Search Results (155)

The recognition of ice thickness on transmission lines serves as a prerequisite for controlling de-icing robots to carry out precise de-icing operations. To address the issue that existing edge computing terminals fail to meet the demands of ice thickness recognition algorithms, this paper introduces an Efficient Edge Computing Network (EECNet) specifically designed for identifying ice thickness on transmission lines. Firstly, pruning is applied to the Efficient Neural Network (ENet), removing redundant components within the encoder to decrease both the computational complexity and the number of parameters in the model. Secondly, a Dilated Asymmetric Bottleneck Module (DABM) is proposed. By integrating different types of convolutions, this module effectively strengthens the model’s capability to extract features from ice-covered transmission lines. Then, an Efficient Partial Conv Module (EPCM) is designed, introducing an adaptive partial convolution selection mechanism that innovatively combines attention mechanisms with partial convolutions. This design enhances the model’s ability to select important feature channels. The method involves segmenting ice-covered images to obtain iced regions and then calculating the ice thickness using the iced area and known cable parameters. Experimental validation on an ice-covered transmission line dataset shows that EECNet achieves a segmentation accuracy of 92.7% in terms of the Mean Intersection over Union (mIoU) and an F1-Score of 96.2%, with an ice thickness recognition error below 3.4%. Compared to ENet, the model’s parameter count is reduced by 41.7%, and the detection speed on OrangePi 5 Pro is improved by 27.3%. After INT8 quantization, the detection speed is increased by 26.3%. These results demonstrate that EECNet not only enhances the recognition speed on edge equipment but also maintains high-precision ice thickness recognition. Full article

Background/Objectives: Evidence on associations between nutrients and dementia risk is limited and inconsistent. We aimed to systematically examine associations between 101 dietary nutrients and dementia incidence with a nutrient-wide association study (EWAS). Methods: We analyzed data from 6280 participants aged 50 years and older from the Health and Retirement Study. Levels of nutrient intake were measured with the food frequency questionnaire. Dementia status was assessed with the Lang–Weir Classification of Cognitive Function. In the EWAS analysis, the Cox proportional hazards regression model was used to estimate associations between each nutrient and dementia incidence, adjusting for multiple comparisons with a false discovery rate (FDR) of 0.05. Nutrients passing the EWAS selection were simultaneously included in the elastic net (ENET) regression model to construct a composite nutrient score (CNS), which was calculated as a weighted sum of the nutrients in the ENET regression model. Results: Over a mean (SD) follow-up of 6.76 (2.14) years, 495 individuals with incident dementia were identified. The results suggested that six nutrients were associated with increased dementia risks and five with decreased dementia risks. Compared with participants at the first tertile of CNS, individuals at the second (hazard ratio [HR] = 1.43, 95% confidence interval [CI] = 1.11 to 1.84) and third tertiles (HR = 1.80, 95% CI = 1.42 to 2.27) had increased risks of dementia. Furthermore, CNS-dementia associations were stronger in females than in males. Conclusions: We found that 11 dietary nutrients and their combinations were associated with dementia risks in middle-aged and older adults. Interventional studies with nutrients were warranted to confirm our findings. Full article

Concentrated apple juice (AJC) is a globally traded commodity, with major producers such as China, Poland, and the United States supplying AJC to markets worldwide. The aim of this article is to determine the competitive position of the main global exporters of concentrated apple juice. It also seeks to analyze changes in their positions over the period from 2005 to 2023. Assessing competitive position is important for several economic and business reasons, including identifying leading exporters and recognizing both growing and declining markets. The competitive position was measured using Market Share (MS) indicators, Gross and Net Export Orientation indicators (E_gr, E_net), and the Revealed Comparative Advantage (RCA) index. The results reveal significant structural shifts in global AJC trade. Most notably, China’s declining competitiveness, reflected across all indicators, led to its loss of market leadership. This raises questions about the reasons for this decline and whether it presents opportunities for other exporters. This signals a broader reconfiguration in the global AJC supply chain, driven in part by domestic economic changes, such as rising consumption and decreasing export orientation. Simultaneously, other countries, particularly in Eastern Europe, show varying degrees of competitive growth, with Moldova and Ukraine emerging as key players. These trends suggest a diversification of supply sources and a more fragmented competitive landscape. Although national differences persist, the analysis indicates that structural and economic transformations, rather than short-term price signals, are driving the evolving global competitiveness in the AJC market. The observed weak correlations between prior-year apple prices and RCA confirm that broader market and policy factors play a more decisive role. Full article

Accurately identifying road conditions, particularly wetness, is crucial for ensuring road safety and enhancing vehicle performance. We conducted road surface classification and road wetness estimation using state-of-the-art deep learning models in this study. Raspberry Pi Model 4 was used to classify road surfaces and estimate road wetness. SqueezeNet, a lightweight convolutional neural network, was used to recognize wet and dry road surfaces with an accuracy of 90%. The ENet model, known for its efficiency in semantic segmentation tasks, was used to estimate the degree of wetness, categorizing roads into damp, wet, and very wet roads. The ENet model showed an accuracy of 90.48%. The efficiency of the deep learning models in road surface wetness monitoring was validated using a confusion matrix created with the margin classifier. A total of 300 images per category were used for training, amounting to 1200 in total. A total of 20 testing images were used for road surface classification and 21 for road wetness estimation. The results highlighted the robustness and applicability of SqueezeNet and ENet models in estimating diverse environmental road conditions. Full article

Proton Exchange Membrane Fuel Cells (PEMFCs), recognised for their high efficiency and zero emissions, represent a promising solution for automotive applications. Despite their potential, durability challenges under real-world automotive operating conditions—arising from chemical, mechanical, catalytic, and thermal degradation processes intensified by contaminants—limit their broader adoption. This review aims to systematically assess recent advancements in understanding and modelling PEMFC degradation mechanisms. The article critically evaluates experimental approaches integrated with advanced physicochemical modelling techniques, such as impedance spectroscopy, microstructural analysis, and hybrid modelling approaches, highlighting their strengths and specific limitations. Experimental studies conducted under dynamic, realistic conditions provide precise data for validating these models. The review explicitly compares physics-based, data-driven, and hybrid modelling strategies, discussing trade-offs between accuracy, computational demand, and generalizability. Key findings emphasise that hybrid models effectively balance precision with computational efficiency. Finally, the article identifies apparent research gaps. It suggests future directions, including developing degradation-resistant materials, improved simulation methodologies, and intelligent control systems to optimise PEMFC performance and enhance operational lifespan. Full article

This study aimed to verify the possibility of stabilizing and focusing a plasma column generated by a plasma cutter. The simulation performed by the COMSOL Multiphysics software is based on the actual configuration and geometry of the burner. This article presented a universal computational method based on FEM simulations, focusing on the deflection of the current of electrically charged particles in a magnetic field within the context of a plasma cutting torch. The simulations estimate the optimal shape and positioning of a focused electron beam for various magnetic lens positions and plasma stream energies, revealing that higher initial electron energies lead to a more even beam focus. Among the configurations tested, positioning the cathode 3 mm above the ring-shaped permanent magnet proved most effective, maintaining beam linearity and minimizing electron scattering, making it suitable for practical implementations. Full article

Background: Approximately 10% of patients with small intestine neuroendocrine neoplasms (SI-NENs) present with locally advanced, unresectable symptomatic disease. The present study analyzed the results of debulking of the mesenteric mass in such patients. Methods: Patients operated on for locally advanced SI-NEN disease were identified from the prospective database of the ENETS Center of Excellence Marburg based on the review of imaging results and operative notes. “Locally advanced” was defined as mesenteric disease involving the mesenteric root above the level of the horizontal part of the duodenum and/or extending into the retroperitoneum. Patient characteristics, operations, and outcomes were retrospectively analyzed. Results: 29 of 202 (14%) operated SI-NEN patients (79% male) operated on, with a median age of 63 (46–78) years, had symptomatic locally advanced disease and presented with either abdominal pain (76%) and/or symptoms of obstruction (38%). Imaging revealed a mesenteric mass >10 mm above the level of the pars descendens duodeni in 15 (52%) patients, with tumor-related obstruction of the superior mesenteric vein in 17 (59%) patients. Fourteen (48%) patients had had previous surgery with primary tumor resection (n = 10) or diagnostic or bypass procedures (n = 4). Debulking of the mesenteric mass with (n = 26) or without (n = 2) bowel resection was performed 28 patients; the remaining patient underwent only resection of the ischemic bowel. Median operating time was 262 (156–411) minutes. Four (14%) patients had clinically relevant postoperative complications; one patient died perioperatively. A total of 27/29 (93%) patients reported improvement in preoperative abdominal symptoms. After a median follow-up of 28 (1–142) months, 21 (72%) patients were alive with disease. Conclusions: Debulking of the mesenteric mass in locally advanced symptomatic SI-NENs is a challenging procedure, but most patients benefit in terms of bowel symptoms. Full article

Crack segmentation in concrete bridge structures is a critical task for ensuring safety and durability. This study focuses on evaluating and improving the performance of various deep learning models for crack segmentation, including U-Net, SegNet, ENet, HRNet, FastFCN, and DeepLab V3+. A novel approach is proposed which integrates a probability map generated by an ensemble of classification models as an additional input channel for segmentation models. This method demonstrated significant improvements in segmentation quality, increasing the IoU by up to 25.91% and F1 score by 15.39% compared with baseline models. These improvements were achieved through the use of additional spatial information provided by the probability map, enabling the models to detect cracks more precisely. Additionally, to evaluate the relevance of this approach, the results were compared with YOLO11x-seg, the latest and largest version for segmentation. These findings highlight that integrating auxiliary data channels into neural network architectures holds promise for enhancing segmentation accuracy in real-world engineering applications. The results of this study provide valuable insights for structural engineers and researchers working on automated crack detection, contributing to the development of reliable tools for structural health monitoring. Full article

With the rapid development of lightweight network models and efficient hardware deployment techniques, the demand for real-time semantic segmentation in areas such as autonomous driving and medical image processing has increased significantly. However, realizing efficient semantic segmentation on resource-constrained embedded platforms still faces many challenges. As a classical lightweight semantic segmentation network, ENet has attracted much attention due to its low computational complexity. In this study, we optimize the ENet semantic segmentation network to significantly reduce its computational complexity through structural simplification and 8-bit quantization and improve its hardware compatibility through the optimization of on-chip data storage and data transfer while maintaining 51.18% mIoU. The optimized network is successfully deployed on hardware accelerator and SoC systems based on Xilinx ZYNQ ZCU104 FPGA. In addition, we optimize the computational units of transposed convolution and dilated convolution and improve the on-chip data storage and data transfer design. The optimized system achieves a frame rate of 130.75 FPS, which meets the real-time processing requirements in areas such as autonomous driving and medical imaging. Meanwhile, the power consumption of the accelerator is 3.479 W, the throughput reaches 460.8 GOPS, and the energy efficiency reaches 132.2 GOPS/W. These results fully demonstrate the effectiveness of the optimization and deployment strategies in achieving a balance between computational efficiency and accuracy, which makes the system well suited for resource-constrained embedded platform applications. Full article

Solid recovered fuel (SRF) is highly suited for thermal treatment, but its low bulk density and other physical properties limit the number of compatible energy systems that can effectively process it. This study presents the findings on SRF energy utilisation, focusing on mechanical treatment and a novel approach to its small-scale co-combustion with certified softwood (SW) pellets and catalytic flue gas control. In this study, the processes of certified SRF feedstock characterisation and mechanical treatment were thoroughly examined. Unique SRF pellets of proper mechanical properties were experimentally prepared for real-scale experiments. Mechanical and chemical properties, such as mechanical resilience, toughness, moisture and heating value, were examined and compared with standard SW A1 class pellets. The prepared SRF pellets possessed an energy density of 30.5 MJ∙kg⁻¹, meeting the strict requirements from multiple perspectives. The influence of pelletisation temperature on pellet quality was investigated. It was found that increased resilience and a water content of 1.59% were achieved at a process temperature equal to 75 °C. Moreover, the moisture resilience was found to be significantly better (0.5 vs. 14.23%) compared with commercial SW pellets, while the hardness and durability values were reasonably similar: 40.7 vs. 45.2 kg and 98.74 vs. 98.99%, respectively. This study demonstrates that SRF pellets, with their improved mechanical and energy properties, are a viable alternative fuel, from a technical standpoint, which can be fully utilised in existing combustion units. Full article

Background: Due to complex signal characteristics and distinct individual differences, the decoding of a motor imagery electroencephalogram (MI-EEG) is limited by the unsatisfactory performance of suboptimal traditional models. Methods: A subject-independent model named MSEI-ENet is proposed for multiple-task MI-EEG decoding. It employs a specially designed multi-scale structure EEG-inception module (MSEI) for comprehensive feature learning. The encoder module further helps to detect discriminative information by its multi-head self-attention layer with a larger receptive field, which enhances feature representation and improves recognition efficacy. Results: The experimental results on Competition IV dataset 2a showed that our proposed model yielded an overall accuracy of 94.30%, MF1 score of 94.31%, and Kappa of 0.92. Conclusions: A performance comparison with state-of-the-art methods demonstrated the effectiveness and generalizability of the proposed model on challenging multi-task MI-EEG decoding. Full article

Accurate estimation of pasture yield in grasslands is crucial for the sustainable utilization of pasture resources and the optimization of grassland management. This study leveraged the capabilities of machine learning techniques, supported by Google Earth Engine (GEE), to assess pasture yield in the temperate grasslands of northern China. Utilizing Landsat-8 data, band reflectances, vegetation indexes (VIs), and soil water index (SWI) were extracted from 1000 field samples across Xilingol. These data, combined with field-measured pasture yields, were employed to construct models using four machine learning algorithms: elastic net regression (Enet), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM). Among the models, XGBoost demonstrated the best performance for pasture yield estimation, with a coefficient of determination (R²) of 0.94 and a precision of 76.3%. Additionally, models that incorporated multiple VIs demonstrated superior prediction accuracy compared to those using individual VI, and including soil moisture data further enhanced predictive precision. The XGBoost model was subsequently applied to map the spatial patterns of pasture yield in the Xilingol grassland for the years 2014 and 2019. The estimated average annual pasture yield in the Xilingol grassland was 1042.38 and 1013.49 kg/ha in 2014 and 2019, respectively, showing a general decreasing trend from the northeast to the southwest. This study explored the effectiveness of common machine learning algorithms in predicting pasture yield of temperate grasslands utilizing Landsat-8 data and ground sample data and provided the valuable support for long-term historical monitoring of pasture resources. The findings also highlighted the importance of predictor selection in optimizing model performance, except for the reflectance and vegetation indices characterizing vegetation canopy information, the inclusion of soil moisture information could appropriately improve the accuracy of model predictions, especially for grasslands with relatively low vegetation cover. Full article

Using agricultural waste biomass pyrolysis to produce energy sources and biochar may support local economies in rural areas and enhance sustainability in the agricultural sector, reducing dependence on traditional energy sources and fertilisers. To obtain liquid and gaseous forms of biomass fuel, wheat straw pellets were pyrolysed in a screw reactor at temperatures of 300, 400, 500, 600, and 700 °C. An analysis was conducted to assess the influence of process temperature on the physicochemical composition of the raw material and the resulting biochar, pyrolysis liquid, and synthesis gas. The presence of potentially harmful substances in the biochar, whose addition to soil can improve soil properties, was assessed by quantitatively determining polycyclic aromatic hydrocarbons (PAHs). Similar tests were carried out for pyrolysis fluid. The assessments were based on the standards for the most dangerous PAHs: fluorene, anthracene, fluoranthene, benzo[b]fluorine, benz[a]anthracene, chrysene, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, dibenz[a,h]anthracene, benzo[g,h,i]perylene, and indeno[1,2,3-cd]pyrene. The results indicated that the total content of polycyclic aromatic hydrocarbons in the biochar ranged from 346.81 µg·kg⁻¹ at 300 °C to 1660.87 µg·kg⁻¹ (700 °C). In the pyrolytic fluid, the PAH content ranged from 58,240.7 µg·kg⁻¹ (300 °C) to 101,889.0 µg·kg⁻¹ (600 °C). It was found that the increase in PAH content in both the biochar and the liquid progressed with increasing pyrolysis temperature. After finding a correlation between the increase in the PAH content in biochar and the increase in the content of high-energy gases in the synthesis gas, it was concluded that it is difficult to reconcile the production of PAH-free biochar in the pyrolysis of biomass with obtaining high-energy gas and pyrolysis oil. Full article

Straw mulch quantity is an important indicator in the detection of straw returned to the field in conservation tillage, but there is a lack of large-scale automated measurement methods. In this study, we estimated global straw mulch quantity and completed the detection of straw returned to the field. We used an unmanned aerial vehicle (UAV) carrying a multispectral camera to acquire remote sensing images of straw in the field. First, the spectral index was selected using the Elastic-net (ENET) algorithm. Then, we used the Genetic Algorithm Hybrid Particle Swarm Optimization (GA-HPSO) algorithm, which embeds crossover and mutation operators from the Genetic Algorithm (GA) into the improved Particle Swarm Optimization (PSO) algorithm to solve the problem of machine learning model prediction performance being greatly affected by parameters. Finally, we used the Monte Carlo method to achieve a global estimation of straw mulch quantity and complete the rapid detection of field plots. The results indicate that the inversion model optimized using the GA-HPSO algorithm performed the best, with the coefficient of determination (R²) reaching 0.75 and the root mean square error (RMSE) only being 0.044. At the same time, the Monte Carlo estimation method achieved an average accuracy of 88.69% for the estimation of global straw mulch quantity, which was effective and applicable in the detection of global mulch quantity. This study provides a scientific reference for the detection of straw mulch quantity in conservation tillage and also provides a reliable model inversion estimation method for the estimation of straw mulch quantity in other crops. Full article

In this paper, the dataset is collected from the fluidic muscle datasheet. This dataset is then used to train models predicting the pressure, force, and contraction length of the fluidic muscle, as three separate outputs. This modeling is performed with four algorithms—extreme gradient boosted trees (XGB), ElasticNet (ENet), support vector regressor (SVR), and multilayer perceptron (MLP) artificial neural network. Each of the four models of fluidic muscles (5-100N, 10-100N, 20-200N, 40-400N) is modeled separately: First, for a later comparison. Then, the combined dataset consisting of data from all the listed datasets is used for training. The results show that it is possible to achieve quality regression performance with the listed algorithms, especially with the general model, which performs better than individual models. Still, room for improvement exists, due to the high variance of the results across validation sets, possibly caused by non-normal data distributions. Full article