Thermal spray coating technologies have emerged as a critical solution for enhancing the performance and durability of engineering components across industries such as aerospace, automotive, and energy. This review systematically examines recent advancements in thermal spray techniques, including flame spray, plasma spray, high-velocity oxy-fuel (HVOF), and wire arc spray, focusing on their working principles, advantages, and inherent limitations. It emphasizes innovative advancements in plasma spraying, including ultra-hot metallic droplets and in-situ deoxidizing mechanisms, which markedly enhance coating quality and interfacial adhesion. Thermal barrier coatings provide thermal insulation by reducing heat transfer and substrate temperature, whereas high-temperature protective coatings mainly protect materials against oxidation, corrosion, wear, and chemical degradation at elevated temperatures. The methodology involves a comprehensive analysis of characterization techniques, including microstructural evaluation, mechanical property assessment, and performance testing, to provide a holistic understanding of coating behavior under operational conditions. Our work contributes to the field by consolidating recent breakthroughs in bulk-like dense metal coatings and nanostructured coatings, offering insights into their engineering applications. The findings underscore the potential of advanced thermal spray technologies to address challenges in extreme environments, thereby enabling the development of high-performance coatings with tailored properties. Researchers and practitioners interested in using these technologies for next-gen material solutions will find this review invaluable.
The present investigation explored an innovative approach for the complex surface removal selectively using electric discharge machining (EDM) process. The parametric effects on the selective material removal of Al 6061 surface is performed by modified mild steel tool electrode where the influence peak current, pulse on-time and pulse off-time on the output responses such as TWR, MRR, Ra and ED is evaluated. The outcomes of this investigation demonstrated that the TWR, MRR, Ra and ED increase with an increase in I p and T on whereas decrease with a decrease in pulse off-time. I p has a significant contribution of 50.97% for TWR, 59.02 % for MRR 56.88 % for Ra and 54.50 % for ED as per ANOVA results. The optimal parameters selection is done using the multi-criteria decision-making TOmada de Decisao Interativa Multicriterio (TODIM) method. For equal weightages of all the output responses the common optimum condition is obtained at 6A of I p 400μs of T on and 75μs of T off .
It is important to accurately predict and optimize the machining parameters in the dynamic working conditions to enhance the surface quality, energy saving, and the productivity of the manufacturing process in CNC machining. In this study, a novel integrated intelligent CNC machining parameters prediction framework is proposed, which integrates an advanced data preprocessing method, hybrid metaheuristic optimization, and a hybrid deep learning model (HLK-Net). A preprocessing process of ARIMA-ZKF filtering and KNN imputing is applied to enhance the quality of sensor data, and a hybrid Fox Optimization–Snow Geese Algorithm (FO–SGA) is proposed for an effective balance between global exploration and local exploitation in optimizing the machining parameters. Moreover, the proposed HLK-Net combines HRNet, LSTM, and Kernel Extreme Learning Machine (KELM), which has the capability of capturing spatial, temporal, and nonlinear relationships in machining data. The superiority of the proposed framework over the existing deep learning models is demonstrated by experimental evaluation on a set of CNC Milling and CNC Turning data. The proposed framework was experimentally evaluated using CNC milling and turning data and it was verified that the proposed framework obtained 0.172 MAE, 0.231 RMSE, 0.947 R 2 , and 0.0387 MPL. The proposed model, HLK-Net, resulted in a 13.13% reduction in MAE, a 13.48% reduction in RMSE, a 25.57% decrease in prediction latency and a 2.82% increase in the R 2 compared to the conventional LSTM model. In addition, the number of iterations of convergence of FO–SGA was 39 compared to 54.12% less iterations than GA. The optimized machining parameters also reduced normalized energy consumption and surface roughness, yet another indication of the practical benefits of the proposed framework. In summary, the proposed framework can accurately, robustly, and efficiently predict and optimize the real-time intelligent CNC machining and smart manufacturing.
In machining, the boring process suffers from an increase in tool vibrations, surface roughness and cutting force due to the high slenderness ratio of the boring tool holder. A magnetorheological (MR) fluid damper was used in the boring process to enhance cutting performance during machining. However, the performance of the MR fluid damper is affected by the sedimentation of iron particles in the MR fluid. In this study, an MR fluid damper enhanced with oleic acid as a surfactant was used to enhance cutting performance and reduce the sedimentation of iron particles in the boring process. The study also investigates the influence of oil grade, iron particle size, and surfactant concentration, using oleic acid, on the rheological and damping performance of the MR fluid. The presence of oleic acid as a surfactant in MR fluid significantly improved stability by reducing particle agglomeration and sedimentation, which ensured consistent rheological behavior under a magnetic field. The experimental results showed that reduced tool vibration, improved surface finish and a decrease in cutting force were achieved when 7.5 wt.% oleic acid with 5W-30 oil grade and 60 [Formula: see text]m iron particles were used in the MR fluid damper. The study confirmed that the oleic acid effectively stabilized the MR fluid by minimizing particle sedimentation and ensuring consistent yield stress behavior under cyclic magnetic activation.
This study aims to explore and identify the critical enablers for the successful deployment of Quality 4.0 in the context of manufacturing micro, small, and medium enterprises, with explicit focus on the Indian perspective. The qualitative phase of the study comprises a narrative review followed by an in-depth interview with the experts. Semi-structured questionnaires were designed and modified to incorporate variables aligned with the background of Indian manufacturing micro, small, and medium enterprises to determine the critical enablers. The enabling factors were further prioritized using the Best-Worst Method as a reference framework to guide individuals in the successful deployment of Quality 4.0. The framework developed in this study is based on the theory of dynamic capability. The results revealed that among the major critical grouped enablers, innovation-focused culture has obtained the highest priority, followed by workforce up-skilling and competency. Quality intelligence resembles the third priority, whereas both quality culture and top management leadership resemble the fourth priority. Partnerships, networking, financial support, and organizational culture were ranked fifth and sixth, respectively. Support from top management has been perceived as an essential enabler for Quality 4.0 implementation. The critical factors identified provide a systematic approach in assessing complex organizational issues and evaluation of the efficiency of Quality 4.0 in digital transformation processes. The present research develops a systematic multi-stage methodology through combining Exploratory Factor Analysis and Best Worst Method and contributes to both theory and practice by contextualizing Quality 4.0 in an emerging economy from the context of Indian MSMEs, while providing a structured prioritization of critical enablers.
This paper introduces an intelligent visual inspection method for Industry 4.0, integrating principal component analysis (PCA) for shape alignment with a geometric center shift-based anomaly detection (GCS-AD). This computer vision-based approach is focused on external contours and is highly robust to lighting differences, rotations, and deformations in the part that can occur in industrial contexts. The experimental results presented here showed that the method achieved an average 98% classification accuracy (95% CI: [93.0%, 99.8%]) on a controlled screw dataset of 100 samples, which is better than K-nearest neighbors (87%; 95% CI: [78.8%, 92.7%]) and support vector machines (89%; 95% CI: [81.0%, 94.3%]). The descriptors formed from PCA-aligned contours proved highly interpretable, providing a straightforward visualization for quality control and decision-making. With an average processing time of approximately 0.05 s per image, the method shows potential for near-real-time use in resource-constrained industrial environments. Here, the application in screw inspection is showcased as a proof-of-concept study under controlled laboratory conditions. The method was evaluated on screws and qualitatively assessed on nuts and gears, demonstrating robustness to lighting and alignment changes in controlled tests. Despite the limited dataset, the results indicate that GCS-AD maintains high interpretability and reliability for near-real-time quality inspection. Broader industrial validation on larger and more diverse datasets is required to confirm generalizability.
The study emphasizes on optimization and evaluation of significant process parameter for Abrasive Water Jet Machining of Al7075-TiB 2 metal matrix composite. Al-TiB 2 metal matrix composite is developed through stir casting using in-situ technique. Optimization of machining parameters is done utilizing Taguchi's L 25 orthogonal array for the design of experiments, for four input parameters such as Nozzle speed, stand-off distance and abrasive flow rate and jet pressure at five different levels ranging from 100-500mm/min, 0.5-2.5mm, 100-300gms/min and 2000-4000bars respectively. Effect of machining parameters on material removal rate, surface roughness and dimensional error was determined using Analysis of Variance (ANOVA) method and the nozzle traverse speed majorly influence the Volumetric Material Removal Rate (VMRR) (99%) as well as the surface roughness (91%). The dimensional accuracy is critically dictated by jet pressure, abrasive flow rate and nozzle speed contributing to 35%, 30% and 30% correspondingly. The study also includes the comparison of Univariate optimization with multi-objective optimization. The novelty of integrating Taguchi design, ANOVA, interaction-effect analysis, and desirability-based multi-objective optimization helps in developing a robust machining strategy for Al7075-TiB 2 composites, a material for which optimization studies remain limited. The results obtained through both the models are compared with verification experiments for validation indicated the multi-objective optimization a relatively reliable technique with deviations in VMRR 0.91%, surface roughness 1.80%- and 20.16%-dimensional error that are lower than that of the single variable optimization.
Robot-assisted welding has become a key enabling technology in modern manufacturing due to its capability to enhance welding precision, productivity, and operational safety compared with conventional manual welding processes. This review provides a comprehensive analysis of robotassisted welding technologies within the framework of Industry 4.0 manufacturing systems. A systematic literature investigation based on the PRISMA methodology was conducted, leading to the detailed examination of 169 research articles covering robot configurations, kinematic modeling techniques, trajectory planning approaches, visionbased sensing systems, and advanced control strategies. The analysis reveals that articulated robotic manipulators with six or more degrees of freedom dominate industrial welding applications due to their superior dexterity and workspace flexibility. The study also highlights the growing importance of visionguided seam tracking, multisensor monitoring, and intelligent control techniques for improving weld quality and process stability. Furthermore, the integration of artificial intelligence, deep learning, and optimization algorithms has significantly improved defect detection, trajectory planning, and adaptive welding control. Despite these advancements, several challenges remain, including limited adaptability of traditional teachandplayback programming, difficulties in realtime process monitoring, and the need for robust multirobot coordination in dynamic manufacturing environments. The findings demonstrate that the convergence of robotic welding with Industry 4.0 technologies such as IoT, big data analytics, and intelligent automation will play a crucial role in developing autonomous and selfoptimizing welding systems. This review synthesizes current research trends, identifies key technological gaps, and provides future research directions to support the development of intelligent robotic welding systems for nextgeneration manufacturing.
This study considers a Physics-Informed Machine Learning (PIML) framework to analyze the mechanical cutting dynamics during radial drilling of Al-TiB2 Metal Matrix Composites (MMCs). During machining, MMCs show high tool vibration along with fluctuating cutting forces due to their heterogeneous structure and abrasive reinforcement particles. These effects make it difficult to achieve consistent dimensional accuracy in radial drilling. Sixty fractional factorial drilling trials with varying process parameters, i.e. the feed rate and the spindle speed, are conducted on Al-TiB2 composites having reinforcements up to 5.5 wt.%. Cutting force signals from these runs are processed to get four target variables, viz., maximum entry impact force, steady-state cutting force, active force variance, and diametric hole expansion. Instead of treating this process as a purely data-driven one and directly using conventional algorithms, a separate prediction of the deterministic mechanics and random micro-chatter during drilling is done. A modified version of the Kienzle power law is used for pre-processing the process parameters. On these parameters, Bayesian ridge regression is used to predict the physical baseline. The remaining nonlinear chatter is predicted using a decoupled Multi-Kernel Gaussian Process Regression (MK-GPR) with a Matern covariance function. The whole process was conducted within a strict Leave-One-Out Cross-Validation (LOOCV) loop so that no data leak happens across the 60 run dataset. The final error metrics are tested against conventional algorithms like Random Forest, XGBoost, SVR, and MLP. The proposed framework achieves an average R-2 of 0.75 and outperforms the other methods by cutting absolute prediction errors (RMSE) by 14-39%. The relationship between steady-state cutting force and process parameters is also studied using the dataset. This variation is validated using the micrographs of the chip morphology. A combined use of statistical modeling and machining physics allows more reliable and interpretable optimization of drilling parameters in MMCs.
Laser cladding has emerged as a transformative surface engineering technology capable of depositing metallurgically bonded coatings with tailored microstructures and enhanced functional performance. This comprehensive review synthesizes to establish systematic correlations between laser processing parameters, melt-pool dynamics, solidification behavior, microstructural evolution, and the resultant mechanical, corrosion, and hydrogen embrittlement properties of clad coatings. The review critically examines how key parameters, namely, laser power, scanning speed, beam diameter, powder/wire feed rate, dilution, and overlap ratio, govern grain morphology, phase constitution, defect density, and residual stress states across Ni-based, Fe-based, Co-based, high-entropy alloy (HEA), and ceramic-reinforced composite systems. Optimized processing conditions consistently yield refined cellular or equiaxed microstructures with enhanced hardness, wear resistance, and corrosion performance. At the same time, excessive heat input promotes grain coarsening, dilution-induced phase destabilization, and increased defect susceptibility. Particular emphasis is placed on hydrogen embrittlement mechanisms in laser-clad layers, revealing a multiscale interplay between hydrogen diffusion, microstructural trapping sites (grain boundaries, dislocations, phase interfaces), residual stress fields, and micro-porosity. The review demonstrates that FCC-dominated Ni/Co-based and HEA coatings exhibit superior hydrogen resistance due to stable passive films, low hydrogen diffusivity, and efficient trapping. In contrast, Fe-based systems require stringent parameter control to limit martensitic transformation and crack propagation. Current challenges and future directions are outlined, including AI-driven process optimization, in-situ monitoring, functionally graded multimaterial designs, and hydrogen-safe coating development for emerging energy infrastructures. This review provides an integrated framework for designing next-generation laser-clad coatings with optimized performance in corrosive, wear-intensive, and hydrogen-containing environments.
In the context of Industry 4.0 and digital transformation, data assets (DA) have emerged as the critical strategic resources through which circulation enterprises can develop sustainable competitive advantages. Using a sample of Chinese A-share listed circulation enterprises from 2012 to 2023, this study examines the impact of DA on capital market performance. The results show that DA not only enhances the stock liquidity but also exerts a significantly positive effect on Tobin’s [Formula: see text] after controlling for traditional factors such as firm size, highlighting the valuation premium as a new production factor. Mechanistic analysis indicates that supply chain resilience (SCR) plays a partial mediating role in the relationship between DA and capital market performance, whereas corporate digitalization levels, environmental, social, and governance (ESG) performance, and regional marketization degree significantly increase the value realization of DA. Heterogeneity analysis further reveals that, in the short-term liquidity dimension, state-owned enterprises (SOEs) exhibit stronger effects on DA, whereas in the long-term valuation dimension, nonstate-owned enterprises (nonSOEs) perform better. Moreover, firms with high technological absorptive capacity and those located in “East Data, West Computing” hub nodes exhibit more pronounced capital market benefits driven by DA. Enterprises occupying core supply chain positions are more likely to translate the DA into immediate liquidity improvements, whereas peripheral firms demonstrate stronger marginal contributions to the long-term valuation. This study deepens the understanding of value-creation pathways for data elements, enriches research on the economic consequences of DA and SCR, and provides both evidence and practical insights for circulation enterprises seeking to formulate differentiated data asset accumulation strategies and build resilient supply chains. It also offers policy insights for governments aiming to implement targeted policy support, improve the market-oriented allocation of data elements, and advance digital infrastructure construction.
The traditional wire-cut electrical discharge machining process suffers from several critical drawbacks, such as uneven surface quality, suboptimal material hardness, and ineffective tuning of process parameters, due to the multi-variable nature of machining dynamics. The research introduces a new hybrid optimization system, which combines the Piranha Predation Optimization Algorithm- Brownian Motion-Levy Flight Mechanism (PPOA-BM-LFM) and a Multi-level Stacked Houghless Network-based Bidirectional Long Short-Term Memory (MultiSHTM) for predictive modeling. The main contribution of this research is the development of an intelligent system that combines a deep learning model with a piranha-based hybrid predation optimization algorithm to enable adaptive data-driven control and immediate adjustments to wire-cut electrical discharge machining parameters. Using historical data design and MATLAB-based simulation, key wire-cut electrical discharge machining parameters- current, pulse-on time, pulse-off time, and wire speed are optimized to minimize surface roughness and maximize hardness. Complex proportional assessment-based multi-criteria analysis identified pulse on time and wire speed as major factors influencing surface roughness and hardness, achieving R 2 = 0.8914 and R 2 = 0.9521, respectively. The optimal setting (1 A, 12.83 μs T ON , 6 μs T OFF , 5 m/min wire speed) yielded 0.274 μm surface roughness, 123.43 hardness, and 0.96 desirability. The proposed model attained 97.6% accuracy and the lowest mean absolute percentage error of 1.35%, outperforming existing models with superior precision and computational efficiency.
Smart manufacturing is a key pillar of Industry 4.0, but it faces major challenges in real-time fault diagnosis and predictive maintenance due to large-scale data processing, latency constraints, and system complexity. Traditional approaches based on either edge computing or cloud computing alone suffer from limitations in accuracy, scalability, integration, and energy efficiency. They also often fail to adequately address network latency and real-time operational demands. To overcome these issues, this study proposes a Hybrid IoT-Embedded Diagnostic Framework that integrates edge-based anomaly detection using a 1D CNN with cloud-based efficiency classification using an LSTM-Attention model. The main contribution of this framework is the seamless integration of edge and cloud computing to enable real-time fault detection and predictive maintenance. Experimental results show that the proposed hybrid model achieves 98.07% accuracy, 98.09% precision, 98.07% recall, and an F1-score of 98.07%, outperforming cloud-only systems with 93.5% accuracy. Additionally, it reduces latency by 34% and energy consumption by 29%, making it more efficient and suitable for resource-constrained environments. Overall, the proposed approach provides a scalable, energy-efficient, and highly accurate solution for smart manufacturing, supporting the transition toward more intelligent and responsive Industry 4.0 systems
Contact-rich manipulation in industrial robotics faces significant challenges in skill transfer, where conventional vision-based systems rely on indirect force inference from visual observations. To address this limitation, this study developed a vision-force fusion framework combining visual and haptic measurements through learned attention mechanisms to generate manipulation trajectories satisfying contact constraints. The approach employs diffusion-based iterative refinement conditioned on multimodal observations from RGB-D cameras and force sensors. Experimental validation employed 387 demonstrations from diverse contact-rich assembly tasks using a collaborative robot with multimodal perception. The approach achieved 87.3% success rate across target task variants, outperforming Diffusion Policy (78.6%), Action Chunking Transformer (73.9%), and Behavior Cloning (62.3%), representing a 25 percentage point improvement over the baseline. Ablation studies confirmed the necessity of multimodal fusion, with vision-only achieving 67.8% and force-only 59.1%. Attention-based fusion demonstrated 8.2 percentage points higher performance than linear weight combinations while maintaining force tracking root mean square error (RMSE) of 1.38 N. Cross-task generalization experiments revealed consistent performance above 86% across geometric variations. Robustness evaluation under sensor degradation maintained 78.4% success with force noise and 75.8% with vision impairment, while achieving 3.09 s inference time suitable for real-time control. These results establish that explicit integration of haptic measurements addresses limitations in vision-based force estimation, enabling more precise contact regulation for industrial assembly operations with tight tolerances and variable component geometries.
In this study, the surface morphology and array formation and their biocompatibility of Ti-6Al-4V alloy were analyzed. Several surface texturing techniques have been developed to modify surface characteristics with the aim of minimizing bacterial adhesion and colonization. The Wire Electrical Discharge Machining (WEDM) process was employed to fabricate square array micro pillar and cylindrical surfaces on Ti-6Al-4V alloy, given its extensive use in biomedical applications. Pillars and a cylindrical array help to regulate the surface cell interaction. The study aimed to optimize machining parameters of Ti-6Al-4V alloy by minimizing kerf width and overcut to enhance surface quality. Response Surface Methodology (RSM) with a Box-Behnken design was employed for experiments and process optimization. The optimized machining parameters obtained through multi-objective optimization were TON 13.51 mu s, wire feed 6 m/min, and wire tension 7.35 gf, resulting in a minimum kerf width of 319.60 mu m and overcut of 34.80 mu m, with experimental values of 310.35 and 33.54 mu m, showing less than 5% deviation. The fabricated microstructures exhibited well-defined geometries, with square micropillars ranging from 145 mu m to 170 mu m and cylindrical features of approximately 231 +/- 10 mu m. Surface morphology analysis revealed the presence of recast layers, debris, micro-cracks, and edge rounding due to re-solidified material, while XRD results confirmed the retention of crystalline phases along with the formation of TiO2 on the machined surfaces. Antibacterial evaluation showed a significant reduction in bacterial adhesion, where untreated surfaces exhibited 1100 CFU, WEDM-treated planar surfaces showed 680 CFU, and square micro-array surfaces demonstrated only 97 CFU, corresponding to an approximate 88% reduction, while cylindrical structures showed about 61% reduction in bacterial growth.
Based on Autodesk Moldflow software, this paper studies the system simulation and process optimization of injection molding process of typical moving parts by constructing a closed-loop "design-simulation-feedback-optimization-verification" model. The results show that the introduction of Moldflow significantly improves the manufacturing accuracy and efficiency. In terms of parameter optimization, through automated script and gradient descent algorithm, multi-round iterative analysis of key parameters such as melt temperature, mold temperature and filling speed was realized, and the optimal combination was screened out, which reduced the length of the welding line to 6.8 mm and stabilized the filling pressure at 101.3 MPa. After optimization, the maximum value of warping deformation is reduced from 0.85 mm to 0.49 mm, a decrease of 42.35%; The shrinkage grade decreased from grades 3 to 2, and the improvement rate reached 33.33%; The molding cycle was shortened from 28.5 s to 24.0 s, and the efficiency was improved by 15.79%; The yield rate increased from 87% to 96%, and the overall manufacturing stability was significantly enhanced. In terms of the cooling system, by redesigning the waterway structure and simulating the temperature field, the mold temperature difference is reduced to 8.9(degrees)C, and the cooling time difference is controlled within 2.1 s, effectively reducing thermal stress concentration and dimensional deformation. The final energy consumption analysis shows that the energy consumption of the optimized scheme is reduced by 32.9%, and the energy saving effect is remarkable. To sum up, Moldflow software, in conjunction with automatic optimization process, provides strong data support and decision-making basis for high-quality manufacturing of injection molding moving parts, and has a wide engineering application prospect.
This paper examines the effect of carbon nanofiber (CNF) reinforcement on the surface finish and dimensional accuracy of the polylactic acid (PLA) composite produced using fused deposition modeling (FDM). Twin-screw compounding was used to prepare PLA-CNF composite filaments with 0-3 wt.% CNF with extrusion of filaments ( L / D ratio 30:1, 120 rpm). To prepare rectangular specimens ( 1 0 & times; 1 0 & times; 2 0 mm), the filaments were printed with the nozzle size of 0.4 mm, the layer thickness of 0.1 mm, the infill density of 100%, and the print speed of 5 mm/s. Dimensional accuracy was assessed by a digital vernier caliper with a resolution of +/- 0.01 mm, and surface roughness (Ra) was measured by a digital surface tester that was stylus-based. Every measurement was averaged three times, and the statistical examination of percent deviation and variance analysis of compositions was done. Findings show that CNF addition was important with respect to the dimensional stability and surface finish. The volume variation of the print decreased to 0.20% of PLA/3 wt.% CNF compared to 0.30% of neat PLA, which is a 33% smaller variation of the dimensional accuracy. The surface roughness also reduced as 0.70 mu m of neat PLA was reduced to 0.42 mu m of PLA/3 wt.% CNF, which is a 40% decrease. It was statistically tested that trends of consistent improvement were observed between compositions. The findings indicate that CNF reinforcement results in a higher level of geometric accuracy and less surface morphology in FDM-printed polymer composites.
Aiming to address the dual challenges of multi-objective optimization and complex demand forecasting in logistics inventory management, this study proposes an intelligent decision-making system that integrates a multi-objective flora optimization algorithm and a dense attention network. By introducing an improved bacterial foraging optimization algorithm, the three-objective collaborative optimization of inventory cost, out-of-stock risk, and service level is achieved, and a deep learning model is constructed by combining a dense attention mechanism to enhance the accuracy of demand forecasting and dynamic adjustment ability. Experimental data show that the system has verified remarkable effects in real logistics scenarios: The total inventory cost is reduced by 23.6%, the out-of-stock rate is reduced by 18.2%, and the prediction accuracy is increased by 15.8%; Under the scale of 100,000 SKUs, the decision response time has been shortened to 3.2 s, which is 40% faster than the traditional method. Through the actual measurement of a specific large-scale e-commerce platform, the annual turnover rate increased by 28%, storage space utilization rose by 19%, and the order satisfaction rate reached 98.7%; it should be noted that these results are case-specific, and their external validity in diverse logistics environments requires further verification. The innovation of this research lies in combining biological heuristic optimization with deep learning to build a two-tier decision-making architecture. A dynamic weight allocation mechanism is proposed to achieve an adaptive balance among multiple objectives. The attention mechanism is introduced to strengthen the extraction of key features and improve the robustness of the prediction model. Numerical experiments and practical cases demonstrate the effectiveness and scalability of the system in a complex logistics environment, providing a new theoretical method and practical approach for intelligent inventory management.
This paper addresses the concurrent scheduling of machines, tools, and automated guided vehicles (AGVs), within a multi-machine flexible manufacturing system (FMS). It aims to minimize makespan (MKSN) by utilizing fewer copies of each tool category, while accounting for transport times across machines to avoid tool delays. Many machines share a central tool magazine (CTM) for storing tools. AGVs and tool transporters (TT) move jobs and tools between machines. The parallel scheduling problem is inherently complex, as it requires computing the minimum number of tool copies for each type, allocating tool copies and AGVs to job operations (jb-opns), sequencing jb-opns on machines, and managing associated trip operations, including both empty and loaded AGV trip times. This study proposes a crow search algorithm (CSA) based on the crow’s intelligence and a nonlinear mixed integer programming (MIP) framework to represent and solve the concurrent scheduling problem. An industrial problem at a manufacturing firm is used to verify the process. The findings indicate that using an extra copy for one tool category, along with a copy for each subsequent tool type, would avoid tool delay and reduce MKSN, and CSA surpasses the performance of the Symbiotic Organisms Search Algorithm (SOSA).
