International Journal of Industrial Engineering: Theory, Applications and Practice

Serial Manipulator Kinematics Problem and Robot Tool Weaving Motion Analysis

2025-04-20T19:01:09+00:00

Understanding the kinematics and dynamics of an industrial robotic arm is vital for optimizing its movements, ensuring safety, planning efficient paths, and developing advanced control strategies. This study provides valuable insights into the behavior of the six degrees of freedom robotic manipulator, considering the variations in joint angles and the corresponding range of motion in relation to the nature of the traversed path. Both the forward and inverse kinematic problems are addressed, and numerical approach is utilized. Using a weaving motion with the welding torch to deposit filler metal across the joint is considered a method for analysis. Throughout different welding trajectories featuring sections containing curved and oblique trajectories, the analysis of the robotic arm is conducted. This study can contribute to effectively develop a welding robotic system in various manufacturing settings and can lead to improvements in quality by refining precision and reducing errors while also to drive cost reductions through the increased efficiency.

The Impact of Sitting Posture on Prolonged Reading: Analyzing Chair Depth and Height in Classroom Settings

2026-02-17T18:55:46+00:00

This study aimed to investigate the influence of sitting posture on the distraction levels during prolonged reading sessions among college students. The environmental variables examined were confined to adjustments in classroom seating, specifically seat depth and backrest height. The participants are university students with normal vision and no history of musculoskeletal disorders. The findings revealed no significant interaction effect between seat depth and backrest height on attention during extended reading. Nevertheless, participants seated on the front third of the chair or without backrest support exhibited higher levels of distraction. Based on the statistical findings, sitting in the front two-thirds of the seat, with support above the lower back, has proven more effective during extended periods of reading. As a result, schools should consider purchasing chairs with adjustable backrests or mandate the use of lumbar pads. These measures allow students to adjust seat depth, promoting better posture and enhancing reading performance during prolonged classroom sessions.

A Multicategory Classification Model with Regularized Pairwise Comparison

2026-02-04T20:12:29+00:00

Multicategory classification poses a significant challenge in machine learning because real-world problems often involve multiple classes, while many algorithms are tailored for binary classification. Among the methods for extending binary classifiers to multicategory classification problems, two representative approaches are the “One-versus-The-Rest” (OVT) and “One-versus-One” (OVO) methods. We demonstrate that the OVT method can encounter masking issues in certain situations and propose a new algorithm using the OVO approach to overcome this problem. In particular, we develop a method that integrates the OVO approach with the Mallows-Bradley-Terry model to estimate pairwise probabilities and incorporates regularization techniques into a single optimization framework to reduce computational cost. The penalty functions considered include LASSO and Ridge-LASSO penalties. Experimental results and real data analysis indicate that the proposed method outperforms the OVT method and is less affected by the masking problem.

A Multi-Objective Model for Job Shop Flexible Scheduling Problem by Considering The Flexibility in Sequencing and Constraints of Activity in Uncertainty Condition

2025-09-24T18:36:17+00:00

The current research aims to develop a multi-objective model for the flexible job shop scheduling problem (FJSP) by incorporating flexibility in sequencing and activity constraints under conditions of uncertainty. To achieve this, a comprehensive study was conducted to identify research gaps. Based on these gaps, the assumptions and research objectives were formulated. The proposed multi-objective mathematical programming model (MOMPM) focuses on three primary goals: minimizing makespan, minimizing labor workload, and minimizing weighted delays. The contributions of this study include flexibility in sequencing, activity constraints, uncertainty in processing times, and sequence-dependent setup times. The model is solved using a multi-objective genetic algorithm (MOGA), and the results indicate that flexibility in activities does not lead to suboptimal solutions, nor do activity constraints. However, uncertainty in processing times can adversely affect the optimal solutions across all objectives.

An Expert System with Layered Evolution-Based Feature Extraction for Epidemic Monitoring

2026-02-04T20:08:42+00:00

Enhancing epidemic monitoring capabilities to support policymaking has become a global health priority. This study proposes a layered evolution-based expert system that integrates feature extraction, feature importance analysis, and robustness-oriented design for epidemic trend monitoring. The proposed framework combines genetic algorithm (GA)-based evolutionary mechanisms with Taguchi method-based optimal operating condition design to enhance analytical stability under varying configurations. Using publicly available COVID-19 data from Japan, the proposed system monitors epidemic dynamics across four mortality levels while simultaneously identifying representative features and estimating their relative importance. The results demonstrate that the system reduces the feature set by approximately 27% while maintaining stable predictive performance under more granular severity definitions. Robustness is further evaluated through statistical validation under varying dataset compositions and scales. Overall, the proposed expert system provides an interpretable and robust framework for epidemic trend monitoring that complements existing prediction-oriented approaches and may support informed public health decision processes through feature-level insights.

Efficient Scheduling for Machining Processes Considering Machine Unavailability Constraints

2026-02-04T20:09:34+00:00

This paper presents an integrated scheduling framework for single-machine machining processes to deal with the challenges posed by machine availability constraints due to preventive maintenance, cooling intervals, and unexpected breakdowns. Our primary objective was to minimize the makespan while accounting for job-specific processing and release times. The NP-hard nature of the problem rendered our initial mixed-integer programming model computationally prohibitive for large instances. Thus, we tailored a branch-and-bound algorithm to derive optimal solutions for small to medium-sized problems by leveraging specific upper and lower bounds. We then employed a genetic algorithm using a permutation-based chromosome and problem-specific decoder to generate near-optimal solutions for large-scale instances. This hybrid approach balances solution quality with computational efficiency, rendering it highly suitable for a wide range of practical industrial manufacturing applications.

A Comparative Study of Metaheuristic Algorithms for Scheduling on Unrelated Parallel Machines: Minimizing Weighted Earliness–Tardiness with Non-Zero Release Times and Distinct Due Dates

2026-01-11T07:05:38+00:00

This study addresses the unrelated parallel machine scheduling problem in a just-in-time manufacturing context, aiming to minimize total weighted earliness and tardiness. The problem formulation incorporates non-zero release times and distinct due dates, reflecting realistic industrial environments. Three hybrid metaheuristic approaches: Genetic Algorithm, Tabu Search, and Variable Neighborhood Search, are proposed and analyzed. The main contribution of this work lies in integrating a linear-programming-based decoding procedure into each metaheuristic to determine job start times and accurately evaluate solution quality, while preserving the general structure of the unrelated parallel machine scheduling problem. The proposed methods are evaluated using a set of medium- and large-scale instances generated for this study. Computational analysis reveals differences in performance among the metaheuristics, with Tabu Search exhibiting the most consistent and effective behavior in terms of solution quality and convergence speed.

Predictive AI-Based Maintenance Model for Slurry Pumps Using ISO 10816-3 Vibration Standard

2026-01-11T06:59:53+00:00

This study proposes a predictive maintenance model for slurry pumps used in the mold flux drying process by integrating the ISO 10816-3 vibration standard with AI-based machine learning algorithms. Real-world sensor data on vibration, current, pressure, revolutions per minute (RPM), and temperature are collected and processed for training and evaluating AI models based on eXtreme Gradient Boosting (XGBoost), Long Short-Term Memory (LSTM), and Random Forest (RF) algorithms. The vibration zones defined by the ISO 10816-3 standard are employed as ground truth labels to enhance the interpretability and reliability of the predictive maintenance models. Among the AI models, the XGBoost-based model demonstrates the best predictive performance. A hybrid diagnostic system is developed by integrating the ISO-defined vibration thresholds with the XGBoost algorithm, which provides effective early warning alerts for potential failures in real time.

Dynamic Traffic Density Prediction in Urban Areas Using Machine Learning Models: Comparative Analysis of Classification Performance

2026-01-11T06:48:32+00:00

Accurate traffic density prediction is essential for reducing congestion, saving time and fuel, lowering emissions, and improving urban mobility. It also supports decision-making for drivers and traffic management authorities. Various dynamic factors influence traffic density and rely heavily on both historical and real-time data. This study explores the application of machine learning (ML) techniques in dynamic traffic density prediction and introduces a data-driven model. Three ML algorithms—Random Forest (RF), Light Gradient Boosting Machine (LGBM), and Extreme Gradient Boosting (XGBoost)—were evaluated using a real-world dataset. The RF algorithm demonstrated superior performance, especially in accurately classifying different levels of traffic congestion. The model's performance was assessed using comprehensive metrics beyond standard accuracy, including F1-score, Matthews Correlation Coefficient, and AUC. These findings highlight the potential of ML-based classification models to enhance traffic forecasting and support intelligent transportation systems in urban environments.

Modeling and Precise Prediction of Unobservable Pouring Processes Based on Machine Learning

2026-02-04T20:06:28+00:00

Accurate online measurement of pouring speed and charge amount remains a critical challenge in complex component manufacturing. This challenge is primarily due to the retention and adhesion of molten material in pipelines and cavities, preventing direct measurement of outlet flow. As a result, process quality heavily relies on operator experience, leading to inconsistent results and frequent defects. To address this, we propose a machine learning (ML)-based soft-sensing approach combined with multi-sensor data fusion. By integrating process parameters—such as material viscosity, temperature, and vacuum level—with structural characteristics of the pouring system, we develop a nonlinear mapping model that transforms unmeasurable outlet parameters into a regression task. Six algorithms were evaluated under identical conditions, with three showing superior accuracy in capturing both global trends and local fluctuations. This framework enables high-precision, real-time prediction of pouring parameters, laying the foundation for intelligent pouring and improved process control.

Adoption Decision of Low-Carbon Technology in The Offshore Green Supply Chain Considering Government Subsidies

2026-02-18T19:33:17+00:00

The adoption of low-carbon technology in the offshore green supply chain is being increasingly affected by government subsidies. To address this issue, we construct an offshoring green supply chain system, which is composed of one OEM and two CMs, and examine how government subsidies impact the adoption of low-carbon technology and social welfare. First, we construct three benchmark models in the case of no government subsidies, and three extended models in the case of government subsidies. All six models consider three different scenarios: a) The situation when none of the members adopt the low-carbon technology; b) The situation when only one CM adopts low-carbon technology; c) The situation when two CMs adopt the low-carbon technology. Based on the above six game theoretical analysis models, we find that in the case of no government subsidy, the wholesale price is proportional to the carbon tax rate of its own country and inversely proportional to the carbon tax rate of the country where the other CM is located. The selling price is proportional to the carbon tax rate of each country. When the government decides to subsidies, the most important factors for the government are the emission reduction rate and carbon emissions, and the government subsidy coefficient can reach a peak when both the emission reduction rate and carbon emissions are high. In order to mitigate the carbon taxes faced by the OEM, we design a cost-sharing contract and demonstrate that all members of the supply chain can benefit from this contract. The analysis in this paper provides important guidance for the optimal decisions of the members in the offshoring green supply chain.

Closed-Loop Supply Chain Cost Optimization for Make-to-Order Production: A Case Study of A Shaft Manufacturing Enterprise

2026-02-03T20:08:21+00:00

Make-to-order (MTO) systems require flexible production schedules and inventory management. The outcome of this study is a hybrid MTO Closed Loop Supply Chain (CLSC) model that combines raw/recycled materials for cost minimization. In this study, a mixed-integer programming (MIP) model and implement two solution approaches have been developed: a simplified mixed-integer linear programming (MILP) for computational efficiency and a genetic algorithm to solve full mixed-integer nonlinear programming MINLP. Monte Carlo simulation was used to account for variability in demand, return rates, and production processes. The contributions of this work include: CLSC model tailored to MTO with dual MIP formulations with comparative analysis of MILP and MINLP performance, practical insights into implementing cost-efficient hybrid CLSC. There is a research gap, which most studies assume deterministic conditions in MTO CLSC. The presented work covers this gap and models the complexity of MTO where demand, lead times, and return rates are uncertain. The model in this work is developed based on shaft manufacturing industrial setups. However, the proposed methodology can be applied in other contexts with needed customization and changes to accommodate the specifications of other context which are characterized by high-value component recovery.

The Impact of Supply Chain Digitalization on Enterprises' Foreign Trade

2026-02-03T20:14:15+00:00

Examining the influence of supply chain digital construction upon enterprise export trade furnishes empirical substantiation for Chinese firms' global expansion. This analysis dissects the mechanism underpinning supply chain digitalization across three vectors: synergistic interplay, capital expenditure mitigation, and Information transparency. It further probes heterogeneous manifestations contingent on external contingencies, encompassing enterprise market standing and pollution categorization. Findings reveal that supply chain digitalization markedly propels enterprise foreign commerce development, principally via synergies, capital cost amelioration, and Information transparency enhancements. External environmental constituents, notably digital backbone frameworks and market fragmentation extent, moderate this impact on export trade. Consequent deductions necessitate Chinese enterprises expediting supply chain digitalization, harnessing digital modalities to augment supply chain collaborative proficiency and reactivity, curtail capital outlays, and elevate informational lucidity, thereby bolstering international market standing. Concurrently, policymakers must enact supportive statutes facilitating enterprise digital metamorphosis within supply chains. Additionally, refining the external milieu—fortifying digital infrastructure and diminishing market segmentation—proves indispensable for unleashing the complete affirmative function of supply chain digitalization in enterprise export trade.

The Impact of Green Subsidy and Carbon Quota Allocation Modes on Optimal Decisions in A Low-Carbon Supply Chain

2026-02-18T19:38:54+00:00

This study investigates a two-echelon low-carbon supply chain comprising one manufacturer and one retailer, considering distinct green subsidy and carbon quota allocation modes. Four Stackelberg game models are developed, including the GT (grandfathering rule combined with green subsidies based on emission abatement cost), BT (benchmarking rule combined with green subsidies based on emission abatement cost), GA (grandfathering rule combined with green subsidies based on emission abatement amount) and BA (benchmarking rule combined with green subsidies based on emission abatement amount) game models. The joint impacts of carbon-related and subsidy-related parameters on optimal solutions, profitability, environmental benefits, and social welfare of the proposed models are analyzed. A comparative analysis is proposed in terms of optimal equilibrium solutions and model performance. Moreover, numerical simulations are conducted to verify the research results and to seek more managerial insights. The results reveal that a high degree of consumers’ low-carbon preference is conducive to enhancing performance for the entire supply chain. The carbon trading price has different impacts on the chain members’ profits. Moreover, the AB game model is the best option to improve the profitability of the retailer and the social welfare of the entire supply chain, and the manufacturer prefers the GA game model from a profitability perspective. The research findings not only serve to optimize business strategies and performance of enterprises but also provide a theoretical foundation and practical guidance for the effective implementation of low-carbon policies for policymakers.

Bearing Fault Diagnosis Using One-Dimensional Convolutional Neural Network-Informer-Multi-Head Attention Mechanism Parallel Modeling

2026-02-03T20:06:43+00:00

Based on the complex characteristics of rolling bearing vibration signals, such as non-stationarity, high noise, and long-sequence dependencies, this paper proposes a parallel fault diagnosis model that integrates a one-dimensional convolutional neural network (1DCNN), Informer, and MHA. This model effectively combines the local time-frequency feature extraction capabilities of 1DCNN, the long-sequence modeling advantages of Informer based on the ProbSparse mechanism, and the dynamic fusion of multi-scale features by the multi-head attention mechanism (MHA) to achieve efficient feature representation. To enhance the model's noise immunity and feature relevance, a dual preprocessing method of variational mode decomposition (VMD) and fast Fourier transform (FFT) is introduced to strengthen the capture of fault features in non-stationary vibration signals. Comparative experimental results on the Case Western Reserve University (CWRU) standard dataset demonstrate that the proposed model exhibits significant diagnostic performance. The model achieves an average accuracy of 99.9% and an of 0.99 under ten fault conditions, with minimal standard deviation. It outperforms several existing methods (such as CNN-LSTM, TCN, and Transformer), demonstrating its superior accuracy, stability, and robustness. Further t-SNE visualization analysis intuitively demonstrates the model's excellent classification boundaries and cluster separability in high-dimensional feature spaces. This research presents a highly accurate, efficient, and generalizable solution for intelligent fault diagnosis under complex operating conditions, offering significant reference value for promoting theoretical development and industrial applications in the field of intelligent diagnosis.

A Decision Model for Service-Oriented Manufacturing Firms Based on The Product Lifecycle

2026-02-18T19:36:59+00:00

Facing saturated product markets, manufacturers are transforming by embedding services across the product lifecycle. This study develops a lifecycle-based game-theoretic model to determine the optimal product price, service price, and service quality for service-oriented manufacturers. Using backward induction, the model identifies optimal decisions under distinct profitability scenarios and examines the influence of key market factors through sensitivity analysis. Findings reveal that the trade-off between product and service profits is shaped by the warranty period and the ratio of consumer sensitivity to product price versus service utility. Accordingly, three profitability regimes—product-dominated, mixed, and service-dominated—are defined. A shorter warranty or higher sensitivity ratio promotes high service quality, whereas a longer warranty or lower ratio favors lower quality. Extending the model to include partial in-warranty service charging shows that charging a fraction of service fees maintains profitability while supporting extended warranty coverage, offering actionable insights for lifecycle-based decision-making.