Search Results (405)

The transition toward sustainable and resilient new-type power systems requires robust diagnostic frameworks for terminal power supply units to ensure continuous grid stability. To ensure the resilience of modern power systems, this paper proposes a multi-source domain deep Transfer Learning method for the abnormal condition diagnosis of low-voltage distribution nodes within a cloud-edge collaborative framework. This approach integrates feature selection based on the Categorical Boosting (CatBoost) algorithm with a hybrid architecture combining a Convolutional Neural Network (CNN) and a Residual Network (ResNet). Additionally, it utilizes a multi-loss adaptation strategy consisting of Multi-Kernel Maximum Mean Difference (MK-MMD), Local Maximum Mean Difference (LMMD), and Mean Squared Error (MSE) to effectively bridge domain gaps and ensure diagnostic consistency. By balancing global commonality with local adaptation, the framework optimizes resource efficiency, reducing collaborative training time by 19.3%. Experimental results confirm that the method effectively prevents equipment failure, achieving diagnostic accuracies of 98.29% for low-voltage anomalies and 88.96% for three-phase imbalance conditions. Full article

High-precision short-term photovoltaic (PV) power prediction has become a critical technology in ensuring grid accommodation capacity, optimizing dispatching decisions, and enhancing plant economic benefits. This paper proposes a long short-term memory (LSTM)-based short-term PV power prediction method with the genetic algorithm (GA)-optimized adaptive fusion of space-based cloud imagery and ground-based meteorological data. The effective integration of satellite cloud imagery is conducted in the PV power prediction system, and the proposed method addresses the issues of low accuracy, poor robustness, and inadequate adaptation to complex weather associated with using a single type of meteorological data for PV power prediction. The multi-source heterogeneous data are preprocessed through outlier detection and missing value imputation. Spearman correlation analysis is employed to identify meteorological attributes highly correlated with PV power output. A dedicated dataset compatible with LSTM algorithm-based prediction models is constructed. An LSTM prediction model with a GA algorithm-based adaptive multi-source heterogeneous data fusion method is proposed, and the ability to construct a precise short-term PV power prediction model is demonstrated. Experimental results demonstrate that the proposed method outperforms single-source LSTM, single-source CNN-LSTM, and dual-source CNN-Transformer models in prediction accuracy, achieving an RMSE of 0.807 kWh and an MAPE of 6.74% on a critical test day. The proposed method enables real-time precision forecasting for grid dispatch centers and lightweight edge deployment at PV plants, enhancing renewable energy integration while effectively mitigating grid instability from power fluctuations. Full article

To achieve low-cost and flexible wheel angles measurement, we propose a novel strategy that integrates wheel segmentation network with 3D vision. In this framework, a semantic segmentation network is first employed to extract the wheel rim, followed by angle estimation through ICP-based point cloud registration. Since wheel rim extraction is closely tied to angle computation accuracy, we introduce APCS-SwinUnet, a segmentation network built on the SwinUnet architecture and enhanced with ASPP, CBAM, and a hybrid loss function. Compared with traditional image processing methods in wheel alignment, APCS-SwinUnet delivers more accurate and refined segmentation, especially at wheel boundaries. Moreover, it demonstrates strong adaptability across diverse tire types and lighting conditions. Based on the segmented mask, the wheel rim point cloud is extracted, and an iterative closest point algorithm is then employed to register the target point cloud with a reference one. Taking the zero-angle condition as the reference, the rotation and translation matrices are obtained through point cloud registration. These matrices are subsequently converted into toe and camber angles via matrix-to-angle transformation. Experimental results verify that the proposed solution enables accurate angle measurement in a cost-effective, simple, and flexible manner. Furthermore, repeated experiments further validate its robustness and stability. Full article

As core infrastructure of power transmission networks, power towers require high-precision 3D models, which are critical for intelligent inspection and digital twin applications of power transmission lines. Traditional reconstruction methods, such as LiDAR scanning and oblique photogrammetry, suffer from issues including high operational risks, low modeling efficiency, and loss of fine details. To address these limitations, this paper proposes a 3D Gaussian Splatting (3DGS)-based method for power tower 3D reconstruction to enhance reconstruction efficiency and detail preservation capability. First, a multi-view data acquisition scheme combining “unmanned aerial vehicle + oblique photogrammetry” was designed to capture RGB images acquired by Unmanned Aerial Vehicle (UAV) platforms, which are used as the primary input for 3D reconstruction. Second, a sparse point cloud was generated via Structure from Motion. Finally, based on 3DGS, Gaussian model initialization, differentiable rendering, and adaptive density control were performed to produce high-precision 3D models of power towers. Taking two typical power tower types as experimental subjects, comparisons were made with the oblique photogrammetry + ContextCapture method. Experimental results demonstrate that 3DGS not only achieves high model completeness (with the reconstructed model nearly indistinguishable from the original images) but also excels in preserving fine details such as angle steels and cables. Additionally, the final modeling time is reduced by over 70% compared to traditional oblique photogrammetry. 3DGS enables efficient and high-precision reconstruction of power tower 3D models, providing a reliable technical foundation for digital twin applications in power transmission lines. By significantly improving reconstruction efficiency and reducing operational costs, the proposed method supports sustainable power infrastructure inspection, asset lifecycle management, and energy-efficient digital twin applications. Full article

Establishing digital twin scenes facilitates the understanding of geospatial phenomena, representing a significant research focus for GIS scientists and engineers. However, current research on digital twin scenes modeling relies on manual intervention or the overlay of static models, resulting in low modeling efficiency and poor standardization. To address these challenges, this paper proposes a knowledge graph-guided and multimodal data fusion-driven rapid modeling method for digital twin scenes, using bridge tower construction as an illustrative example. We first constructed a knowledge graph linking the three domains of “event-object-data” in bridge tower construction. Guided by this graph, we designed a knowledge graph-guided multimodal data association and fusion algorithm. Then a rapid modeling method for bridge tower construction scenes based on dynamic data was established. Finally, a prototype system was developed, and a case study area was selected for analysis. Experimental results show that the knowledge graph we built clearly captures all elements and their relationships in bridge tower construction scenes. Our method enables precise fusion of 5 types of multimodal data: BIM, DEM, images, videos, and point clouds. It improves spatial registration accuracy by 21.83%, increases temporal fusion efficiency by 65.6%, and reduces feature fusion error rates by 70.9%. Local updates of the 3D geographic scene take less than 30 ms, supporting millisecond-level digital twin modeling. This provides a practical reference for building geographic digital twin scenes. Full article

The thermal forcing of the Tibetan Plateau (TP) significantly influences the Asian summer monsoon. However, its interaction with cloud feedbacks remains unclear due to the limitations of synchronous analysis and traditional cloud classification over the TP. By applying an improved cloud-classification algorithm—which integrates cloud microphysical properties to improve low-cloud detection—to CERES data (2001–2023), we generated a long-term cloud-type dataset. Combined with ERA5 reanalysis data, we systematically analyzed the trends and lead–lag relationships among cloud vertical structure, surface radiation, cloud radiative forcing (CRF), heat fluxes, snowfall, and the TP Monsoon Index (TPMI). Results indicate a vertical cloud redistribution over the TP, with high cloud cover (HCC) decreasing and low cloud cover (LCC) increasing. HCC is strongly synchronized with snowfall and significantly affects surface radiation, while net CRF and sensible heat flux show delayed responses, peaking when HCC leads by about one month. A composite analysis of winter low-HCC events reveals that reduced HCC suppresses snowfall, weakens net CRF, and reduces sensible heat flux after approximately 1–2 months, while the TPMI shows a significant response around month zero. These findings highlight the key role of cloud–radiation–snowfall interactions in modulating TP thermal forcing. Full article

Although RGB images contain rich details, they lack 3D depth information. Depth data, while providing spatial positioning, is often affected by noise and suffers from sparsity or missing data at key feature points, leading to low accuracy and high computational complexity in traditional visual localization. To address this, this paper proposes a high-precision, sub-pixel-level localization method for workpiece feature points based on RGB-D data fusion. The method specifically targets two types of localization objects: planar corner keypoints and sharp-corner keypoints. It employs the YOLOv10 model combined with a Background Misdetection Filtering Module (BMFM) to classify and identify feature points in RGB images. An improved Prewitt operator (using 5 × 5 convolution kernels in 8 directions) and sub-pixel refinement techniques are utilized to enhance 2D localization accuracy. The 2D feature boundaries are then mapped into 3D point cloud space based on camera extrinsic parameters. After coarse error detection in the point cloud and local quadric surface fitting, 3D localization is achieved by intersecting spatial rays with the fitted surfaces. Experimental results demonstrate that the proposed method achieves a mean absolute error (MAE) of 0.17 mm for localizing flat, free-form, and grooved components, with a maximum error of less than 0.22 mm, meeting the requirements of high-precision industrial applications such as precision manufacturing and quality inspection. Full article

This paper presents the development and cloud deployment of a system for the segmentation of electromyographic (EMG) signals oriented toward muscle fatigue monitoring in the biceps and triceps. A dataset of 30 subjects was used, resulting in 120 EMG and gyroscope files containing between four and six strength exercise series each. After a quality assessment, approximately 80% of the signals (95 files) were classified as level 1 or 2 and considered suitable for segmentation and subsequent analysis. A near real-time segmentation algorithm was designed based on signal envelopes, sliding windows, and quantile thresholds, complemented with discrete wavelet transform (DWT) filtering. Using EMG alone, segmentation accuracy reached 83% for biceps and 54% for triceps; after incorporating DWT preprocessing, accuracy increased to 87.5% and 71%, respectively. By exploiting the gyroscope’s X-axis signal as a low-noise reference, the optimal configuration achieved an overall accuracy of 80%, with 83.3% for biceps and 76.2% for triceps. The prototype was deployed on Amazon Web Services (AWS) using EC2 instances and SQS queues, and its computational cost was evaluated across four server types. On a t2.micro instance, the maximum memory usage was approximately 219 MB with a dedicated CPU and a maximum processing time of 0.98 s per signal, demonstrating the feasibility of near real-time operation under conditions with limited resources. Full article

Grasslands occupy 24% of the Earth’s surface. In most areas of the world these are either destroyed, fragmented or converted into cultivated fields. In Africa, their biodiversity is still insufficiently known. This study reports on the avian assemblages associated with grasslands in South African Highveld and Lesotho Drakensberg. Special attention was paid to the species richness, diversity, and population densities and dominance of particular species. Birds were counted by means of the Line Transect Method in three distinguished grassland types: Dry Cymbopogon-Themeda Grassland (transect length: 28 km), Wet Cymbopogo-Themeda Grassland (27 km) km, and Mountain Themeda-Festuca Grassland (31 km). In total, 86 bird species were recorded. While cumulative dominance was similar between the Dry and Wet Grassland (61–65%), these two were much different from that in the Mountain Grassland (46%). However the dominance index was similar in all three grassland types compared (0.25–0.33). Only one species, the long-tailed widow Euplectes orix was a common dominant species for all three grassland types. African stonechat, wing-snapping cisticola Cisticola ayresii, Levaillant’s cisticola Cisticola tinniens and yellow bishop Euplectes capensis were dominant only in the Mountain Grassland; northern black korhaan Afrotis afroides and the eastern clapper lark Mirafra fasciolata—only in the Dry and Wet Grassland; ostrich Struthio camelus, cloud cisticola Cisticola textrix, African quailfinch Ortygozpiza atricollis and pied starling Spreo bicolor—only in the Dry Grassland, while the helmeted guineafowl Numida meleagris, zitting cisticola Cisticola juncidis and African pipit Anthus cinnamomeus—only in the Wet Grassland. Despite these obvious differences in dominance and population densities of species, Diversity and evenness indices were similar in all three grassland types. Shannon’s Diversity Index (H′) varied between 1.22 and 1.35; Simpson Diversity Index between 0.91 and 0.94, while Pielou’s Evenness Index (J′) varied between 0.33 and 0.36. However, Sørensen Similarity Index between the three grassland types was low, ranging between 0.07 and 0.26. Proportions of ecological guilds were similar in the Dry and Wet Grassland but differed from mountain Grassland. In comparison with other tropical grassland, avian communities in southern Africa are characterized by higher species richness and higher its variance between particular grassland types. Full article

Low clouds significantly influence Earth’s radiation budget, but their climate feedback remains highly uncertain due to complex interactions with meteorological conditions across spatial and temporal scales. The cloud controlling factor framework is widely used to link meteorological variables with cloud properties. However, most studies assume a static, linear relationship, potentially obscuring the timescale-dependent responses. In this study, we apply the Ensemble Empirical Mode Decomposition method to ISCCP-H cloud observations and ERA5 data (1987–2016) to isolate low cloud amount across multiple intrinsic timescales and trends over global land and ocean. The trends show a nonlinear increase in stratocumulus (Sc) and a significant nonlinear decline in cumulus (Cu), while stratus (St) exhibits weaker trends. We categorize timescales short (≤1 year) for annual variations, medium (1–8 years) for interannual variability such as ENSO, and long (>8 years) for decadal and longer-term climate changes. It is found that Sc and Cu over land are primarily influenced by near-surface heating, while sea surface temperature and surface sensible heat flux (SHF) dominate over ocean at short timescales. SHF becomes dominant over land at medium timescales, largely reflecting ENSO-related induced surface anomalies. At long timescales, atmospheric stability and wind speed influence continental clouds, while SHF remains dominant over ocean. Trend components reveal that Sc and Cu are most sensitive to temperature changes, whereas St responds to mid-level humidity over ocean and SHF over land. These findings underscore the importance of timescale-dependent cloud–meteorology relationships to improve cloud parameterizations and reduce climate projection uncertainties. Overall, our results demonstrate that low cloud variability and trends cannot be explained by a single linear mechanism but instead arise from distinct meteorological controls that change across timescales, cloud types, and surface regimes. Full article

LiDAR is used in autonomous driving for navigation, obstacle avoidance, and environment mapping. However, adverse weather conditions introduce noise into sensor data, potentially degrading the performance of perception algorithms and compromising the safety and reliability of autonomous driving systems. Hence, in this paper, we investigate the limitations of LiDAR perception algorithms in adverse weather conditions, explore ways to mitigate the effects of noise, and propose future research directions to achieve all-weather autonomy with LiDAR sensors. Using real-world datasets and synthetically generated dense fog, we characterize the noise in adverse weather such as snow, rain, and fog; their effect on sensor data; and how to effectively mitigate the noise for tasks like object detection, localization, and SLAM. Specifically, we investigate point cloud filtering methods and compare them based on their ability to denoise point clouds, focusing on processing time, accuracy, and limitations. Additionally, we evaluate the impact of adverse weather on state-of-the-art 3D object detection, localization, and SLAM methods, as well as the effect of point cloud filtering on the algorithms’ performance. We find that point cloud filtering methods are partially successful at removing noise due to adverse weather, but must be fine-tuned for the specific LiDAR, application scenario, and type of adverse weather. 3D object detection was negatively affected by adverse weather, but performance improved with dynamic filtering algorithms. We found that heavy snowfall does not affect localization when using a map constructed in clear weather, but it fails in dense fog due to a low number of feature points. SLAM also failed in thick fog outdoors, but it performed well in heavy snowfall. Filtering algorithms have varied effects on SLAM performance depending on the type of scan-matching algorithm. Full article

Atrial fibrillation (AF) is the most common type of heart rhythm disorder worldwide. Early recognition of brief episodes of atrial fibrillation can provide important diagnostic information and lead to prompt treatment. AF is mainly characterized by an irregular heartbeat. Today, many personal devices such as smartphones, smartwatches, smart rings, or small wearable medical devices can detect heart rhythm. Sensors can acquire different types of heart-related signals and extract the sequence of inter-beat intervals, i.e., the instantaneous heart rate. Various algorithms, some of which are very complex and require significant computational resources, are used to recognize AF based on inter-beat intervals (RR). This study aims to verify the possibility of using neural networks algorithms directly on a microcontroller connected to sensors for AF detection. Sequences of 25, 50, and 100 RR were extracted from a public database of electrocardiographic signals with annotated episodes of atrial fibrillation. A custom 1D convolutional neural network (1D-CNN) was designed and then validated via a 5-fold subject-wise split cross-validation scheme. In each fold, the model was tested on a set of 3 randomly selected subjects, which had not previously been used for training, to ensure a subject-independent evaluation of model performance. Across all folds, all models achieved high and stable performance, with test accuracies of 0.963 ± 0.031, 0.976 ± 0.022, and 0.980 ± 0.023, respectively, for models using 25 RR, 50 RR, and 100 RR sequences. Precision, recall, F1-score, and AUC-ROC exhibited similarly high performance, confirming robust generalization across unseen subjects. Performance systematically improved with longer RR windows, indicating that richer temporal context enhances discrimination of AF rhythm irregularities. A complete Edge AI prototype integrating a low-power ECG analog front-end, an ARM Cortex M7 microcontroller and an IoT transmitting module was utilized for realistic tests. Inferencing time, peak RAM usage, flash usage and current absorption were measured. The results obtained show the possibility of using neural network algorithms directly on microcontrollers for real-time AF recognition with very low power consumption. The prototype is also capable of sending the suspicious ECG trace to the cloud for final validation by a physician. The proposed methodology can be used for personal screening not only with ECG signals but with any other signal that reproduces the sequence of heartbeats (e.g., photoplethysmographic, pulse oximetric, pressure, accelerometric, etc.). Full article

Oil film cavitation in the friction pairs of wet clutches significantly compromises transmission stability and component durability. This study investigates the cavitation evolution across three microtexture types—hexahedral, cylindrical, and hemispherical—with texture ratios ranging from 3.205% to 12.917% and a constant depth of 0.0564 mm, under a 6000 rpm operating condition. A finite element model of the oil film was established to analyze the cavitation volume fraction, pressure field, and gas-phase mass transfer rate. The numerical simulations were complemented by visualization experiments, where high-speed imaging (550–1050 rpm) captured the cavitation bubble dynamics, and the transmitted torque was measured. The results indicate that microtexture parameters profoundly influence cavitation intensity. Hemispherical textures with a 6.41% texture ratio yielded the highest cavitation volume fraction (0.020215), substantially exceeding that of hexahedral textures (0.0015197). Cavitation initiates within the texture dimples, with hemispherical geometries facilitating its diffusion into non-textured regions. A threshold effect of the texture ratio was identified, where cavitation intensity peaks at 6.41% but diminishes at 12.917%, attributable to flow homogenization. Optimized designs can effectively suppress cavitation: either increasing the texture depth or adopting a high texture ratio (>45%) with hexahedral or cylindrical geometries reduces the pressure drop in low-pressure zones by over 30%. Experimental validation confirmed that an increased texture ratio reduces torque by 20%, correlating with the shrinkage of the oil film at the outer diameter. High-speed imaging revealed a periodic cavitation evolution, with the collapse of sheet-to-cloud cavitation occupying 15.2% of the cycle, which aligns with the simulated peak in mass transfer at t = 0.003 s. In conclusion, cavitation can be effectively controlled by optimizing the texture ratio, depth, and geometry to maintain a stable oil film pressure gradient. This study provides a theoretical foundation for the microtexture design of wet clutches, thereby enhancing their reliability in power-shift applications. Full article

A game-changing strategy for increasing crop productivity while preserving vital resources is precision agriculture. The development of cloud computing and the Internet of Things (IoT) has made it possible and efficient to monitor soil and environmental data in real time. In order to monitor temperature, soil moisture, humidity, and light intensity, this work proposes an inexpensive, IoT-enabled smart agriculture system that uses low-cost sensors. The real-time data is wirelessly transmitted by an ESP32 edge computing device and stored and analyzed on cloud platforms like Firebase or ThingSpeak. A rule-based algorithm generates alerts when sensor values surpass predefined thresholds, enabling prompt and informed decision-making. Field experiments reveal that the proposed system is accurate, economical, and energy-efficient, making it ideal for automation and remote monitoring in precision agriculture. A user-friendly dashboard allows farmers to easily visualize data trends and receive timely notifications. The system supports scalability and can be adapted to different crop types and soil conditions with minimal effort. Moreover, by optimizing water and resource usage, the system contributes to sustainable farming practices and environmental conservation. This deployable solution offers a practical and affordable pathway for small- and medium-sized farmers to adopt smart agriculture technologies and improve crop yield outcomes efficiently. Full article

In this paper, we present the design, prototyping, and working of a cost-effective, energy-efficient, and scalable air quality monitoring system (AQMS), enabled by a Low-power, long-Range Wide-Area Network (LoRaWAN), an Internet of Things (IoT) technology designed to provide connectivity for massive machine-type communication applications. The growing threat of air pollution necessitates outdoor and mobile environmental monitoring systems to provide real-time, location-specific data, which is unfortunately not possible using fixed monitoring devices. For our AQMS, we have developed two custom-built sensor nodes. The first node is equipped with a Nucleo-WL55JC1 microcontroller and sensors to measure temperature, humidity, and carbon dioxide (CO₂), while the other node is equipped with an Arduino MKR WAN 1310 controller with sensors to measure carbon monoxide (CO), ammonia (NH₃), and particulate matter (PM2.5 and PM10). These sensor nodes connect to a WisGate Edge LoRaWAN gateway, which aggregates and forwards the sensor data to The Things Network (TTN) for processing and cloud storage. The final visualization is handled via the Ubidots IoT platform, allowing for real-time visualization of environmental data. Besides environmental data, we were able to acquire a received signal strength indicator, signal-to-noise ratio, as well as a frame counter, which shows the number of packets received by the gateway. We performed laboratory testing, which confirmed reliable communication, with a packet delivery rate of 98% and a minimal average latency of 2.5 s. Both nodes operated efficiently on battery power, with the Nucleo-WL55JC1 consuming an average of 20 mA in active mode, while the Arduino MKR WAN 1310 operated at 15 mA. These values ensured extended operation for remote deployment. The system’s low power consumption and modular architecture make it viable for smart city applications and large-scale deployments in resource-constrained areas. Full article