the proposed method consistently achieved 6%–10% higher Hypervolume (HV) and 45%–48% lower Inverted Generational Distance (IGD). In comparison with the classical MA, it achieved 1%–6% higher HV, 27%–31% lower IGD, and generated 30%–50% more non-dominated solutions.
Search and Rescue (SAR) path planning is critical in post-disaster scenarios with road damage and noisy demand information. This paper proposes a dynamic post-disaster SAR path planning model that comprehensively considers road damage severity, casualty uncertainty, and life-saving priority. A three-stage decision-making framework is developed: first, an improved K-means clustering algorithm based on silhouette analysis determines the optimal number of rescue clusters and establishes an unstructured post-disaster environment model. Second, SAR path pre-planning is implemented: a hybrid heuristic integrating priority greedy hierarchical allocation, priority-aware nearest neighbor heuristic and constraint-preserving 2-opt algorithm generates high-quality initial solutions, and a motified dynamic adaptive large neighborhood search (M-D-ALNS) algorithm constructs pre-planned SAR routes with minimal priority violations and shortest completion time. Third, a bidirectional event-driven response mechanism based on local and global (BEDRM-LG) is proposed to mitigate discrepancies between pre-planned results and actual rescue conditions. Experimental validation on modified Solomon benchmark instances shows the proposed initialization strategy generates optimal initial solutions fastest; M-D-ALNS reduces violation rate to near zero with over 90% optimization efficiency; BEDRM-LG outperforms common benchmarks and even reaches the theoretical upper bound, demonstrating superior performance over static planning and continuous full-information re-optimization in post-disaster SAR tasks.
In this paper, we propose an exact method for solving the problem of scheduling multiprocessor tasks on two dedicated processors with release dates. In this problem, we have three subsets of tasks, where these tasks need one or two dedicated processors to be performed. The aim is to minimize the total completion time, where each task has a release date. In computational complexity theory, this problem is shown to be NP-hard in the strong sense. The contribution of our work is to construct a mathematical model, to give an improved lower bound and to propose a branch-and-bound algorithm to solve the studied problem.
In many real group decision-making problems, decisions must be made with limited data and busy Decision Makers (DMs). In such settings, asking for detailed scores or trade-off weights can be unrealistic and can obscure, rather than clarify, how a recommendation emerges. We propose Simple Group Choice (SGC), a deliberately lightweight approach to multi-criteria group choice that relies only on two sets of binary questions. SGC then aggregates by simple counting: wins on a criterion are weighted by how many people consider that criterion key and are summed across the group. The result is a transparent support score for each alternative and a top choice set, accompanied by compact diagnostics that help interpret and screen. Because incomplete answers are common, we extend SGC to uncertainty by treating missing entries as unknown binary bits and examining all feasible completions. This yields clear, robust recommendations without introducing parametric assumptions. A didactic urban-logistics study illustrates the approach under complete information and under uncertainty. Overall, SGC offers a practical, explainable, and low-burden tool for contexts like shortlisting in early-phase group decisions, and a natural front end to more detailed analyses when needed.
Identifying the optimal locations of fire stations is a critical problem in urban planning and emergency management, which directly affects the efficiency of emergency services. However, fire station location decisions are often made under incomplete information environments. This incompleteness is primarily manifested in two aspects: one is human cognitive uncertainty due to a lack of historical data (referred to as human uncertainty), and the other is inherently ambiguous probability distributions of randomness itself. Therefore, during the location process, factors such as travel time, operating cost, and fixed cost often exhibit certain degrees of randomness with ambiguous probability distributions or human uncertainty. To address these challenges, this paper presents a multi-objective chance constrained programming model that simultaneously considers four factors: distance, travel time, annual operating cost, and fixed cost. It also introduces sublinear expectation theory and uncertainty theory to handle randomness with ambiguous probability distributions and human uncertainty, respectively. This paper offers significant theoretical insights and practical guidance for urban fire safety planning and emergency management under an incomplete information environment. One example is that fire station location plans in Changqing District of Jinan City are derived by applying a genetic algorithm (GA) to solve the model.
When unexpected temporary reductions in airspace capacity occur, flights may need to be rescheduled, often resulting in delays and associated costs. Reallocating limited airspace resources in a way that equitably mitigates delay costs across airlines, while directly involving them in the decision-making process, remains a significant operational challenge. Although several approaches have been proposed in the literature, the complexity of tactical air traffic flow management and the diverse requirements of stakeholders have so far limited their practical applicability. In this work, we introduce a novel inter-airline slot trading framework that provides a fair mechanism aligned with airlines’ operational needs, while remaining sufficiently simple for real-world deployment. The framework is based on an integer programming formulation, and we show that the resulting trade selection problem can be expressed as a Maximum Weight Independent Set (MWIS) and solved efficiently. To cope with the growing computational burden associated with large-scale instances, we develop a three-step machine-learning-based pre-processing heuristic. This heuristic uses neural networks to discard flight combinations unlikely to generate beneficial trades and to estimate the value of the remaining ones, thereby restricting the search space to the most promising candidates. Computational experiments demonstrate that the proposed framework effectively reduces total delay costs while ensuring an equitable distribution of benefits among airlines. Moreover, the neural-network-based pre-processing component enables the efficient solution of instances that would otherwise be computationally intractable, substantially enhancing the practical applicability of the approach.
