Search Results (2,057)

Fault diagnosis in modern power systems increasingly depends on unstructured operation and maintenance (O&M) logs, yet real-world logs are often small in scale and highly imbalanced across fault types, which degrades the generalizability of standard neural models. This paper proposes a graph-augmented diagnostic framework that integrates imbalance-aware knowledge extraction with interpretable reasoning. The framework consists of three stages: (1) domain adaptation of a BERT–BiLSTM–CRF NER model and a BERT–MLP RE model using an imbalance-aware training recipe that combines Low-Rank Adaptation (LoRA), a mixed focal–range loss, and undersampling; (2) construction of a power-system knowledge graph that organizes extracted entities and relations (e.g., fault devices, abnormal phenomena, causes, and handling measures); and (3) a graph-augmented assistant agent that reuses the NER model as a graph-aware retriever within a retrieval-augmented generation (RAG) architecture to support contextualized and interpretable diagnostic reasoning. Experiments on 3921 real-world fault-processing logs show consistent gains: NER reaches 92.0% accuracy and 71.3% Macro-F1 (vs. 80.3% and 63.2%), and RE achieves 88.0% accuracy and 70.1% F1 (vs. 82.1% and 60.4%), while reducing average training time per epoch by about 18%. These results demonstrate an efficient and practical path toward robust log-based fault diagnosis under scarce and imbalanced data. Full article

Ensuring infrared images are of super-high resolution is crucial for enhancing thermal imaging systems’ visual perception, yet existing methods struggle to recover sharp edges and textual details. Therefore, in this study, we aimed to address the following issues: over-smoothed edges, distorted radiometric contrast in diffusion-based approaches, and scanning artifacts introduced by efficient state-space models like Mamba. We propose a novel edge-guided diffusion framework named EGDM-IRSR. Its core methodology integrates a multi-modal scanning mechanism employing complementary scan paths with content-aware modulation to mitigate directional artifacts, along with an edge guidance branch with learnable direction-aware convolutions, complemented by edge-frequency composite loss. Extensive experiments conducted on public benchmarks demonstrate that our method significantly outperforms state-of-the-art alternatives in quantitative metrics and exhibits superior visual fidelity by effectively preserving edges and fine structures. Ablation studies validate the effectiveness of each proposed component. We conclude that EGDM-IRSR provides a more robust and detail-enriched solution for acquiring super-resolution infrared images by synergistically integrating edge guidance with enhanced sequential modeling. Full article

A state of terminal stagnation is often reached by software projects despite the presence of advanced tools, and these occurrences are defined within this study as software engineering deadpoints, where the cost of system recovery is frequently found to be higher than the actual value of the software. While many factors are seen to lead toward project failure, it is suggested by the evidence that technical debts are the main cause of such failures. A significant number (23.5%) of these fatal issues is created during the early architectural phases of development, and it is noted that these problems often remain hidden until they become unrecoverable. The data collected during this research show that projects facing technical obstacles (Recovery Score: 4.24) are much harder to save than those suffering with process obstacles (Recovery Score: 5.38). It was also observed that a steady reluctance to refactor old logic and an excessive number of code revisions are seen as the most reliable signs that a project is approaching a point of no return. Because these warning signs are often overlooked by management, the eventual failure of the system is often viewed as an unexpected event rather than a predictable outcome of poor early choices. By defining these terminal states, this work provides those in leadership roles with a method to differentiate between minor delays and total failure, thereby assisting teams in avoiding the heavy economic losses associated with unproductive development paths. Full article

Slow-moving landslides can trigger severe disasters when activated by earthquakes, torrential rains, or typhoons. Early detection is crucial for mitigating loss of life and property damage. Interferometric Synthetic Aperture Radar (InSAR) technology is among the most effective techniques for detecting slow-moving landslides, though its accuracy can be further improved through integration with optical imagery and Digital Elevation Models (DEM). Current machine learning approaches that combine InSAR and optical data suffer from limited efficiency, poor transferability, and challenges in regional-scale application. To address these limitations, this study proposes a multimodal dual-path network that integrates InSAR products with textural information from optical imagery to detect slow-moving landslides. One path processes InSAR deformation rates and topographic factors, while the other incorporates texture information and auxiliary data. Together, these paths extract semantic information from high-dimensional spatial features and condense it into low-dimensional representations. A pyramid pooling module is employed to capture multi-scale features during low-level semantic extraction. For feature fusion, a rate-constrained adaptive module is introduced to enhance the contribution of deformation rates to slow-moving landslides. According to the results, the proposed method improves the F1-score for landslide detection by 6% compared to using InSAR products alone. These results provide reliable technical support for regional landslide inventory compilation and disaster management, as well as new insights for regional-scale surveys in slow-moving landslide-prone areas. Full article

With the widespread adoption of container technology, its shared kernel architecture has made abnormal file access behavior a key precursor to container escape and lateral attacks, necessitating precise and efficient runtime detection mechanisms. However, existing monitoring methods typically suffer from issues such as insufficient granularity in data collection, limited path semantic modeling capabilities, and low anomaly detection accuracy. To address these challenges, this paper proposes an eBPF-based method for detecting abnormal file access in containers. A lightweight kernel-level monitoring mechanism is constructed to capture access behavior in real time at the system call level, effectively enhancing both the granularity of data collection and the completeness of context. At the feature modeling layer, a multimodal path semantic representation method is designed, combining risk-layer rules and semantic vectorization strategies to enhance the hierarchical expression of path structures and improve context modeling ability. In the detection layer, an attention-enhanced autoencoder model is introduced, achieving high-precision identification of abnormal access behavior and low false-positive monitoring under unsupervised conditions through a path segment attention mechanism and weighted reconstruction loss function. Experiments in real container environments show that the proposed method achieves a recall rate of $82.0\%$, a false-positive rate of $0.79\%$, and a Matthews correlation coefficient of 0.852, significantly outperforming mainstream unsupervised detection methods such as Isolation Forest, One-Class SVM, and Local Outlier Factor. These results verify the advantages of the proposed method in terms of detection accuracy, real-time performance, and system friendliness, providing an efficient and feasible solution for enhancing the detection of unknown attacks in container runtimes. Full article

We developed a Bluetooth Low Energy-based indoor asset-tracking system oriented toward elderly care and cultural heritage stewardship. The system stabilizes the noisy received signal strength indicator using a Kalman filter, adapts a logarithmic path loss model to local attenuation via dynamic calibration, and estimates positions with an inverse distance weighted centroid. Built on inexpensive beacons and commodity gateways, it supports real-time updates and map-based visualization while remaining easy to deploy and scale across rooms and facilities. We validate the pipeline in a laboratory grid and discuss applicability to workflows such as geofenced reminders, caregiver situational awareness, and collection movement oversight, offering an affordable, interoperable path to reliable indoor tracking for care institutions, museums, and smart buildings. Full article

Deep learning has demonstrated outstanding potential for the spectral super-resolution (S²R) reconstruction of multispectral images (MSIs). However, it is still a challenge to alleviate spectral distortion during S²R reconstruction. Given the superiority of a graph, a graph-regularized double-path interactive S²R network (GDIS²Net) consisting of two parallel branches is proposed to reconstruct hyperspectral images (HSIs) from MSIs. An interactive residual module is carefully schemed as the backbone of the S²R network to facilitate the feature interaction between the two branches, while an enhanced residual module is constructed for further feature fusion. In addition, a new loss function considering the spectral continuity is proposed to optimize the proposed GDIS²Net. Experimental analyses show that the proposed GDIS²Net outperforms state-of-the-art methods on both simulated and real datasets. Full article

To address the tendency of the traditional Children’s Drawing Development Optimization (CDDO) algorithm to fall into local optima and converge slowly in global optimization and fire-field robot path planning, this study proposes a Multi-Strategy Enhanced Children’s Drawing Development Optimization (MECDDO) algorithm. The algorithm achieves performance improvements through three core strategies: (1) an adaptive cooperative search strategy that integrates information from the global best, worst, and random individuals and guides updates via dynamic weighting, expanding the exploration of the solution space; (2) a multi-strategy adaptive selection mechanism that constructs a pool of four differentiated strategies and dynamically adjusts selection probabilities based on strategy success rates, balancing exploration and exploitation; and (3) a global-optimum guided boundary repair strategy that reduces the loss of high-quality information from out-of-bounds solutions, enhancing local exploitation efficiency. Experiments on the CEC2017 benchmark suite demonstrate that MECDDO achieves outstanding performance across 30-, 50-, and 100-dimensional spaces. Statistical significance was evaluated using the Friedman test and Wilcoxon signed-rank test at a 0.05 significance level. The Friedman test mean rankings (M.R.) are 1.63, 2.20, and 2.70, respectively, consistently outperforming traditional CDDO (M.R. = 9.83, 9.93, 9.73, ranked 10th). Applied to mobile robot path planning, MECDDO achieves an average path length of 27.95483 in 20 × 20 grid environments (rank 1), shortening paths by 8.83% compared with CDDO (30.66212, rank 10), and 61.15516 in 40 × 40 grids (rank 1), reducing paths by 37.19% versus CDDO (97.20336, rank 9), providing trajectories free of redundant turns and convergence speeds 2–3 times faster than competing algorithms. These results validate MECDDO’s significant advantages in numerical optimization accuracy and practical robot path planning. Full article

To address the problems of severe fruit occlusion, large variations in target scale, and many small-scale goals being overlooked in the recognition of trellised watermelons under complex agricultural scenarios, this study proposes an improved RT-DETR-based detection model, termed RT-DETR-Watermelon. A context-guided (CG) module is embedded into the backbone network. A dedicated P2 detection layer is added to enhance the model’s sensitivity to small objects. A scale sequence feature fusion (SSFF) module and a triple feature encoder (TFE) module are introduced into the model to improve the model’s capability to detect targets at multiple scales. The original bounding box regression loss is replaced with MPDIoU (Multiple Path Distance Intersection over Union) loss, which accelerates model convergence and improves localization precision. Finally, the number of channels is adjusted to reduce parameter count, computational complexity, and storage size. The experimental results show that, compared with the original RT-DETR model, the proposed RT-DETR-Watermelon model increases precision, recall, and mean Average Precision (mAP@0.5) by 0.4, 1.8, and 1.0 percentage points, while reducing the number of parameters, computational cost, and model size by 53.5%, 23.5%, and 53.2%, respectively. Full article

This study investigates radio-frequency signal propagation in underground metro tunnels with a focus on distributed antenna system (DAS) deployment. Deterministic simulations were performed using Altair WinProp 2024.1 (ProMan) with a 3D ray-tracing engine (GO + UTD) at 2.4 GHz in a reinforced concrete tunnel model of 900 m length. Two antenna configurations (B3: 8 dBi directional; B8: 5 dBi wide-beam) were evaluated under identical geometric and material conditions. Results show that path loss varies from 42 to 65 dB over 850 m, with estimated attenuation exponents lower than free-space values due to quasi-waveguide effects. The B3 configuration provides higher near-field received power (up to −7.5 dBm) but exhibits stronger attenuation over long distances. In contrast, the B8 configuration ensures a more uniform spatial power distribution and a reduced path-loss growth rate beyond 500 m. The findings confirm that antenna radiation pattern significantly influences underground communication performance and demonstrate the engineering suitability of distributed antenna systems for stable metro tunnel coverage. Full article

Background: Balance impairment and falls are a major health concern in older adults. Beyond vestibular and visual factors, growing evidence indicates that age-related hearing loss contributes to postural instability through altered multisensory integration. However, interventions addressing the interaction between auditory input and postural control remain limited. This study examined whether integrating Taopatch^® nanotechnology, based on localized photobiomodulation, into conventional hearing aids could influence postural control in individuals with hearing loss. Methods: Forty experienced hearing aid users (mean age 77.3 ± 15.6 years) completed five postural assessments using a SensorMedica^® baropodometric platform. Four sessions employed a placebo patch identical in appearance to the active device, and the fifth used Taopatch^®. Static and stabilometric parameters were analyzed under open- and closed-eye conditions. Results: Significant improvements were observed with the Taopatch^®-integrated device. Sway path length (−8%, p = 0.002), mean velocity (−8%, p = 0.002), and low-frequency sway (−30%, p = 0.04) decreased, indicating smoother and more efficient postural control. A lateral redistribution of plantar load and an increase in contact surface area (up to +15%) were also found. These effects were less evident without visual input. Conclusions: Preliminary findings suggest that localized photobiomodulation integrated into hearing aids may positively influence postural stability in older adults with hearing impairment, possibly by supporting sensory integration processes. Further controlled studies are needed to confirm these effects and clarify the underlying mechanisms. Full article

In response to the challenges of low detection efficiency, high omission rate in small target detection and high model complexity in wood surface defect detection, this study proposes a lightweight detection model based on YOLO, which integrates a dual-path integrated attention network (DFA-Net). The model is built on the enhanced YOLOv5 framework and achieves a balance of accuracy and efficiency through the collaborative optimization of multiple modules. Specifically, this paper designs a dual-path downsampling convolutional module (DP-DCM), combining wavelet transform with dual-path feature fusion to improve multi-scale feature extraction capabilities while reducing the number of parameters. Next, a fusion attention module (FAM) is designed to dynamically focus on defect features in complex backgrounds through channel and spatial attention mechanisms. Furthermore, a focal modulation network (FMNet) is introduced to enhance the robustness of the augmentation model in detecting small defects. Finally, the NWD Loss function is used to mitigate the localization bias of small targets. Experimental results show that the improved model achieves a 92.8% mAP rate on five types of defect datasets (dead knots, live knots, cracks, notches, and marrow). Compared with the baseline model, YOLOv5s, the performance of this model has been improved by 6.5%. The model runs at a detection speed of 105 FPS, and the number of parameters is only 5.8 million, which is better than models such as YOLOv8 and YOLOv9-t. While maintaining a lightweight design, this method achieves high precision and real-time performance on a consumer-grade GPU platform, indicating its practical applicability in automated wood inspection scenarios. The proposed approach provides an efficient solution for intelligent wood sorting, contributing to improved wood utilization and enhanced processing automation. Full article

Conventional configurations for neutral-point-clamped multilevel converters exhibit an uneven power loss distribution among their power semiconductor devices. In particular, this imbalance increases the temperature of power semiconductor devices situated at specific locations within the converter layout, thereby reducing the reliability of the system. To mitigate this issue, neutral-point-clamped multilevel converters can be implemented using a switching-cell array design. This design allows for multiple topology configurations and inherently introduces redundant conduction paths, thus reducing conduction losses and also providing greater flexibility in distributing the switching losses. This work analyzes the thermal behavior of various configurations of a four-level dc–ac neutral-point-clamped converter based on a switching-cell array. An active thermal control strategy is used to distribute the switching losses, in order achieve a more uniform temperature distribution across the converter. Experimental results confirm that, compared to conventional neutral-point-clamped converter implementations, configurations based on a switching-cell array combined with active thermal control achieve a more uniform distribution of power losses. This leads to significantly improved temperature uniformity across the converter, thereby reducing thermal stress and enhancing overall system reliability. Full article

This paper proposed a reconfigurable intelligent surface (RIS)-empowered reference-modulated differential chaos shift keying (RM-DCSK) wireless powered communication (WPC) system. As a noncoherent chaotic communication scheme, the proposed system exploits the reference reuse property of RM-DCSK, where the reference signal simultaneously carries data information, thereby improving spectral efficiency while maintaining noncoherent and channel-estimation-free reception with low receiver circuit complexity. Furthermore, RIS is utilized to reconfigure the propagation environment and mitigate the path loss effect of WPC links. At the user equipment (UE), a harvest–store–use (HSU) energy harvesting and finite-buffer model is developed, and a threshold-based on/off transmission policy is adopted to enable sustainable uplink transmission. To quantify the gain of energy buffering and management, a bufferless baseline system is further established. Closed-form bit error rate (BER) expressions are obtained under multi-path Rayleigh fading channels for both the proposed RIS-RM-DCSK-WPC system and bufferless baseline system. Finally, simulation results validate the analysis and demonstrate that the proposed system achieves superior BER performance compared with representative benchmarks, including existing RIS-aided DCSK-WPC, RM-DCSK-WPC, and bufferless RIS-RM-DCSK-WPC systems. Full article

Mitigating climate change necessitates a thorough understanding of soil organic carbon (SOC) decomposition and its response to warming. The overuse of synthetic fertilizers can alter SOC composition and affect carbon cycling, potentially changing the temperature sensitivity (Q₁₀) of SOC decomposition. This study evaluated the Q₁₀ of SOC decomposition after long-term (37-year) excessive fertilization in a loess soil. Four treatments were compared: control (no nutrient input, CK); recommended rates of synthetic nitrogen (N) and phosphorus (P) fertilizers (CF_r); excessive rates of N and P fertilizers (CF_h); and CF_h plus organic manure (MCF_h). The Q₁₀ of SOC decomposition was investigated via an incubation experiment at temperatures of 15 °C, 25 °C and 35 °C within 63 days. Compared with CF_r, long-term CF_h and MCF_h significantly increased SOC contents by 14% and 67%, and this increase was driven primarily by rises in mineral-associated organic carbon (MOC) of 28% and 62% and particulate organic carbon (POC) of 32% and 79%, respectively, under CF_h and MCF_h. While CF_h and MCF_h did not change the SOC composition, they increased the proportions of fine POC (fPOC) to SOC by 10% and 91%, and the ratio of light POC to SOC by 78% and 143%, respectively. Q₁₀ values ranged from 2.18 to 3.00 across all treatments, with a mean of 2.64. Both CF_h and MCF_h drastically enhanced the Q₁₀ values by 38% and 25% compared with CF_r at 15–25 °C. However, MCF_h significantly decreased the Q₁₀ value by 31% relative to CF_r at 25–35 °C. Partial least squares path modeling showed that soil physicochemical properties and labile carbon fractions differed significantly among treatments, with physical properties regulating labile carbon fractions. Fertilization significantly increased the Q₁₀ value at 15–25 °C owing to increased proportion of labile carbon fractions and decreased labile carbon content. Our results suggest that SOC gains from continuous addition of synthetic fertilizers are vulnerable to loss under warming. However, this loss could be alleviated by incorporation of organic manure. Thus, integration of organic manure into nutrient management practices could be an efficient way to counteract warming-induced SOC decomposition. Full article