Search Results (4,901)

Driven by the dual-carbon goals, photovoltaic (PV) battery systems at renewable energy stations are increasingly clustered on the distribution side. The rapid expansion of these clusters, together with the pronounced uncertainty and spatio-temporal heterogeneity of PV generation, degrades battery utilization and forces conservative dispatch. To address this, we propose a “spatio-temporal clustering–deep estimation” framework for short-term interval forecasting of PV clusters. First, a graph is built from meteorological–geographical similarity and partitioned into sub-clusters by a self-supervised DAEGC. Second, an attention-based spatio-temporal graph convolutional network (ASTGCN) is trained independently for each sub-cluster to capture local dynamics; the individual forecasts are then aggregated to yield the cluster-wide point prediction. Finally, kernel density estimation (KDE) non-parametrically models the residuals, producing probabilistic power intervals for the entire cluster. At the 90% confidence level, the proposed framework improves PICP by 4.01% and reduces PINAW by 7.20% compared with the ASTGCN-KDE baseline without spatio-temporal clustering, demonstrating enhanced interval forecasting performance. Full article

As oilfield development enters the mid-to-late stages, conventional water flooding techniques face increasing challenges such as high water cut and limited improvement in recovery efficiency. Gas flooding has gradually become a critical method for enhancing oil recovery (EOR). However, significant heterogeneity in pore structures within complex reservoirs severely affects flow capacity and development performance during gas flooding processes. To elucidate the microscale flow mechanisms influenced by heterogeneity, this study constructs a series of two-dimensional pore network models with varying degrees of heterogeneity based on an improved Quartet Structure Generation Set algorithm. Gas-oil two-phase flow simulations were conducted using the multiphase flow module of COMSOL Multiphysics® 6.2. By adjusting the bimodal pore size ratio and pore distribution parameters, the heterogeneity level of the reservoir was systematically controlled, and relative permeability curves were extracted to inform macro-scale development strategy design. Simulation results indicate that (1) strong heterogeneity reduces the stability of the displacement front, leading to pronounced gas channeling; (2) in strongly heterogeneous pore structures, residual oil saturation significantly increases, with small pore regions forming residual oil-enriched zones that are difficult to mobilize; (3) relative permeability curves vary markedly under different heterogeneity conditions—oil-phase permeability declines rapidly during displacement, while gas-phase permeability rises sharply at high gas saturation levels. This study systematically investigates, for the first time, the microscale impact of pore structure heterogeneity on gas flooding behavior and applies pore-scale simulation outcomes to optimize macro-scale development strategies. The findings offer theoretical support and a technical pathway for gas injection design in complex heterogeneous reservoirs. While two-dimensional pore-network models enable controlled mechanistic and sensitivity analyses of heterogeneity, they do not fully capture three-dimensional connectivity and tortuosity. Accordingly, our results are positioned as mechanistic priors that are calibrated to field data during upscaling. Full article

Cholangiocarcinoma (CCA), an aggressive biliary tract cancer whose prevalence is rising, particularly in Thailand, is marked by elevated oxidative stress driven by chronic inflammation, parasitic infections, and dysregulated redox signaling. This study investigates the anticancer potential of tiliacorinine using a silico approach, including drug-likeness, ADMET, network pharmacology, molecular docking, and dynamics simulations. Tiliacorinine and 216 predicted targets were identified, with 79 overlapping CCA-related genes from GeneCards. GO and KEGG analyses revealed involvement in cell migration, membrane structure, kinase activity, and cancer-associated pathways. Network and PPI analyses identified ten key targets, including SRC, HIF1A, HSP90AA1, NFKB1, MTOR, MMP9, MMP2, PIK3CA, ICAM1, and MAPK1. Tiliacorinine showed the strongest affinity for MTOR (−10.78 kcal/mol, K_i = 12.62 nM), binding at the same site as known inhibitors with superior energy and specificity, supported by hydrogen bonding at ASP950 and hydrophobic interactions. Tiliacorinine also demonstrated strong binding to SRC, MMP9, and MAPK1. Molecular dynamics simulations revealed stable binding of tiliacorinine to MTOR, particularly at residues ASP950, TRP1086, and PHE1087. Comparative analysis with the MTOR–GDC-0980 complex confirmed consistent interaction patterns, reinforcing the structural stability and specificity of tiliacorinine. These results highlight its strong pharmacological potential and support its candidacy as a promising lead compound for cholangiocarcinoma therapy. Full article

The operational state of high-voltage disconnectors plays a critical role in ensuring the safety, stability, and power supply reliability of electrical systems. To enable accurate identification of the operational status of high-voltage disconnectors, this paper proposes a fault diagnosis method based on a Multi-Scale Residual Depthwise Separable Convolutional Neural Network (MSRDSN). First, wavelet transform is applied to vibration signals to perform multi-scale analysis and enhance detail resolution. Then, a novel network architecture, referred to as RDSN, is constructed to extract discriminative high-level features from vibration signals by integrating residual learning blocks and depthwise separable convolution blocks. Furthermore, a combined loss function is introduced to optimize the RDSN, which simultaneously maximizes inter-class distance, minimizes intra-class distance, and reduces feature redundancy. Experimental results show that the proposed method achieves a top accuracy of 99.44% on a balanced dataset, outperforming the sub-optimal approach by 1.11%. This study offers a novel and effective solution for fault diagnosis in high-voltage disconnectors. Full article

This study focuses on human awareness, a critical component in human–robot interaction, particularly within agricultural environments where interactions are enriched by complex contextual information. The main objective is identifying human activities occurring during collaborative harvesting tasks involving humans and robots. To achieve this, we propose a novel and lightweight deep learning model, named 1D-ResNeXt, designed explicitly for recognizing activities in agriculture-related human–robot collaboration. The model is built as an end-to-end architecture incorporating feature fusion and a multi-kernel convolutional block strategy. It utilizes residual connections and a split–transform–merge mechanism to mitigate performance degradation and reduce model complexity by limiting the number of trainable parameters. Sensor data were collected from twenty individuals with five wearable devices placed on different body parts. Each sensor was embedded with tri-axial accelerometers, gyroscopes, and magnetometers. Under real field conditions, the participants performed several sub-tasks commonly associated with agricultural labor, such as lifting and carrying loads. Before classification, the raw sensor signals were pre-processed to eliminate noise. The cleaned time-series data were then input into the proposed deep learning network for sequential pattern recognition. Experimental results showed that the chest-mounted sensor achieved the highest F1-score of 99.86%, outperforming other sensor placements and combinations. An analysis of temporal window sizes (0.5, 1.0, 1.5, and 2.0 s) demonstrated that the 0.5 s window provided the best recognition performance, indicating that key activity features in agriculture can be captured over short intervals. Moreover, a comprehensive evaluation of sensor modalities revealed that multimodal fusion of accelerometer, gyroscope, and magnetometer data yielded the best accuracy at 99.92%. The combination of accelerometer and gyroscope data offered an optimal compromise, achieving 99.49% accuracy while maintaining lower system complexity. These findings highlight the importance of strategic sensor placement and data fusion in enhancing activity recognition performance while reducing the need for extensive data and computational resources. This work contributes to developing intelligent, efficient, and adaptive collaborative systems, offering promising applications in agriculture and beyond, with improved safety, cost-efficiency, and real-time operational capability. Full article

For the fault issues in the workshop environmental control system sensor network, a fault diagnosis and recovery method based on medium-to-long-term predictions is proposed. Firstly, a temperature observer based on the Informer model is established. Then, the predicted data temporarily replaces the missing real data, and the model predicts the state of the sensor system within the step size. Secondly, the predicted data is combined with the measured temperature series, and residuals are utilized for real-time detection of sensor faults. Finally, the predicted data at the time of the fault replaces the real data, enabling the recovery of fault data; experiments are conducted to verify the effectiveness of the proposed method. The results indicate that when the prediction horizon is 1, 5, 10, 20, and 50, the average fault diagnosis rates under four fault levels are 94.40%, 95.28%, 94.79%, 92.52%, and 93.35%, respectively. The average coefficients of determination for data recovery are 0.999, 0.997, 0.995, 0.985, and 0.915, respectively. This achieves medium-to-long-term predictions in the field of sensor fault diagnosis. Full article

High-resolution meteorological satellite cloud imagery plays a crucial role in diagnosing and forecasting severe convective weather phenomena characterized by suddenness and locality, such as tropical cyclones. However, constrained by imaging principles and various internal/external interferences during satellite data acquisition, current satellite imagery often fails to meet the spatiotemporal resolution requirements for fine-scale monitoring of these weather systems. Particularly for real-time tracking of tropical cyclone genesis-evolution dynamics and capturing detailed cloud structure variations within cyclone cores, existing spatial resolutions remain insufficient. Therefore, developing super-resolution techniques for meteorological satellite cloud imagery through software-based approaches holds significant application potential. This paper proposes a Multi-scale Residual Deformable Attention Model (MRDAM) based on Generative Adversarial Networks (GANs), specifically designed for satellite cloud image super-resolution tasks considering their morphological diversity and non-rigid deformation characteristics. The generator architecture incorporates two key components: a Multi-scale Feature Progressive Fusion Module (MFPFM), which enhances texture detail preservation and spectral consistency in reconstructed images, and a Deformable Attention Additive Fusion Module (DAAFM), which captures irregular cloud pattern features through adaptive spatial-attention mechanisms. Comparative experiments against multiple GAN-based super-resolution baselines demonstrate that MRDAM achieves superior performance in both objective evaluation metrics (PSNR/SSIM) and subjective visual quality, proving its superior performance for satellite cloud image super-resolution tasks. Full article

Residual Networks (ResNet) address the vanishing gradient problem through skip connections and have become a fundamental architecture for computer vision tasks. However, standard convolutional layers exhibit limited capacity in modeling complex nonlinear relationships. We present EKAResNet, a residual backbone enhanced with a spline-based Kolmogorov–Arnold Network (KAN) head. Specifically, we introduce a KAN-based Feature Classification Module (KAN-FCM) that replaces a portion of the traditional fully connected classifier. This module employs piecewise polynomial (spline) approximation to achieve adaptive nonlinear mapping while maintaining a controlled parameter budget. We evaluate EKAResNet on CIFAR-10 and CIFAR-100, achieving top accuracies of 95.84% and 80.06%, respectively. Importantly, the model maintains a parameter count comparable to strong ResNet and WideResNet baselines. Ablation studies on spline configurations further confirm the contribution of the KAN head. These results demonstrate the effectiveness of integrating KAN structures into ResNet for modeling high-dimensional, complex features. Our work highlights a promising direction for designing deep learning architectures that balance accuracy and computational efficiency. Full article

Apple jam is a widely used all-season product. The quality of the jam is closely related to its sugar concentration, which affects its taste, texture, shelf life, and legal compliance with production requirements. Although traditional methods for measuring sugar, such as titration, enzymatic methods, and chromatography, are accurate, they are also invasive, destructive, and unsuitable for rapid screening. This study investigates a non-destructive and non-invasive alternative method that uses hyperspectral imaging (HSI) in combination with machine learning to estimate the sugar content in processed apple products. Eight cultivars were selected from the Central Asian region, recognized as the origin of apples and known for its rich diversity of apple cultivars. A total of 88 jam samples were prepared with sugar concentrations ranging from 25% to 75%. For each sample, several hyperspectral images were obtained using a visible-to-near-infrared (VNIR) camera. The acquired spectral data were then processed and analyzed using regression models, including the support vector machine (SVM), eXtreme gradient boosting (XGBoost), and a one-dimensional residual network (1D ResNet). Among them, ResNet achieved the highest prediction accuracy of R² = 0.948. The results highlight the potential of HSI and machine learning for a fast, accurate, and non-invasive assessment of the sugar content in processed foods. Full article

This study investigates the combined effect of wheat straw particle size and mixing ratio on the anaerobic co-digestion (ACD) of waste activated sludge under mesophilic conditions. Ten batch digesters were tested with varying straw-to-sludge ratios (0–1.5%) and particle sizes (5 cm, 1 cm, and <2 mm). Fine straw particles (<2 mm) at 1.5% loading achieved the highest removal efficiencies for TS (43.55%), TVS (47.83%), and COD (51.52%), resulting in a 140% increase in biogas yield and methane content of 60.15%. The energy recovery reached 14.37 kWh/kg, almost double the control. The developed Recurrent Neural Network (RNN) model (3 layers, 13 neurons, 500 epochs) predicted biogas production with 99.8% accuracy, a root mean square error (RMSE) of 0.0038, a mean absolute error (MAE) of 0.0093, and an R² close to 1. These results confirm the potential of integrating agricultural residues into wastewater treatment for renewable energy recovery and emission reduction. This study uniquely integrates mechanical pretreatment of wheat straw with RNN-based modeling to enhance biogas generation and predictive accuracy. It establishes a dual-experimental AI framework for optimizing sludge–straw co-digestion systems. This approach provides a scalable, data-driven solution for sustainable waste-to-energy applications. Full article

Surgical treatment is the only option for cholesteatoma; however, the recurrence rate is high, and the incidence of residual cholesteatoma recurrence largely depends on the surgeon’s skill. Training deep neural network (DNN) models typically requires large datasets, but the prevalence of cholesteatoma is low (1 in 25,000 people). It also remains difficult to treat. Developing analytical methods to improve accuracy with limited datasets remains a significant challenge in medical artificial intelligence (AI) research. This study introduces an AI-based system for detecting residual cholesteatoma in surgical field videos. A retrospective analysis was conducted on 144 cases from 88 patients who underwent surgery. The training dataset comprised videos of cholesteatoma lesions recorded during surgery and intact middle ear mucosa after lesion removal. These videos were captured using both an endoscope and a surgical microscope for AI model development. The diagnostic accuracy was approximately 80% for both endoscopic and microscopic images. Although the diagnostic accuracy for microscopic images was slightly lower, focusing on the lesion center improved the accuracy to a level comparable to that of endoscopic images. This study demonstrates the diagnostic feasibility of AI-based cholesteatoma detection despite a limited sample size, highlighting the value of proof-of-concept studies in clarifying technical requirements for future clinical systems. To our knowledge, this is the first AI study to use videos from both modalities. Full article

Network robustness optimization is crucial for enhancing the resilience of industrial networks and social systems against malicious attacks. Existing studies typically evaluate the robustness by simulating the sequential removal of nodes or edges and recording the residual connectivity at each step. However, the attack simulation is computationally expensive and becomes impractical for large-scale networks. Therefore, this paper proposes a multiobjective evolutionary algorithm assisted by a graph isomorphism network (GIN)-based surrogate model to efficiently optimize network robustness. First, the robustness optimization task is formulated as a multiobjective problem that simultaneously considers network robustness against attacks and the structural modification cost. Then, a GIN-based surrogate model is constructed to approximate the robustness, replacing the expensive simulation assessments. Finally, the multiobjective evolutionary algorithm is employed to explore promising network structures guided by the surrogate model, which is continuously updated via online learning to improve both prediction accuracy and optimization performance. Experimental results in various synthetic and real-world networks demonstrate that the proposed algorithm reduces the computational cost of the robustness evaluation by about $65\%$ while achieving comparable or even superior robustness optimization performance compared with those of baseline algorithms. These results indicate that the proposed method is practical and scalable and can be applied to enhance the robustness of industrial and social networks. Full article

Molecular simulation is central to modern drug discovery but is often limited by high computational cost and the complexity of molecular interactions. Deep-learning drug–target interaction (DTI) prediction can accelerate screening; however, many models underuse the local functional structure features—binding motifs, reactive groups, and residue-level fragments—that drive recognition. We present LoF-DTI, a framework that explicitly represents and couples such local features. Drugs are converted from SMILES into molecular graphs and targets from sequences into feature representations. On the drug side, a Jumping Knowledge (JK) enhanced Graph Isomorphism Network (GIN) extracts atom- and neighborhood-level patterns; on the target side, residual CNN blocks with progressively enlarged receptive fields, augmented by N-mer substructural statistics, capture multi-scale local motifs. A Gated Cross-Attention (GCA) module then performs atom-to-residue interaction learning, highlighting decisive local pairs and providing token-level interpretability through attention scores. By prioritizing locality during both encoding and interaction, LoF-DTI delivers competitive results across multiple benchmarks and improves early retrieval relevant to virtual screening. Case analyses show that the model recovers known functional binding sites, suggesting strong potential to provide mechanism-aware guidance for molecular simulation and to streamline the drug design pipeline. Full article

Protein S-palmitoylation is a reversible lipid modification that regulates various aspects of protein function, including membrane association, subcellular localization, trafficking, stability, and activity. The depalmitoylase ABHD17A removes palmitate from multiple substrates, but its cellular positioning and the role of its own palmitoylation in regulating its function remain unclear. This study identifies a palmitoylation code within the conserved N-terminal cysteine cluster of ABHD17A, which governs its intracellular distribution and plasma membrane (PM) targeting. N-terminal palmitoylation is essential for PM localization. Through the use of code-restricted mutants, we found that modifications in the middle region (C14, C15) are critical for PM targeting and catalytic activity, while modifications at the front (C10, C11) and rear (C18) influence endosomal routing and delivery to the PM. Alanine scanning revealed that adjacent hydrophobic residues, particularly L9 and F13, are crucial for initial engagement with endomembranes. Sequence analysis and mutagenesis identified two tyrosine-based YXXØ motifs within the alpha/beta hydrolase fold; disruption of the proximal motif (L115A) decreased surface abundance and redirected ABHD17A to autophagosomes, indicating a need for YXXØ-dependent endosomal sorting, likely at the trans-Golgi network. Biochemical assays demonstrated a continuum of acylation states influenced by the palmitoylation code. This requirement for the middle region was conserved in ABHD17B and ABHD17C. Overall, our findings suggest a stepwise mechanism for ABHD17A delivery to the PM, enabling its depalmitoylase activity on membrane-bound substrates. Full article

6G low-power and lossy network (6G LLN) is a kind of distributed network designed for IoT and edge computing scenarios of the sixth-generation mobile communication technology. Its routing technologies should fully consider characteristics of green and low carbon, constrained nodes, lossy links, etc. This paper proposes an improved routing protocol for low-power and lossy networks (I-RPL) to better suit the characteristics of 6G LLN and meet its application requirements. I-RPL has designed new context-aware routing metrics, which include the residual energy indicator, buffer utilization ratio, ETX, delay, and hop count to meet multi-dimensional network QoS requirements. The candidate parent and its preferred parent’s residual energy indicator and buffer utilization ratio are calculated recursively to reduce the effect of upstream parents. ETX and delay calculating methods are improved to ensure a better performance. Moreover, I-RPL has optimized the network construction process to improve energy and protocol efficiency. I-RPL has designed scientific multiple routing metrics evaluation theories (Lagrangian multiplier theories), proposed new rank computing and optimal route selecting mechanisms to simplify protocol, and optimized broadcast suppression and network reliability. Finally, theoretical analysis and experiment results show that the average end-to-end delay of I-RPL is 13% lower than that of RPL; the average alive node number increased 11% and so on. So, I-RPL can be applied well to the 6G LLN and is superior to RPL and its improvements. Full article