This study examines maintenance capacity adjustment strategies under split acquisition, a common practice in defense procurement driven by budgetary and production constraints. Using the Republic of Korea Air Force (ROKAF) F-35A as a case, the study investigates how time gaps between acquisition phases, combined with bathtub-shaped failure patterns, generate non-linear shocks in maintenance workload that threaten operational availability. A System Dynamics (SD) model is developed to simulate a 40:20 split acquisition scenario and to compare three capacity adjustment strategies: Level (fixed), Chase (reactive), and Lead (proactive). The strategies are evaluated under review intervals of 3, 5, and 10 years to reflect realistic defense planning environments. The simulation results show that the reactive Chase strategy consistently fails to achieve the availability target of 75 % due to workforce training delays. In contrast, the proactive Lead strategy successfully defends availability by anticipating future workload changes. A 3-year review interval achieves the highest cost-effectiveness, while a 5-year interval, aligned with the Mid-term Defense Plan cycle, is identified as a realistic threshold for sustaining availability. However, under a 10-year review interval, even proactive strategies lose effectiveness. The findings highlight the importance of proactive, forecast-based decision-making and suggest that synchronizing maintenance review cycles with acquisition schedules within a 5-year horizon is essential for robust maintenance policy design.
The rapid growth of artificial intelligence (AI) has shifted the development of wearable health monitoring devices toward intelligent design. However, conventional approaches still rely heavily on experience, involve inefficient user requirement acquisition, and lack systematic support for innovation. To address these limitations, this study proposes a data-driven framework for intelligent product development that integrates deep learning, Kansei Engineering, and TRIZ theory. The framework establishes an AI-assisted work-flow comprising online data acquisition, requirement identification, knowledge mapping, concept generation, and engineering optimization. First, web crawlers collect user reviews and product images. A convolutional neural network (CNN) identifies form features, while a generative adversarial network (GAN) generates diverse design concepts. Kansei Engineering then maps user requirements to design elements, and TRIZ theory resolves engineering contradictions to support systematic optimization. A case study of smart health watches, using a dataset of nearly 30,000 images and more than 12,000 reviews, demonstrates that the framework can effectively automate requirement extraction, concept generation, and design optimization. Compared with traditional experience-driven methods, the proposed framework significantly improves decision-making efficiency and supports product innovation. This study extends the application of deep learning to engineering design support and provides a transferable methodology for data-driven intelligent product development.
Raw material ordering under demand uncertainty is often hindered by inaccurate forecasts, resulting in excess inventory, stockouts, and increased operating costs. Traditional Newsvendor models typically assume static demand distributions and therefore have limited ability to capture dynamic changes in demand. To address this limitation, this study proposes a Markov chain–based Newsvendor decision framework incorporating an [L, U] inventory boundary policy. The main contribution of the proposed framework is the integration of state-transition-based probabilistic demand forecasting with inventory optimization. Specifically, raw material demand is classified into a finite set of states, and the transition probability matrix is estimated from historical demand-state observations. The resulting predicted demand distribution is then incorporated into the Newsvendor model to determine the optimal order-up-to level and the corresponding replenishment quantity. A simulation-based case study of an electronics manufacturing firm is conducted using a generated 52-week demand dataset. The results show that the proposed MC-NM model reduces the expected total ordering cost from 4887.68 yuan to 4269.92 yuan, corresponding to a cost reduction of 12.6 %. These findings indicate that the proposed framework can reduce inventory-related costs while maintaining responsiveness to demand fluctuations, thereby providing practical decision support for raw material procurement under uncertainty.
The mixing stage of automotive battery production requires reliable monitoring of raw material types, personnel actions, and correct tool use. However, accurate multi-scale object detection in complex scenes remains challenging because of background interference and the requirements of embedded deployment and real-time operation. This study proposes an enhanced YOLOv5 framework for industrial multi-object detection. The method adopts a dual-stage feature-enhancement strategy designed to improve robustness while limiting parameter count and computational complexity. First, the Convolutional Block Attention Module (CBAM) is embedded in the Backbone and Neck of YOLOv5 to provide multi-granularity feature enhancement and improve the detection of small objects, such as tools. Second, the conventional CIOU loss is replaced with the Focal-EIOU loss function to optimize bounding-box regression, reduce false detections across multiple target scales, and accelerate model convergence. Finally, K-means clustering is applied to target geometric features to generate specialized anchor-box parameters better suited to industrial scenarios. Experimental results on an automotive battery production-site dataset show that the improved model's mAP@0.5 increased by 2.4 % compared with the original model, reaching 95.2 %, while mAP@0.5:0.95 improved by 3.6 %, reaching 78.5 %. The proposed framework provides a lightweight and reliable solution for multi-object detection in complex industrial environments and supports the development of intelligent visual monitoring systems for smart manufacturing.
Equipment reliability is critical to maintaining production efficiency and controlling manufacturing costs. Predictive maintenance (PdM), based on machine condition prediction, can effectively reduce the risk of machine failure; consequently, machine degradation assessment and the prediction of remaining maintenance life (RML) are crucial for maintenance decision-making. Moreover, because equipment condition affects production planning, PdM should be integrated into the traditional economic production quantity (EPQ) model. The main contribution of this study is the introduction of RML as a decision-support metric linking equipment degradation prediction with production lot-sizing decisions, thereby enabling the joint optimization of EPQ and PdM policies. To minimize expected average cost, maintenance decisions are integrated into the production lot-sizing model to determine the optimal production lot size and maintenance policy. This study considers a single-machine production process in which the ARMA method is used to forecast the machine degradation index. Cox’s proportional hazard model (PHM) is then employed to estimate machine reliability based on the predicted degradation index. Based on this reliability assessment, remaining maintenance life (RML), rather than the traditional remaining useful life (RUL), is employed to link degradation prediction with the EPQ model and to characterize the machine deterioration process. An integrated EPQ–PdM model is developed to jointly determine the optimal production lot size and maintenance policy while minimizing the expected average cost (EAC) over the production cycle. Finally, a case study of an automotive bumper factory demonstrates the effectiveness of the proposed framework. The results show that the framework reduces EAC by 19.6 % relative to the current production strategy and identifies an optimal maintenance threshold of Rsafe = 0.4 with six maintenance cycles.
This study presents a data-to-decision workflow that combines surrogate modeling, multi-objective optimization, and decision-making in a single procedure. Four surrogate models, including Kolmogorov-Arnold Networks (KAN), Cat-Boost (CAT), Gradient Boosting Regressor (GBR), and LightGBM (LGB), were trained under a unified preprocessing and Bayesian tuning scheme and evaluated on held-out data. The retained models were then embedded in NSGA-III to generate the Pareto front for the trade-off between surface roughness (Ra) and material removal rate (MRR). To move from the Pareto set to a single operating condition, the candidate solutions were further assessed using multiple MCDA methods under different objective weighting schemes, and the resulting rankings were combined through rank aggregation (Borda, Copeland, Kemeny-Young, Robust Rank Aggregation). A turning case study on C3604 free-cutting brass, using cutting speed, feed, depth of cut, nose radius, and coolant condition as inputs, showed that the final recommendation remained stable across different weighting and ranking settings. Experimental verification at selected Pareto points agreed well with the predicted values, with relative errors of about 5 % for both Ra and MRR. The results show that the proposed workflow provides a practical and consistent route from limited machining data to a final operating decision.
Intelligent siting of electric vehicle (EV) fast charging stations is of great significance for the development of urban transportation. This study comprehensively considers factors related to the placement of EV fast charging stations, including differences in travel behavior between fuel vehicle travelers and EV travelers, road congestion, vehicle energy consumption, and the geographical location of charging stations. Based on these factors, an advanced bilevel programming model integrating dynamic traffic assignment (DUE) and charging station location selection is developed. Specifically, the upper level uses a genetic algorithm (GA) to solve a system optimization model aimed at minimizing total travel time while accounting for the construction and operation costs of charging stations. Corresponding to the upper-level decisions, the lower level captures the joint selection behaviors of fuel and electric vehicles and is solved by a DUE procedure combined with the method of successive averages (MSA). This lower-level model incorporates travelers’ on-route fast-charging behaviors, including departure time choice, charging facility preference, and route selection. The integrated approach provides a sophisticated framework for analyzing and optimizing the interaction between charging station placement, travel time, and associated costs. An iterative dynamic traffic flow algorithm integrating EV charging queue simulation is proposed. Finally, numerical studies conducted on an illustrative network derive optimal location schemes at different EV adoption stages and analyze the complex operational characteristics of the traffic network under the coexistence of EVs and fuel vehicles.
This study investigates the collaborative optimization of routing and charging in a distribution network comprising electric trucks (ETs) and autonomous unmanned vehicles (AUVs), supported by mobile photovoltaic (PV) storage charging. A comprehensive optimization model is developed to minimize the total daily operating cost of the logistics enterprise, encompassing vehicle acquisition, staffing, energy consumption, charging, and penalties. The model simultaneously determines coordinated ET-AUV delivery routes, mobile charging schedules, and parking node locations. The PV-storage system is incorporated as a green power-supply constraint, directly influencing charging costs. To address the model's complexity, an enhanced hybrid frog-leaping algorithm is proposed. This algorithm incorporates an initial solution construction method to improve population quality, an advanced local deep search to increase search efficiency, and diversity control strategies with clone selection programs to maintain population diversity. The effectiveness of the developed algorithm is validated through multiple case studies with varying customer sizes. Computational experiments on instances with up to 60 customers indicate that, compared with the ET-only mode, the proposed ET-AUV collaborative mode reduces total daily operating costs by 18.61 %, decreases staffing costs by 62.5 %, and lowers penalty costs by 23.21 %, thereby enhancing customer satisfaction and operational resilience. Sensitivity analysis shows that system efficiency depends on several operational parameters. Increases in ET payload and range are the main drivers of cost reduction. Additionally, the average speeds of both vehicle types have a critical U-shaped effect on total costs.
In metropolitan instant delivery systems, rising operational costs and poor delivery timeliness pose significant challenges. This study addresses these issues by investigating optimal order allocation and route optimization in a hybrid delivery system that integrates autonomous delivery vehicles (ADVs) with human riders. Analysis of historical order data indicates that inefficient order assignment between ADVs and riders is a major operational bottleneck. To address this problem, a collaborative human-ADV delivery model is formulated with the objective of minimizing total logistics costs under constraints related to delivery time windows, vehicle capacity, and routing requirements. The proposed model is applied to a real-world case involving QX Fresh Supermarket, comprising 80 orders and 16 community transfer points. An improved genetic algorithm is developed to solve the optimization problem efficiently. Empirical results for off-peak, normal, and peak periods show that the collaborative delivery approach significantly improves ADV utilization and reduces total delivery costs by more than 30 % without compromising timeliness. These findings provide both a sound theoretical basis and practical guidance for the advancement of human-machine collaborative logistics.
Reliable defect evaluation is essential in ceramic sanitaryware manufacturing, where inspection outcomes directly influence rework decisions, process control, and delivery performance. In practice, defect assessment is often affected by operational variability, class imbalance, and human-dependent inspection procedures, which limit the repeatability and consistency of quality control decisions. This study formulates multi-class defect classification as a practical quality control problem and investigates the robustness of production-data-based decision-support in an industrial environment. The analysis is based on a real-world dataset comprising 11,071 production records collected under routine operating conditions. Defect labels were assigned through a two-stage quality control procedure involving trained inspectors and supervisory verification. Multinomial logistic regression, support vector machines with radial basis function kernels, and CatBoost were evaluated as base classifiers. A probability-based voting ensemble was developed to integrate the complementary decision structures, and posterior probabilities were calibrated using Platt scaling prior to aggregation to improve decision consistency. Experimental results show that the proposed calibrated ensemble improves Macro-F1 from 0.7126 (logistic regression) to 0.7211 and enhances minority-class recall, leading to more balanced performance under severe class imbalance. The findings indicate that probability calibration and ensemble integration contribute to improved stability and interpretability of defect evaluation. Overall, the proposed framework provides a practically deployable decision-support layer that supports more consistent rework decisions and more reliable quality control in industrial production settings.
This paper presents the development and experimental validation of an adaptive pneumatic gripper for collaborative robotic palletizing of packages with varying mass and surface characteristics. The main objective is to determine the optimal gripping-force and minimum operating pressure required to ensure stable and safe handling without slippage. A dynamic mathematical model was developed, incorporating the effects of package mass, friction coefficient, contact surface area, and inertial forces during manipulation. Numerical analysis was performed for different friction conditions (& micro; = 0.30-0.90) and contact configurations, enabling the determination of the minimum required gripping-forces and corresponding operating pressures. Experimental validation was conducted on a real industrial system with a collaborative robot. The results show a linear relationship between pressure and gripping-force, described by F = 22.152 p-17.535, with a high correlation coefficient (R2 approximate to 0.998). The maximum experimentally obtained gripping-force was approximately 70-75 N at a pressure of around 4 bar. Quantitative deviations between numerical and experimental results (65-75 %) were observed and corrected by introducing a calibration factor (kcorr approximate to 0.30). The proposed model and experimental system enable reliable optimization of gripping-force and improve manipulation stability under real industrial conditions. The main contribution of this study lies in the integration of analytical modelling, numerical optimization, and industrial experimental validation for collaborative robotic palletizing systems
In modern industrial production, electronic component manufacturing imposes increasingly stringent requirements on quality inspection. To address the susceptibility of traditional detection methods to noise interference, the difficulty of localizing abnormal regions in images, and insufficient feature extraction capability, an improved industrial image anomaly detection method based on the original AGUR-Net (Attention-Guided Unsupervised Representation Network) model is proposed. This method enhances the recognition of abnormal regions by introducing attention gate modules and preactivated fusion residual blocks, while simultaneously combining stacked sparse denoising autoencoders to enhance the model's robustness to noise and suppress redundant information. The experimental results show that the proposed method achieves an accuracy of 84.26 % in locating abnormal regions on the MVTecAD dataset after 185 iterations, which is higher than the dual attention generative adversarial network (79.21 %), long short-term memory non-destructive testing network (79.31 %), and bidirectional long short-term memory network (81.63 %). At 400 iterations, the average loss value of this method for detecting abnormal images of industrial products is 0.016. In addition, the processing time for single images of Class A, B, and C defects in the actual factory environment is 372 ms, 329 ms, and 378 ms, respectively, demonstrating high detection efficiency. Overall, this method can accurately identify and locate abnormal images in the surface treatment and quality inspection stages of electronic component production, which helps to improve the quality control level in the manufacturing process and provides effective technical support for intelligent industrial production.
This paper investigates optimal decision-making and coordination in a material supply chain under multidimensional stochastic risks, including uncertain demand, yield, and processing activities. By analyzing a system consisting of a manufacturer and a supplier, we derive the unique optimal purchasing and stocking decisions that maximize the expected profit of the integrated supply chain. A wholesale price contract is proposed to coordinate decentralized participants, incorporating risk allocation through a linear combination of production cost risk and sales profit risk. The key findings indicate that elevated risk dimensions lead to an increase in optimal order and inventory quantities. However, they also result in a reduction in system profit and a narrowing of the coordination price range. Numerical analyses demonstrate that, when faced with heightened multidimensional risks, enterprises should increase safety stocks and ordering levels to maintain supply stability. At the same time, supply chain members should negotiate wholesale prices within a more restricted interval to achieve effective coordination. Contract flexibility and risk sharing become more significant in maintaining efficiency under high-uncertainty conditions. Sensitivity analysis demonstrates the robustness of the proposed mechanism, emphasizing its flexibility in profit sharing under stochastic conditions. This study contributes to supply chain risk management by providing a generalized contract framework that aligns decentralized decisions with centralized optimization, thereby ensuring stability and efficiency in high-risk industrial environments.
Sudden disruptions, including equipment failures, supply interruptions, and extreme events, can quickly make shop-floor schedules infeasible. Yet emergency production dispatching is often treated as a stand-alone rescheduling problem that overlooks coordination and enforcement. This study frames disruption response in industrial parks as a coupled production-control and governance problem and develops a tripartite evolutionary game model involving a local coordination authority, manufacturing firms, and an upper-level government. The model examines how proactive dispatching, cooperative rescheduling, and supervision co-evolve under bounded rationality. Using replicator dynamics, Jacobian stability analysis, and numerical simulations, we identify the conditions under which cooperative emergency dispatching becomes stable. The findings are based on a simulation-based evolutionary game model rather than on calibrated industrial case data. Results show that resource support, cooperation gains, and credible penalties promote proactive dispatching, whereas high adjustment costs weaken firms’ willingness to cooperate. Rising monitoring costs reduce the attractiveness of strict supervision, while effective horizontal cooperation partly substitutes for vertical enforcement. From a practical perspective, emergency dispatching is more effective when firms’ adjustment burdens are reduced, repeated cooperation is rewarded, passive behavior is credibly disciplined, and supervision remains effective but sustainable.
When multiple orders with different requirements must be fulfilled under ited capacity, order selection and maintenance planning for multi-component systems become strongly interdependent. To address this problem, this study proposes a condition-based maintenance strategy for multi-component systems and a joint optimization model for multi-order batch production that incorporates customer satisfaction. First, the deterioration trend of each component described using the proportional hazards model, and the health status of component is represented by virtual age. Different maintenance modes are determined according to real-time monitoring results and maintenance thresholds after the completion of each production batch. With profit maximization the objective, the condition-based maintenance strategy is integrated with capacitated batch production model and a customer satisfaction model to velop a joint optimization framework for maintenance and multi-order production. Finally, the model is solved using simulated annealing and particle swarm optimization. A case study is used to verify the effectiveness of the proposed approach and to demonstrate its practical relevance, and a sensitivity analysis of the key model parameters is also conducted.
Industry 4.0 presents a modern concept in production management that applies digital technologies and enables more efficient and faster production with minimal waste. This paper presents the concept of Industry 4.0, its development over the years, and related technologies for digitalization and automation of production management. A comprehensive systematic literature review based on the scholarly database Scopus was conducted, using VOSviewer software, along with additional analysis of selected articles by the authors of this paper. The main objective of the paper is to define the technologies most applied in production management and their importance and impact on production. Based on these results, the most important and commonly applied technologies in the manufacturing industry are defined: the Internet of Things (IoT), Artificial Intelligence (AI), and Big Data (BD). This paper highlights the advantages, disadvantages, and potential improvements of each technology in manufacturing companies. The intention of this article is to highlight the importance of applying technologies for digitalization and automation, as well as the concept of Industry 4.0, in manufacturing companies through the presentation of the literature review results. This paper is of high importance for manufacturing companies and managers in supporting decision-making regarding the application of technologies for digitalization, automation, and business improvement.
To tackle the pronounced temporal dynamics and intricate interdependencies within process manufacturing knowledge, this paper introduces an innovative framework: the Adaptive Multi-Scale Temporal Path Fusion Network (AMTPFNet). The method constructs short-term (high-frequency) and longterm (low-frequency) historical subgraphs to generate multi-scale temporal representations. It also employs a self-attention mechanism for query-aware temporal path modeling, enabling adaptive weight allocation based on varying time spans. Extensive experiments are conducted on benchmark datasets, including ICEWS18, GDELT, WIKI, and YAGO. Additionally, an application analysis is presented using electromechanical fault data. The results demonstrate that AMTPFNet exhibits remarkable effectiveness and robustness in temporal knowledge graph reasoning tasks, achieving MRR scores of 0.914 on YAGO and 0.838 on WIKI. It achieves high efficiency in predicting future production facts and assessing process quality in industrial workflows. Root causes of failures (e.g., insulation, friction) for motor components (stators, rotors) are accurately predicted, demonstrating the framework's transferability to real-world manufacturing scenarios. Although electromechanical fault data are used as a case study, the framework generalizes to manufacturing quality prediction and is readily transferable to finance, healthcare, and social media analytics.
Abrasive water jet technology is an advanced machining method that combines high-pressure water jet with solid abrasives. Owing to its unique cold-processing characteristics, high flexibility, and environmental bebefits, it has been widely applied in aerospace, medical devices, microelectronics, defense and other fields. Focusing on alumina ceramic plates, this study systematically investigates abrasive water jet (AWJ) milling through an integrated experimental and modeling approach. The research framework consists of three main phases: the development of an experimental design for abrasive water jet milling of alumina ceramics; systematic parameter optimization using single-factor and orthogonal array experiments, with material removal rate and milling depth as key performance indicators; and the application of a radial basis function (RBF) neural network model for milling depth prediction. The experimental results demonstrate that optimal parameter combinations improve machining efficiency by 38 % compared to baseline conditions. The developed RBF model achieves exceptional predictive accuracy, with maximum absolute and relative errors of 0.30 mm and 18.8 %, respectively, and a mean absolute error of 12.01 % across validation trials. This work provides a theoretical foundation for precision machining of advanced ceramics while demonstrating a viable pathway toward intelligent process optimization in AWJ technology.
Urban intelligent transportation systems require real-time, near-optimal routing for autonomous vehicles navigating dynamic and uncertain traffic. We propose a Harris Hawks Optimization-deep reinforcement learning framework (HHO-DRL) that unites HHO's global exploration with DRL's adaptive policy search through (i) a dynamic-weight fusion scheme that continuously balances exploration and exploitation and (ii) a bidirectional experience-feedback loop that exchanges elite solutions between the two solvers. On 23 CEC-2014 benchmark functions and five classical multimodal tests, HHO-DRL lowers mean error by up to three orders of magnitude relative to PSO and adaptive HHO, demonstrating superior robustness and precision. In 30 x 30 grid-world simu- lations with 30 % obstacle density, it generates vehicle routes 35 % shorter than those produced by Grey Wolf Optimization and 25% shorter than adap- tive HHO, while preserving smooth, collision-free trajectories. These results confirm that the proposed dual-mechanism delivers fast, high-quality solutions for high-dimensional, dynamic path-planning and other complex engineering optimization tasks.
This paper presents a Genetic Algorithm (GA) framework for warehouse navigation as a Travelling Salesman Problem (TSP) variant for Automated Guided Vehicles (AGVs). The warehouse layout is represented as a graph, where pick-up locations serve as terminal nodes. A distance matrix, computed via Breadth-First Search (BFS) enables efficient route evaluation. To promote diversity in the initial population, a Hamming distance-based vectorized initialization strategy is employed, ensuring that the chromosomes are maximally distinct. The GA balances exploration and exploitation by dynamically adjusting the fitness function. Early generations emphasize diversity, while later ones focus on solution refinement, improving convergence and avoiding premature stagnation. Our key contribution demonstrates that the Hamming distance-based approach achieves comparable or better results with significantly fewer chromosomes. This reduces computational cost and runtime, making the method well-suited for real-time AGV routing in warehouses. The framework is adaptable to structured environments and shows strong potential for integration into real-world logistics and robotics applications. Future work will focus on optimizing the algorithm and integrating it into the ROS 2 environment. The simplified version of the algorithm can be accessed at: https://github.com/IntoTheVoid-61/Warehouse-Pathfinder.
