Search Results (2,284)

Particle accelerator laboratories, which enable world-class research across many scientific fields, depend on the magnets used to manipulate their particle beams for successful operation. The community employs various techniques, typically based on Hall probes and induction sensors/coils, to verify the performance of these accelerator magnets. When the transverse access around a magnet is restricted, conventional Hall probe systems cannot be deployed or require significant modification, while moving wire/coil systems tend to provide information only on the magnetic field’s integral. This research builds upon a vibrating wire setup first commissioned to locate the magnetic center of quadrupole magnets. Scans up to the n = 200 wire harmonic (∼10 kHz drive frequency) were measured to reconstruct the magnetic field across a wire strung through a test magnet. New software was developed to systematically process the many frequency response scans needed for a detailed field reconstruction. This research investigated the speed and precision of the measurement, identifying limitations due to both instrumentation and nonlinear wire behavior. The vibrating wire data agreed with a reference Hall probe scan on the order of 6%; roughly 0.7% RMS error persisted after calibrating the vibrating wire data to the reference scan via scaling factor. Full article

Presenting information to multiple sensory modalities often facilitates or interferes with processing, yet the mechanisms remain unclear. Using a Stroop-like task, the two reported experiments examined how semantic congruency and incongruency in one sensory modality affect processing and responding in a different modality. Participants were presented with pictures and sounds simultaneously (Experiment 1) or asynchronously (Experiment 2) and had to respond whether the visual or auditory stimulus was an animal or vehicle, while ignoring the other modality. Semantic congruency and incongruency in the unattended modality both affected responses in the attended modality, with visual stimuli having larger effects on auditory processing than the reverse (Experiment 1). Effects of visual input on auditory processing decreased under longer SOAs, while effects of auditory input on visual processing increased over SOAs and were correlated with relative processing speed (Experiment 2). These results suggest that congruence and modality both impact multisensory processing. Full article

Data is needed to understand local air quality, reduce exposure, and mitigate the negative impacts on human health. Measuring local air quality often requires a hybrid monitoring approach consisting of the national air monitoring network and one or more networks of air sensors. However, it can be challenging to combine this data to produce a consistent picture of air quality, largely because sensor data is produced in a variety of formats. Users may have difficulty reformatting, performing basic quality control steps, and using the data for their intended purpose. We developed an R-Shiny application that allows users to import text-based air sensor data, describe the format, perform basic quality control, and export the data to standard formats through a user-friendly interface. Format information can be saved to speed up the processing of additional sensors of the same type. This tool can be used by air quality professionals (e.g., state, local, Tribal air agency staff, consultants, researchers) to more efficiently work with data and perform further analysis in the Air Sensor Network Analysis Tool (ASNAT), Google Earth or Geographic Information System (GIS) programs, the Real Time Geospatial Data Viewer (RETIGO), or other applications they already use for air quality analysis and management. Full article

This study introduces YOLOv13-Cone-Lite, an enhanced algorithm based on YOLOv13s, designed to meet the stringent accuracy and real-time performance demands for traffic cone detection in autonomous formula racing cars on enclosed tracks. We improved detection accuracy by refining the network architecture. Specifically, the DS-C3k2_UIB module, an advanced iteration of the Universal Inverted Bottleneck (UIB), was integrated into the backbone to boost small object feature extraction. Additionally, the Non-Maximum Suppression (NMS)-free ConeDetect head was engineered to eliminate post-processing delays. To accommodate resource-limited onboard terminals, we minimized superfluous parameters through structural reparameterization pruning and performed 8-bit integer (INT8) quantization using the TensorRT toolkit, resulting in a lightweight model. Experimental findings show that YOLOv13-Cone-Lite achieves a mAP₅₀ of 92.9% (a 4.5% enhancement over the original YOLOv13s), a frame rate of 68 Hz (double the original model’s speed), and a parameter size of 8.7 MB (a 52.5% reduction). The proposed algorithm effectively addresses challenges like intricate lighting and long-range detection of small objects and offers the automotive industry a framework to develop more efficient onboard perception systems, while informing object detection in other closed autonomous environments like factory campuses. Notably, the model is optimized for enclosed tracks, with open traffic generalization needing further validation. Full article

Phishing is a form of cyber-attack that aims to steal sensitive information by impersonating a trusted entity. To overcome this threat, various artificial intelligence (AI) methods have been developed to improve the effectiveness of phishing detection. This study evaluates three machine learning models, namely Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM), using an open dataset containing phishing and non-phishing URLs. The research process includes data preprocessing stages such as cleaning, normalization, categorical feature encoding, feature selection, and dividing the dataset into training and test data. The trained models are then evaluated using accuracy, precision, recall, F1-score, and comparison score metrics to determine the best model in phishing classification. The evaluation results show that the Random Forest model has the best performance with higher accuracy and generalization of 98.64% compared to Decision Tree which is only 98.37% and SVM 92.67%. Decision Tree has advantages in speed and interpretability but is susceptible to overfitting. SVM shows good performance on high-dimensional datasets but is less efficient in computing time. Based on the research results, Random Forest is recommended as the most optimal model for machine learning-based phishing detection. Full article

In recent years, generative AI and autonomous driving have been highly popular topics. Additionally, with the increasing global emphasis on carbon emissions and carbon trading, integrating autonomous driving technologies that can instantly perceive environ-mental changes with vehicle-based generative AI would enable vehicles to better under-stand their surroundings and provide drivers with recommendations for more energy-efficient and comfortable driving. This study employed You Only Look Once version11 (YOLOv11) for visual detection of the driving environment, integrating it with vehicle speed data received from the OBD-II system. All information is integrated and processed using the embedded Nvidia Jetson AGX Orin platform. For visual detection validation, part of the test set includes standard Taiwanese road signs. Experimental results show that incorporating Squeeze-and-Excitation Attention (SEAttention), into YOLOv11 improves the mAP50–95 accuracy by 10.1 percentage points. Generative AI processed this information in real time and provided the driver with appropriate driving recommendations, such as gently braking, detecting a pedestrian ahead, or warning of excessive speed. These recommendations are delivered through voice output to prevent driver distraction caused by looking at an interface. When a red light or pedestrian is detected, early deceleration is suggested, effectively reducing fuel consumption while also enhancing driving comfort, ultimately achieving the goal of energy-efficient driving. Full article

In Directed Energy Deposition (DED), modeling the molten pool temperature field is crucial for precise temperature control, process optimization, and quality improvement. However, conventional numerical methods suffer from limitations such as high computational costs and poor transferability. This study proposes a physics-informed neural network with dynamic learning rate (DLR-PINN) model, which integrates transfer learning to enable rapid prediction of 3D temperature fields and dimensions of molten pools across process parameters. Its validity is verified by a finite element method (FEM) calibrated via single-track DED experiments. Results show that DLR-PINN exhibits superior convergence and stability compared to traditional PINN. Combined with transfer learning, training efficiency is significantly enhanced, with a single prediction taking only 10 s. Using the FEM as the benchmark, it achieves a mean absolute percentage error (MAPE) of 0.53% for temperature prediction, and MAPE of 3.69%, 2.48%, and 6.96% for molten pool dimension predictions, respectively. Sensitivity analysis of process parameters reveals that scanning speed has a significantly greater regulatory effect on molten pool characteristics than laser power. Additionally, the temperature field of the flat-top heat source is more uniform than that of the Gaussian heat source, which is more conducive to improving printing quality and efficiency. Full article

Multi-object tracking (MOT) algorithms are a key research direction in the field of computer vision. Among them, the joint detection and embedding (JDE) method, with its excellent speed and accuracy performance, has become the current mainstream solution. However, in complex scenes with dense targets or occlusions, the tracking performance of existing algorithms is often limited, especially in terms of unstable identity assignment and insufficient tracking accuracy. To address these challenges, this paper proposes a new multi-object tracking model—the Reparameterized and Global Context Track (RGTrack). This model is based on the Correlation-Sensitive Track (CSTrack) framework and innovatively introduces multi-branch training and attention mechanisms, combined with reparameterized convolutional networks and global attention modules, significantly enhancing the network’s feature extraction ability in complex scenes, especially in ignoring irrelevant information and focusing on key areas. It adopted a multiple association strategy to better establish the association relationship between targets in consecutive frames. Through this improvement, the Reparameterized and Global Context Track can better handle scenes with dense targets and severe occlusions, providing more accurate target identity matching and continuous tracking. Experimental results show that compared with the Correlation-Sensitive Track, the Reparameterized and Global Context Track has significant improvements in multiple key indicators: multi-object tracking accuracy (MOTA) increased by 1.15%, Identity F1 Score (IDF1) increased by 1.73%, and Mostly Tracked (MT) increased by 6.86%, while ID-switched (ID Sw) decreased by 47.49%. These results indicate that the Reparameterized and Global Context Track not only can stably track targets in more complex scenes but also significantly improves the continuity of target identities. Moreover, the Reparameterized and Global Context Track increased the frames per second (FPS) by 51.48% and reduced the model size by 3.08%, demonstrating its significant advantages in real-time performance and computational efficiency. Therefore, the Reparameterized and Global Context Track model maintains high accuracy while having stronger real-time processing capabilities, making it especially suitable for embedded devices and resource-constrained application environments. Full article

As an important traffic and transportation roadway, tunnel engineering is widely used in important fields such as highways, railways, water conservancy, subways and mining. It is limited by complex geological conditions, harsh construction environments and poor robustness of the monitoring system. If the construction process and monitoring method are not properly designed, it will often directly induce disasters such as tunnel deformation, collapse, leakage and rockburst. This seriously threatens the safety of tunnel construction and operation and the protection of the regional ecological environment. Therefore, based on distributed fiber optic sensing technology, the full–cycle spatiotemporally continuous sensing information of the tunnel structure is obtained in real time. Accordingly, the health status of the tunnel is dynamically grasped, which is of great significance to ensure the intrinsic safety of the whole life cycle for the tunnel project. Firstly, this manuscript systematically sorts out the development and evolution process of the theory and technology of structural health monitoring in tunnel engineering. The scope of application, advantages and disadvantages of mainstream tunnel engineering monitoring equipment and main optical fiber technology are compared and analyzed from the two dimensions of equipment and technology. This provides a new path for clarifying the key points and difficulties of tunnel engineering monitoring. Secondly, the mechanism of action of four typical optical fiber sensing technologies and their application in tunnel engineering are introduced in detail. On this basis, a spatiotemporal continuous perception method for tunnel engineering based on DFOS is proposed. It provides new ideas for safety monitoring and early warning of tunnel engineering structures throughout the life cycle. Finally, a high–speed rail tunnel in northern China is used as the research object to carry out tunnel structure health monitoring. The dynamic changes in the average strain of the tunnel section measurement points during the pouring and curing period and the backfilling period are compared. The force deformation characteristics of different positions of tunnels in different periods have been mastered. Accordingly, scientific guidance is provided for the dynamic adjustment of tunnel engineering construction plans and disaster emergency prevention and control. At the same time, in view of the development and upgrading of new sensors, large models and support processes, an innovative tunnel engineering monitoring method integrating “acoustic, optical and electromagnetic” model is proposed, combining with various machine learning algorithms to train the long–term monitoring data of tunnel engineering. Based on this, a risk assessment model for potential hazards in tunnel engineering is developed. Thus, the potential and disaster effects of future disasters in tunnel engineering are predicted, and the level of disaster prevention, mitigation and relief of tunnel engineering is continuously improved. Full article

Visible and infrared thermal imaging are crucial techniques for detecting structural and temperature anomalies in electrical power system equipment. To meet the demand for multimodal infrared/visible light monitoring of target devices, this paper introduces CBAM-YOLOv4, an improved lightweight object detection model, which leverages a novel synergistic integration of the Convolutional Block Attention Module (CBAM) with YOLOv4. The model employs MobileNet-v3 as the backbone to reduce parameter count, applies depthwise separable convolution to decrease computational complexity, and incorporates the CBAM module to enhance the extraction of critical optical features under complex backgrounds. Furthermore, a feature-level fusion strategy is adopted to integrate visible and infrared image information effectively. Validation on public datasets demonstrates that the proposed model achieves an 18.05 frames per second increase in detection speed over the baseline, a 1.61% improvement in mean average precision (mAP), and a 2 MB reduction in model size, substantially improving both detection accuracy and efficiency through this optimized integration in anomaly inspection of electrical equipment. Validation on a representative edge device, the NVIDIA Jetson Nano, confirms the model’s practical applicability. After INT8 quantization, the model achieves a real-time inference speed of 40.8 FPS with a high mAP of 80.91%, while consuming only 5.2 W of power. Compared to the standard YOLOv4, our model demonstrates a significant improvement in both processing efficiency and detection accuracy, offering a uniquely balanced and deployable solution for mobile inspection platforms. Full article

Digital ink Chinese characters (DICCs) written by international students often contain various errors and irregularities, making the recognition of these characters a highly challenging pattern recognition problem. This paper designs a graph convolutional network with agent attention (GCNAA) for recognizing DICCs written by international students. Each sampling point is treated as a vertex in a graph, with connections between adjacent sampling points within the same stroke serving as edges to create a Chinese character graph structure. The GCNAA is used to process the data of the Chinese character graph structure, implemented by stacking Block modules. In each Block module, the graph agent attention module not only models the global context between graph nodes but also reduces computational complexity, shortens training time, and accelerates inference speed. The graph convolution block module models the local adjacency structure of the graph by aggregating local geometric information from neighboring nodes, while graph pooling is employed to learn multi-resolution features. Finally, the Softmax function is used to generate prediction results. Experiments conducted on public datasets such as CASIA-OLWHDB1.0-1.2, SCUT-COUCH2009 GB1&GB2, and HIT-OR3C-ONLINE demonstrate that the GCNAA performs well even on large-category datasets, showing strong generalization ability and robustness. The recognition accuracy for DICCs written by international students reaches 98.7%. Accurate and efficient handwritten Chinese character recognition technology can provide a solid technical foundation for computer-assisted Chinese character writing for international students, thereby promoting the development of international Chinese character education. Full article

Downhole torsional vibrations, especially high-frequency torsional oscillations (HFTOs) and stick–slip phenomena, pose a serious threat to drilling operations, often resulting in tool damage, prolonged non-productive time, and significant cost increases. Traditional monitoring methods cannot promptly capture complex vibration patterns, so there is an urgent need for advanced early warning systems. This study proposes the DS-DW-TimesNet model, which improves the TimesNet framework by incorporating downsampling technology for efficient data compression, dilated convolution that can expand the temporal receptive field, and a learnable weight normalization method that can stabilize the training process, thereby enhancing the capabilities of feature extraction and long-sequence modeling. Verified using field data from the Fuman Oilfield, the results show that in terms of the mean absolute error (MAE) for 210 s predictions, this model is 77.2% and 21.8% lower than LSTM and Informer, respectively, and the inference speed is increased by 78.5% (reaching 48 milliseconds). It can provide reliable 210 s early warning windows for high-frequency torsional oscillations and 150 s early warning windows for stick–slip, exceeding industry standards and helping to improve the safety and efficiency of drilling operations. Full article

To increase the safety and efficiency of underground mining processes, it is important to advance automation. An important part of that is to achieve autonomous material loading using load–haul–dump (LHD) machines. To be able to autonomously load material from a muck pile, it is crucial to first detect and characterize it in terms of spatial configuration and geometry. Currently, the technologies available on the market that do not require an operator at the stope are only applicable in specific mine layouts or use 2D camera images of the surroundings that can be observed from a control room for teleoperation. However, due to missing depth information, estimating distances is difficult. This work presents a novel approach to muck pile detection developed as part of the EU-funded Next Generation Carbon Neutral Pilots for Smart Intelligent Mining Systems (NEXGEN SIMS) project. It uses a stereo camera mounted on an LHD to gather three-dimensional data of the surroundings. By applying a topological algorithm, a muck pile can be located and its overall shape determined. This system can detect and segment muck piles while driving towards them at full speed. The detected position and shape of the muck pile can then be used to determine an optimal attack point for the machine. This sensor solution was then integrated into a complete system for autonomous loading with an LHD. In two different underground mines, it was tested and demonstrated that the machines were able to reliably load material without human intervention. Full article

Enhancer–promoter interactions (EPIs) play a key role in epigenetic regulation of gene expression, dominating cellular identity and functional diversity. Dissecting these interactions is crucial for understanding transcriptional regulatory networks and their significance in cell differentiation, development, and disease. Here, we propose a novel deep learning framework, EPIFBMC (Enhancer-Promoter Interaction prediction with FBMC network) that leverages DNA sequence and genomic features for accurate EPI prediction. The FBMC network consists of three key modules: the Four-Encoding module first encodes the DNA sequence in multiple dimensions to extract key sequence information; then the BESL (Balanced Ensemble Subset Learning) adopts an integrated subset learning strategy to optimize the feature-learning process of positive and negative samples; finally, the MCANet module completes the training of EPI prediction based on a Multi-channel Network. We evaluated EPIFBMC on three cell line datasets (HeLa, IMR90, and NHEK), and validated its generalizability across three independent datasets (K562, GM12878, HUVEC) through cross-cell-line experiments, comparing favorably with state-of-the-art methods. Notably, EPIFBMC balances genomic feature richness and computational complexity, significantly accelerating training speed. Ablation studies identified two key DNA sequence features—positional conservation and positional specificity score—which showed critical predictive value across a benchmark dataset of six diverse cell lines. The computational testing show that EPIFBMC shows excellent performance in the EPI prediction task, providing a powerful tool for decoding gene regulatory networks. It is believed that it will have important application prospects in developmental biology, disease mechanism research, and therapeutic target discovery. Full article

The continuous annealing process is widely used in the production of advanced high-strength steels. However, to tightly regulate the mechanical properties of the steel, precise control of processing parameters is needed. Although some techniques are available to monitor the mechanical properties of the steel on entry and exit to the furnace, monitoring the evolving microstructure of the steel through installation of sensors in the annealing line is extremely challenging due to the high temperature, high speed of the steel strip and limited space in the furnace. This study presents the development and validation of a multifrequency electromagnetic sensor system for real-time monitoring of microstructural transformations in steel during thermal cycling, intended for deployment in a continuous annealing line. Experiments were conducted on austenitic stainless steel to study the signal response to an increase in resistivity without a change in magnetic permeability. Pure nickel was tested to investigate the response to a change in magnetic permeability and the ferromagnetic-to-paramagnetic transition at its Curie temperature. A ferritic stainless steel was also tested to assess the performance of the system for high-temperature ferromagnetic materials and a higher-temperature ferromagnetic-to-paramagnetic transition. The tests indicate a strong response to material resistivity and permeability changes, with complementary information from different frequencies. Test results are supplemented by a finite element modelling study into the effect of a change in frequency and permeability on sensor response, with a discussion on the implications of experimental and modelling results for future applications. The results show that the developed system has the potential to characterise thermally induced changes in steels, establishing proof of concept for non-destructive, high-temperature electromagnetic sensing in steel processing and setting the foundation for further industrial deployment in phase and recrystallisation monitoring. Full article