Search Results (365)

The rapid expansion of low-altitude unmanned aerial vehicles demands robust obstacle detection and avoidance systems capable of operating under diverse environmental conditions. This paper proposes a multimodal fusion attention network that integrates visual imagery and Light Detection and Ranging (LiDAR) point cloud data for real-time obstacle perception. The architecture incorporates a bidirectional cross-modal attention mechanism that learns dynamic correspondences between heterogeneous sensor modalities, enabling adaptive feature integration based on contextual reliability. An adaptive weighting component automatically modulates modal contributions according to estimated sensor confidence under varying environmental conditions. The network further employs gated fusion units and multi-scale feature pyramids to ensure comprehensive obstacle representation across different distances. A hierarchical avoidance decision framework translates detection outputs into executable control commands through threat assessment and graduated response strategies. Experimental evaluation on both public benchmarks and a purpose-collected low-altitude obstacle dataset demonstrates that the proposed method achieves 84.9% mean Average Precision (mAP) while maintaining 47.3 frames per second (FPS) on Graphics Processing Unit (GPU) hardware and 23.6 FPS on embedded platforms. Ablation studies confirm the contribution of each architectural component, with cross-modal attention providing the most substantial performance improvement. Full article

Manufacturing companies face significant challenges in leveraging artificial intelligence for equipment management due to high infrastructure costs and limited availability of labeled data for failures. While most manufacturing AI applications focus on structured sensor data, vast amounts of unstructured textual information containing valuable maintenance knowledge remain underutilized. This study presents a practical generative AI-based framework for structured information extraction that automatically converts unstructured equipment maintenance texts into predefined semantic fields to support predictive maintenance in manufacturing environments. We adopted and evaluated three representative generative models—Bidirectional and Auto-Regressive Transformers (BART) with KoBART, Text-to-Text Transfer Transformer (T5) with pko-t5-base, and the large language model Qwen—to generate structured outputs by extracting three predefined fields: failed components, failure types, and corrective actions. The framework enables the structuring of equipment management text data from Manufacturing Execution Systems (MES) to build predictive maintenance support systems. We validated the approach using a large-scale MES dataset consisting of 29,736 equipment maintenance records from a major automotive parts manufacturer, from which curated subsets were used for model training and evaluation. Our methodology employs Generative Pre-trained Transformer 4 (GPT-4) for initial dataset construction, followed by domain expert validation to ensure data quality. The trained models achieved promising performance when evaluated using extraction-aligned metrics, including exact match (EM) and token-level precision, recall, and F1-score, which directly assess field-level extraction correctness. ROUGE scores are additionally reported as a supplementary indicator of lexical overlap. Among the evaluated models, Qwen consistently outperformed BART and T5 across all extracted fields. The structured outputs are further processed through domain-specific dictionaries and regular expressions to create a comprehensive analytical database supporting predictive maintenance strategies. We implemented a web-based analytics platform enabling time-series analysis, correlation analysis, frequency analysis, and anomaly detection for equipment maintenance optimization. The proposed system converts tacit knowledge embedded in maintenance texts into explicit, actionable insights without requiring additional sensor installations or infrastructure investments. This research contributes to the manufacturing AI field by demonstrating a comprehensive application of generative language models to equipment maintenance text analysis, providing a cost-effective approach for digital transformation in manufacturing environments. The framework’s scalability and cloud-based deployment model present significant opportunities for widespread adoption in the manufacturing sector, supporting the transition from reactive to predictive maintenance strategies. Full article

This paper tackles the growing uncertainty in distribution networks caused by distributed generation, load fluctuations, and frequent topological changes. It proposes a multi-source data fusion framework using enhanced selective convolution (SKNet) and graph attention networks (GAT). First, heterogeneous measurement data, including Phasor Measurement Unit (PMU) and Supervisory Control and Data Acquisition (SCADA) data, are processed through a unified normalization and outlier elimination technique to ensure data quality. Second, SKNet is utilized to extract spatiotemporal multi-scale features, improving the detection of both rapid disturbances and long-term trends. Third, the extracted features are fed into GAT to model node electrical couplings, while power flow residual constraints are embedded in the loss function to enforce the physical validity of the estimated states. This physics-informed design overcomes a key limitation of pure data-driven models and enables an end-to-end framework that integrates data-driven learning with physical mechanism constraints. Finally, comprehensive validation is performed on the improved IEEE 33-node and IEEE 123-node test systems. The test scenarios include Gaussian measurement noise, data outliers, missing measurements, and topological changes. The results show that the proposed method outperforms baseline models such as Multi-Scale Graph Attention Network (MS-GAT), Bidirectional Long Short-Term Memory (BiLSTM), and traditional weighted least squares (WLS). It achieves Root Mean Square Error (RMSE) reductions of up to 18% and Mean Absolute Error (MAE) reductions of up to 15%. The average inference latency is only 10–18 ms. Even under unknown topological changes, the estimation error increases by only 15–25%. These results demonstrate the superior accuracy, robustness, and real-time performance of the proposed method for intelligent distribution network state estimation. Full article

The widespread connectivity of the Industrial Internet of Things (IIoT) improves the efficiency and functionality of connected devices. However, it also raises serious concerns about cybersecurity threats. Implementing an effective intrusion detection system (IDS) for IIoT is challenging due to heterogeneous data, high feature dimensionality, class imbalance, and the risk of data leakage during evaluation. This paper presents a leakage-safe hybrid intrusion detection framework that combines text-based and numerical network flow features in an IIoT environment. Each network flow is converted into a short text description and encoded using a frozen Large Language Model (LLM) called the Bidirectional Encoder Representations from Transformers (BERT) model to obtain fixed semantic embeddings, while numerical traffic features are standardized in parallel. To improve class separation, class prototypes are computed in Principal Component Analysis (PCA) space, and cosine similarity scores for these prototypes are added to the feature set. Class imbalance is handled only in the training data using the Synthetic Minority Over-sampling Technique (SMOTE). A Random Forest (RF) is used to select the top features, followed by a Histogram-based Gradient Boosting (HGB) classifier for final prediction. The proposed framework is evaluated on the Edge-IIoTset and ToN_IoT datasets and achieves promising results. Empirically, the framework attains 98.19% accuracy on Edge-IIoTset and 99.15% accuracy on ToN_IoT, indicating robust, leakage-safe performance. Full article

Temporal knowledge graphs (TKGs) effectively represent dynamic facts by incorporating a temporal dimension, yet they frequently encounter data incompleteness issues that constrain downstream applications. Concurrently, TKG prediction tasks, which enable reasoning about future events, have garnered significant attention. Existing TKG completion methods often neglect semantic information, underexploit event information from subsequent timestamps, and fail to leverage the structural symmetries inherent in temporal data. To address these limitations, this paper proposes a synergistic approach comprising two models: SiSe for completion and DL-CompGCN for prediction. SiSe integrates semantic and structural embeddings by employing entity text descriptions as semantic signals, utilizing symmetric cross-attention for bidirectional feature fusion and leveraging bidirectional gated recurrent units to capture symmetric temporal influences from both past and future events. On ICEWS14, ICEWS05-15, and GDELT completion datasets, the MRR improves by 1.2, 1.4, and 0.8 percentage points, respectively. DL-CompGCN addresses the accuracy–interpretability trade-off in prediction tasks through a time-aware graph convolutional encoder and a dual-decoder framework that combines bilinear scoring with first-order logical rules to generate interpretable paths while preserving the symmetric properties of temporal relations. It achieves state-of-the-art performance on ICEWS14, ICEWS05-15, and ICEWS18 prediction datasets. The proposed models explicitly incorporate symmetric principles in their architectural design; SiSe employs symmetric bidirectional temporal modeling, while DL-CompGCN maintains symmetry in its graph propagation and rule inference mechanisms. The experimental results demonstrate that both models significantly outperform baseline methods, offering a comprehensive solution for temporal knowledge graph reasoning that respects and exploits the symmetric structures inherent in temporal data. Full article

The high integration of renewables like distributed photovoltaic (PV) into medium- and low-voltage distribution networks causes bidirectional power flows, increased voltage fluctuations, and operational uncertainty. Traditional power flow models struggle to balance efficiency and accuracy for multi-period optimization. This paper proposes a dual-objective voltage optimization method based on a Holomorphic Embedding time-series power flow model. First, a recursive relationship for nodal voltage power series expansion is derived, revealing the linear superposition of first-order coefficients with power injection changes and the rapid decay of higher-order terms. A linearized analytical model neglecting higher-order terms is built, improving the computational efficiency of time-series power flow calculations while maintaining accuracy. Then, integrating energy storage systems and static var compensators, a dual-objective optimization model minimizing voltage deviation and daily operational cost is formulated. Tests on a practical 91-node rural distribution system show that the proposed power flow model maintains a voltage error below 0.25% compared to the Newton–Raphson method across various PV integration scenarios, and the optimization reduces computation time by about 61.3% versus the Second-Order Cone Programming method, validating its advantages in precision and efficiency for balancing voltage quality and economy. Full article

The quality of tea bud harvesting directly affects the final quality of the tea; however, due to the small size of tea buds and the complex natural background, accurately detecting them remains challenging. To address this issue, this paper proposes a lightweight and efficient tea bud detection model named TSF-Net. This model adopts the P2-enhanced bidirectional feature pyramid network (P2A-BiFPN) to enhance the recognition ability of small objects and achieve efficient multi-scale feature fusion. Additionally, coordinate space attention (CSA) is embedded in multiple C3k2 blocks to enhance the feature extraction of key regions, while an A2C2f module based on self-attention is introduced to further improve the fine feature representation. Extensive experiments conducted on the self-built WYTeaBud dataset show that TSF-Net increases mAP@50 by 2.0% and reduces the model parameters to approximately 85% of the baseline, achieving a good balance between detection accuracy and model complexity. Further evaluations on public tea bud datasets and the VisDrone2019 small object benchmark also confirm the effectiveness and generalization ability of the proposed method. Moreover, TSF-Net is converted to the RKNN format and successfully deployed on the RK3588 embedded platform, verifying its practical applicability and deployment potential in intelligent tea bud harvesting. Full article

In this study, we investigate how computationally simplified activation functions affect predictive performance, inference latency, and energy usage in long short-term memory-based temperature prediction for wind turbine generator bearings. We tested three different types of long short-term memory (LSTM) cells, along with bidirectional LSTM (biLSTM) networks, to determine their effectiveness in modeling dynamic changes in gearbox bearing temperatures. We compared several activation-function variants, focusing on variants that are either computationally simple or known to give good performance in deep recurrent networks. The results show that the best-performing architectures achieved root mean squared errors (RMSEs) between $0.0798$ and $0.0822$, corresponding to coefficients of determination in the range of $R^2=0.84$–$0.85$. When applied across five turbines, the best-performing architectures (peephole and bidirectional) achieved root mean squared errors of $0.0898$, $0.0882$, and $0.042$, respectively. The best activation function-enhanced variant (the peephole) improved accuracy by approximately $3\%$ while maintaining low model complexity. These findings provide a practical and efficient solution for embedded predictive maintenance systems, providing high accuracy without incurring the computational cost of deeper or bidirectional architectures. Full article

As control and execution components within hydraulic systems, hydraulic valves are critical to system efficiency and operational safety. However, existing research primarily focuses on specific valve designs, exhibiting limitations in versatility and task coordination that constrain their comprehensive diagnostic capabilities. To address these issues, this paper innovatively proposes a multi-modal feature deep fusion multi-task prediction (CLAF-MTL) model. This model employs a core architecture based on the CNN-LSTM-Additive Attention module and a fully connected network (FCN) for multi-domain features, while simultaneously embedding a multi-task learning mechanism. It resolves the multi-task prediction challenge of leakage rate regression and fault type classification, significantly enhancing diagnostic efficiency and practicality. This model innovatively designs a complementary fusion mechanism of “deep auto-features + multi-domain features” overcoming the limitations of single-modality representation. It integrates leakage rate regression and fault type classification into a unified modeling framework, dynamically optimizing dual-task weights via the MGDA-UB algorithm to achieve bidirectional complementarity between tasks. Experimental results demonstrate that the proposed method achieves an R² of 0.9784 for leakage rate prediction and a fault type identification accuracy of 92.23% on the test set. Compared to traditional approaches, this method is the first to simultaneously address the challenge of accurately predicting both leakage rate and fault type. It exhibits superior robustness and applicability across generic valve scenarios, providing an effective solution for intelligent monitoring of valve leakage faults in hydraulic systems. Full article

The integration of high-penetration renewable energy sources (RESs) and electric vehicles (EVs) increases the risk of voltage fluctuations in distribution networks. Traditional static partitioning strategies struggle to handle the intermittency of wind turbine (WT) and photovoltaic (PV) generation, as well as the spatiotemporal randomness of EV loads. Furthermore, existing scheduling methods typically optimize EV active power or reactive compensation independently, missing opportunities for synergistic regulation. The main novelty of this paper lies in proposing a spatiotemporally coupled voltage-stability optimization framework. This framework, based on an hourly updated electrical distance matrix that accounts for RES uncertainty and EV spatiotemporal transfer characteristics, enables hourly dynamic network partitioning. Simultaneously, coordinated active–reactive optimization control of EVs is achieved by regulating the power factor angle of three-phase six-pulse bidirectional chargers. The framework is embedded within a hierarchical model predictive control (MPC) architecture, where the upper layer performs hourly dynamic partition updates and the lower layer executes a five-minute rolling dispatch for EVs. Simulations conducted on a modified IEEE 33-bus system demonstrate that, compared to uncoordinated charging, the proposed method reduces total daily network losses by 4991.3 kW, corresponding to a decrease of 3.9%. Furthermore, it markedly shrinks the low-voltage area and generally raises node voltages throughout the day. The method effectively enhances voltage uniformity, reduces network losses, and improves renewable energy accommodation capability. Full article

Aiming to address insufficient feature extraction, vanishing gradients, and low prediction accuracy in tool wear prediction, this paper proposes a hybrid deep neural network based on a Convolutional Neural Network (CNN), Residual Network (ResNet) residual connections, the Convolutional Block Attention Module (CBAM), and a Bidirectional Gated Recurrent Unit (BiGRU). First, a 34-dimensional multi-domain feature set covering the time domain, frequency domain, and time–frequency domain is constructed, and multi-sensor signals are standardized using z-score normalization. A CNN–BiGRU backbone is then established, where ResNet-style residual connections are introduced to alleviate training degradation and mitigate vanishing-gradient issues in deep networks. Meanwhile, CBAM is integrated into the feature extraction module to adaptively reweight informative features in both channel and spatial dimensions. In addition, a BiGRU layer is embedded for temporal modeling to capture bidirectional dependencies throughout the wear evolution process. Finally, a fully connected layer is used as a regressor to map high-dimensional representations to tool wear values. Experiments on the PHM2010 dataset demonstrate that the proposed hybrid architecture is more stable and achieves better predictive performance than several mainstream deep learning baselines. Systematic ablation studies further quantify the contribution of each component: compared with the baseline CNN model, the mean absolute error (MAE) is reduced by 47.5%, the root mean square error (RMSE) is reduced by 68.5%, and the coefficient of determination (R²) increases by 14.5%, enabling accurate tool wear prediction. Full article

Accurate traffic flow prediction is a core component of intelligent transportation systems, supporting proactive traffic management, resource optimization, and sustainable urban mobility. However, urban traffic networks exhibit heterogeneous multi-scale periodic patterns and time-varying spatial interactions among road segments, which are not sufficiently captured by many existing spatio-temporal forecasting models. To address this limitation, this paper proposes PDR-STGCN (Periodicity-Aware Dynamic Relational Spatio-Temporal Graph Convolutional Network), an enhanced STGCN framework that jointly models multi-scale periodicity and dynamically evolving spatial dependencies for traffic flow prediction. Specifically, a periodicity-aware embedding module is designed to capture heterogeneous temporal cycles (e.g., daily and weekly patterns) and emphasize dominant social rhythms in traffic systems. In addition, a dynamic relational graph construction module adaptively learns time-varying spatial interactions among road nodes, enabling the model to reflect evolving traffic states. Spatio-temporal feature fusion and prediction are achieved through an attention-based Bidirectional Long Short-Term Memory (BiLSTM) network integrated with graph convolution operations. Extensive experiments are conducted on three datasets, including Metro Traffic Los Angeles (METR-LA), Performance Measurement System Bay Area (PEMS-BAY), and a real-world traffic dataset from Guizhou, China. Experimental results demonstrate that PDR-STGCN consistently outperforms state-of-the-art baseline models. For next-hour traffic forecasting, the proposed model achieves average reductions of 16.50% in RMSE, 9.00% in MAE, and 0.34% in MAPE compared with the second-best baseline. Beyond improved prediction accuracy, PDR-STGCN reveals latent spatio-temporal evolution patterns and dynamic interaction mechanisms, providing interpretable insights for traffic system analysis, simulation, and AI-driven decision-making in urban transportation networks. Full article

The integration of emojis into Early Childhood Education (ECE) presents a promising avenue for enhancing student engagement, emotional expression, and comprehension. While prior studies suggest the benefit of visual aids in learning, systematic frameworks for pedagogically aligned emoji recommendation remain underdeveloped. This paper presents EduEmoji-ECE, a pedagogically annotated dataset of early-childhood learning text segments. Specifically, the proposed model incorporates Bidirectional Encoder Representations from Transformers (BERTs) for contextual embedding extraction, Gated Recurrent Units (GRUs) for sequential pattern recognition, Deep Neural Networks (DNNs) for classification and emoji recommendation, and DECOC for improving emoji class prediction robustness. This hybrid BERT-GRU-DNN-DECOC architecture effectively captures textual semantics, emotional tone, and pedagogical intent, ensuring the alignment of emoji class recommendation with learning objectives. The experimental results show that the system is effective, with an accuracy of 95.3%, a precision of 93%, a recall of 91.8%, and an F1-score of 92.3%, outperforming baseline models in terms of contextual understanding and overall accuracy. This work helps fill a gap in AI-based education by combining learning with visual support for young children. The results suggest an association between emoji-enhanced materials and improved engagement/comprehension indicators in our exploratory classroom setting; however, causal attribution to the AI placement mechanism is not supported by the current study design. Full article

Symmetry and asymmetry are foundational design principles in artificial intelligence, defining the balance between invariance and adaptability in multimodal learning systems. In audio-visual speaker verification, where speech and lip-motion features are jointly modeled to determine whether two utterances belong to the same individual, these principles govern both fairness and discriminative power. In this work, we analyze how symmetry and asymmetry emerge within a gated-fusion architecture that integrates Time-Delay Neural Networks and Bidirectional Long Short-Term Memory encoders for speech, ResNet-based visual lip encoders, and a shared Conformer-based temporal backbone. Structural symmetry is preserved through weight-sharing across paired utterances and symmetric cosine-based scoring, ensuring verification consistency regardless of input order. In contrast, asymmetry is intentionally introduced through modality-dependent temporal encoding, multi-head attention pooling, and a learnable gating mechanism that dynamically re-weights the contribution of audio and visual streams at each timestep. This controlled asymmetry allows the model to rely on visual cues when speech is noisy, and conversely on speech when lip visibility is degraded, yielding adaptive robustness under cross-modal degradation. Experimental results demonstrate that combining symmetric embedding space design with adaptive asymmetric fusion significantly improves generalization, reducing Equal Error Rate (EER) to 3.419% on VoxCeleb-2 test dataset without sacrificing interpretability. The findings show that symmetry ensures stable and fair decision-making, while learnable asymmetry enables modality awareness together forming a principled foundation for next-generation audio-visual speaker verification systems. Full article

This paper introduces and systematically develops the theory of Zappa–Szép skew braces, a novel algebraic structure that provides a unified framework for bidirectional group interactions, thereby generalizing the classical constructions of semidirect skew braces and matched-pair factorizations (ZS1–ZS4, BC1–BC2). We establish the complete axiomatic foundation for these objects, characterizing them through necessary and sufficient compatibility conditions that encode mutual actions between two digroups. Central results include a semidirect embedding theorem, explicit constructions of nontrivial examples—notably a fully mutual brace of order 12 built from $V_4$ and $C_3$—and a detailed analysis of key structural invariants such as the socle, center, and automorphism groups. The framework is further elucidated via universal properties and categorical adjunctions, positioning Zappa–Szép skew braces as fundamental objects within noncommutative algebra. Applications to representation theory, cohomology, and the construction of set-theoretic solutions to the Yang–Baxter equation are derived, demonstrating both the generality and utility of the theory. Full article