Engineering reports : open access最新文献

The demand for miniaturized metallic components in electronics, biomedical devices, and aerospace necessitates sustainable micro-forming solutions. Conventional deep-drawing often suffers from stage complexity, excessive die use, and size-effect limitations. This study aims to optimize stage number, limiting drawing ratio (LDR), and diametrical reduction for sustainable fabrication of copper micro cups. Directionally rolled pure copper strips with 250% deformation (strain −3.5) and an initial thickness of 0.1895 mm were used. Finite element analysis (FEA) was performed to design multi-stage deep-drawing die sequences, with validation through experimental trials. Three strategies were investigated: a 4-stage process (30% reduction per stage), a 6-stage process (15% reduction), and an 8-stage process (15%–10% reductions). Experimental punch load, strain distribution, and thickness profiles were compared against simulation. Results showed that while the 4- and 6-stage processes failed due to thinning and fracture from reduced formability, the 8-stage design achieved defect-free cups with uniform wall thickness. Bidirectional rolling (BDR) yielded higher dimensional accuracy and reduced thinning compared to unidirectional rolling (UDR), as confirmed by ISO 24213 criteria. Optimizing stage number and LDR proved critical in controlling flow stress, minimizing die wear, and improving sustainability. The study focused on copper microparts of specified dimensions. Broader validation across alloys, geometries, and rolling conditions is required. The findings provide industries with a framework to reduce energy, material waste, and die consumption while ensuring micropart quality. This is the first integrated study combining grain-size-controlled copper blanks, FEA-driven multistage die design, and experimental validation for sustainable micro-deep drawing.

This study investigates the effects of heat treatment (HT) and severe shot peening (SSP) on the surface integrity of binder jetting (BJ) manufactured 316L stainless steel. While HT step was chosen for its proven effectiveness in relieving residual stresses in PBF-LB built 316L, it was observed to increase porosity in BJ samples from 2.5% to 7.5%. SSP alone, however, effectively enhanced surface hardness from 145 to 504 HV, introduced beneficial compressive residual stresses reaching −995 MPa at a depth of 91 μm (remaining compressive up to 300 μm), and reduced surface porosity to 0.45%. These improvements indicate a significant enhancement in surface integrity, thus potentially improving wear and fatigue resistance. The findings suggest that SSP is sufficient for optimizing surface properties in BJ components, offering an effective post-processing approach for high-performance applications.

In urban public spaces, maintaining high power quality is very much needed for reliable and efficient energy consumption. This study aims to develop and validate an effective design method focused on improving urban power quality through the integration of renewable wind energy. This research proposes a novel and unique design method for enhancing and improving power quality in urban areas by connecting the wind energy through the utilization of vertical-axis wind turbines (VAWTs). The whole concept of the proposed methods involves a structured methodology comprising system modeling, integration of VAWTs with a Unified Power Quality Conditioner (UPQC), and experimental validation to measure voltage stability, total harmonic distortion (THD) and reactive power performance. The UPQC, an advanced power electronic device, operates by combining series and shunt compensators to address a wide range of power quality disturbances simultaneously. The series compensator handles the whole voltage-related problem and the shunt compensators fully manage and coordinate the current-related issue. This dual compensation approach ensures synchronized mitigation of both voltage and current disturbances, thereby maintaining consistent grid performance. By utilizing wind energy harnessed from VAWTs, the recommended system provides an alternative and renewable source of power, minimizing dependencies on the conventional grids and improving the overall energy efficiencies. The vertical turbines are chosen due to their excellent adaptability and suitability for all urban environments, where space limitations and varying wind directions at all positions face significant challenges. The research contains a detailed analysis of the performance enhancement brought about by the UPQC in parallel with VAWTs, leading on key power quality metrics. Experimental results show a significant minimization in voltage sags and swells, with the normalized sag values improving by up to 75% and swell values by up to 65%. The method improves power stability and promotes sustainability by combining renewable energy with advanced power electronic solutions in urban areas.

Bangladesh's geographical location as well as dense population makes it highly vulnerable to rising maximum temperatures and the associated public health impacts driven by climate change. Although machine learning-based forecasting and correlation analysis are widely applied, most existing models lack regional calibration and consequently fail to capture Bangladesh's distinct climatic variability. In addition, the accuracy of forecasting of maximum temperature cannot be determined since the results are inconsistent. This study addresses these challenges by proposing and evaluating four forecasting models: the Autoregressive Integrated Moving Average (ARIMA) model, the Exponential Smoothing State-Space (ETS) model, the Holt (Double Exponential Smoothing) model, and the Prophet model. The models forecast maximum temperature and examine its correlation with public health issues—particularly diarrhea cases in Bangladesh—over the period from 2017 to 2040. The study compares the forecasting performance of these state-of-the-art models, showing that maximum temperatures are projected to range between 31.39°C and 34.3°C by 2040. ETS and Holt predicted 3786 and 2282 diarrhea cases respectively at 33.71°C, while Prophet estimated 3841 cases at 34.3°C. ARIMA forecasted 4045 diarrhea cases with a predicted maximum temperature of 32.15°C, achieving a moderate error margin. Among the models, ARIMA demonstrated the most balanced and reliable performance, confirming its effectiveness for climate and health forecasting in the Bangladeshi context. Overall, this work outperforms existing approaches by delivering consistent and accurate temperature forecasts. Additionally, it bridges a significant gap in the literature by establishing a clear correlation between maximum temperature and diarrhea cases in Bangladesh. The findings offer actionable insights for policymakers and public health officials, supporting more effective strategies for public health management and climate adaptation planning.

Blockchain networks now support billions of dollars in daily transactions, making reliable and transparent fraud detection essential for maintaining user trust and financial stability. Yet, real-world blockchain datasets are extremely imbalanced, with fraudulent activity representing less than 1% of all transactions. This imbalance causes conventional machine learning models to achieve deceptively high accuracy while still failing to detect a substantial portion of fraudulent events. To address this challenge, this study evaluates the performance and explainability of three models-XGBoost, LightGBM, and Decision Tree-on the Ethereum-based fraud detection data, in which 58% of transactions are identified as fraud. The methodology combines vast feature engineering, k-fold cross-validation, and assorted resampling approaches, such as Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling Nearest Neighbor (ADASYN), to revise the effect of class mismatch. Accuracy, AUC, recall, precision, F1-Score, and Matthews Correlation Coefficient(MCC) are used to measure model performance, and SHapley Additive exPlanations (SHAP) is utilized to give global and local interpretability. Experimental results show that XGBoost combined with SMOTE or ADASYN yields the strongest performance, achieving a recall over 99%, an AUC of 1.000, and a substantially improved MCC compared to training on the raw imbalanced data. LightGBM presents a favourable precision-recall balance, and Decision Trees demonstrate significant gains after resampling, despite their simplicity. SHAP analysis reveals that log-transformed transaction amount, merchant-based encoding, geographic encoding, and temporal features are the primary contributors to fraud risk. These results are important in highlighting two implications: (i) the importance of dealing with extreme class imbalance, rather than choosing increasingly sophisticated approaches, and (ii) the ability to be trusted to be explained is a requirement of responsible working in both financial and blockchain settings. The research offers a pragmatic, interpretable framework on blockchain fraud detection and future directions, including sophisticated hybrid sampling, collective learning, as well as cross-chain generalization to enhance fraud detection in distributed systems.

In this work, the adsorption properties and sensitivity of Sb-doped MoS₂ for the selective detection of various environmental toxic gases (CO, CO₂, NO, NO₂, SO₂, and SO₃) were investigated using Density Functional Theory combined with the Nonequilibrium Green's Function formalism. The Perdew–Burke–Ernzerhof functional within the Generalized Gradient Approximation was employed for the computational analysis. Key parameters such as adsorption energy, charge transfer, bandgap, Density of States, Projected Density of States, optical properties, work function, recovery time, current–voltage characteristics, and sensitivity were examined to understand the adsorption behavior of the Sb–MoS₂ monolayer toward these gases. The results indicate strong adsorption energies for SO₃ and NO₂, with values of −1.71 and −1.50 eV, respectively. SO₃, NO, NO₂, and CO exhibit chemisorption, whereas SO₂ and CO₂ undergo physisorption. The optical analysis reveals noticeable changes in the absorption and reflection of incident photon energy upon gas adsorption. Among all gases, CO shows the highest sensitivity of 68.32% at a bias voltage of 1.7 V, while NO exhibits the lowest sensitivity of 21.86% at 1.8 V. This study lays the groundwork for the development of Sb–MoS₂ monolayers as highly sensitive FET-based sensors for the detection and monitoring of environmental toxic gases.

Urban air pollution, particularly nitrogen dioxide (NO₂), remains a critical environmental and public health concern in rapidly growing cities. This study explores the spatiotemporal patterns of NO₂ concentrations in Chennai and Bengaluru from 2019 to 2023 by integrating satellite-based datasets and statistical modeling on the Google Earth Engine (GEE) platform. Sentinel-5P TROPOMI data were used to assess NO₂ levels, Sentinel-2-derived NDVI represented vegetation cover, and Landsat 8 imagery provided land surface temperature (LST) estimates. Seasonal trends were analyzed for both summer (March–June) and winter (November–February) periods. Results revealed pronounced seasonal variability, with Chennai exhibiting consistently higher NO₂ concentrations in winter, while Bengaluru displayed more stable or decreasing trends. Notably, NO₂ levels in Chennai rose by 15.4% during summers over the study period, whereas Bengaluru saw a 16.6% decrease. A comparative regression analysis showed that the relationship between NO₂ and vegetation cover (NDVI) strengthened in Chennai during winter (R ² = 0.043 in 2023), suggesting reduced green cover may intensify pollutant accumulation. Conversely, Bengaluru showed stronger NO₂–NDVI correlations during summer (R ² = 0.049 in 2023), indicating vegetation's role in pollutant mitigation during active growing seasons. The NO₂–LST relationship also varied: Chennai experienced the strongest positive correlation in summer 2022 (R ² = 0.30), whereas Bengaluru exhibited increasing winter correlations, potentially driven by surface warming and enhanced atmospheric mixing. Although direct meteorological parameters such as rainfall, humidity, wind speed, solar radiation, and visibility were not included in the present analysis, their influence on NO₂ dynamics is acknowledged and warrants future exploration. Overall, the findings underscore the complex, season-specific interactions among urban heat, vegetation, and air pollution in different metropolitan contexts. These insights support the need for tailored, climate-responsive pollution control strategies that integrate urban greening, emission reductions, and adaptive planning.

As a core model for promoting technological industrialization, school-enterprise collaborative innovation faces multi-dimensional risks from technology maturity fluctuations, market policy changes, and multi-party interest games. Traditional risk assessment methods like AHP and SVM rely on static indicators and single-objective optimization, struggling with dynamic constraints. While multi-objective evolutionary algorithms handle objective conflicts, they suffer from low convergence efficiency in high-dimensional spaces and poor real-time performance. Data sparsity and limited emergency scenario generation further restrict industrial applicability. This study proposes a technology transfer risk assessment framework integrating deep adversarial reinforcement learning with multi-objective optimization. We develop a physically-constrained adversarial generation network to simulate technology failure distributions and market fluctuation patterns, generating high-fidelity risk scenarios. Combined with the Proximal Policy Optimization algorithm, we design a dynamic decision-making mechanism that simultaneously optimizes risk control costs, technology transfer efficiency, and patent revenue. Experiments in semiconductor manufacturing and new energy batteries demonstrate significantly improved assessment accuracy and decision-making speed. The framework achieves a 92% decision correction rate with 1.2-h response delay in emergencies, overcoming traditional methods' limitations in dynamic multi-objective collaboration. Adaptive reference point strategy and adversarial training effectively address data distribution bias and noise interference, providing practical intelligent decision support for school-enterprise innovation.

Inspection of surface defects in industrial components is vital to ensuring product quality and operational safety. This study presents an improved YOLOv8s-based deep learning model designed to detect fatigue and linear cracks on part surfaces with high localization and classification accuracy. Unlike conventional manual inspection, the proposed system achieves real-time detection while maintaining low computational cost. A custom dataset comprising high-resolution images of concrete and metallic textures captured under diverse environmental conditions was used for training and evaluation. Data augmentation techniques such as Mosaic and Contrast Limited Adaptive Histogram Equalization (CLAHE) were employed to enhance generalization, and the model was trained for 100 epochs to ensure stable convergence. The enhanced YOLOv8s architecture integrates the Convolutional Block Attention Module (CBAM) and Ant Colony Optimization (ACO) for intelligent feature selection, resulting in improved learning efficiency and reduced overfitting. Experimental results show that the model achieves mAP@IoU = 0.5 of 0.9626 and [email protected]:0.95 of 0.7803, with validation box and class losses of 0.6849 and 0.5377, respectively. Precision and recall values of 0.926 and 0.923 demonstrate excellent detection completeness and low false positives. Comparative analysis with BsS-YOLO and Efficient YOLOv8-ES confirms the superior accuracy and efficiency of the proposed approach. Visual inspection further validates the model's robustness in identifying diverse crack patterns under challenging surface conditions. The improved YOLOv8s model thus offers a scalable, real-time, and accurate solution for automated defect detection in industrial applications. Future work will focus on multi-class defect recognition and deployment on lightweight edge-computing devices.

Suicide ranks as the 18th leading cause of death worldwide among young adults, claiming over 720,000 lives each year. Early detection of individuals at risk is essential for timely intervention. This study introduces a Triple-Layer Ensemble (TLE) model that predicts suicidal behavior using machine learning techniques. The proposed model combines Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) to enhance prediction accuracy. Experimental results show that the TLE model surpasses individual classifiers and traditional ensemble methods, achieving 94.81% accuracy, 98.15% ROC-AUC, and a Matthews Correlation Coefficient (MCC) of 92.23%. To improve interpretability, Explainable AI (XAI) methods, Local Interpretable Model-Agnostic Explanations (LIME), and Shapley Additive Explanations (SHAP) highlight key predictors such as mental support, stress levels, and self-harm history. Additionally, a web-based platform incorporating the TLE model provides real-time suicide risk assessment, enabling healthcare professionals to implement personalized interventions. The proposed framework delivers high predictive performance with transparency and interpretability, offering a scalable solution for early suicide risk prediction and prevention.