Search Results (1,714)

Illegal angling activities pose significant threats to aquatic ecosystem conservation and sustainable water resource management by disrupting ecological balance and aquatic resource protection, emphasizing the need for effective intelligent monitoring approaches. Illegal angling detection in complex aquatic environments remains challenging due to small target sizes, diverse human postures, and severe interference from shoreline vegetation, water reflections, and other complex backgrounds. To address these issues, this paper proposes an improved YOLOv8-based illegal angling detection framework, termed ERFA–YOLO. To enhance the discriminative representation capability of slender targets in complex scenes, an Enhanced Receptive Field Attention mechanism (ERFAConv) is introduced. By leveraging adaptive contextual perception and spatial geometric feature modeling, the proposed mechanism effectively enhances fishing-related target features while suppressing false activations from background noise. Furthermore, a temporal consistency-based post-processing strategy is introduced to reduce false positives caused by transient prediction noise and improve detection stability in dynamic aquatic environments. In addition, a dedicated illegal angling dataset covering multiple time periods, weather conditions, and complex shoreline environments is constructed to improve the generalization capability of the model in real-world natural scenarios. Experimental results demonstrate that, compared with the original YOLOv8 baseline, ERFA–YOLO achieves a Precision of 93.19% (+4.67%), a Recall of 86.24% (+1.34%), an ${\mathrm{mAP}}_{50}$ of 93.49% (+3.47%), and an ${\mathrm{mAP}}_{50:95}$ of 60.53%, while achieving real-time inference performance of 75.34 FPS on an NVIDIA RTX 4090 GPU. Compared with several mainstream object detection algorithms, the proposed method exhibits superior robustness and detection stability in complex natural environments, demonstrating the potential of ERFA–YOLO for intelligent illegal angling monitoring in sustainable aquatic resource management scenarios. Full article

This paper introduces a novel physical layer authentication technique for Internet of Things (IoT) networks, leveraging channel state information (CSI) data from Wi-Fi signals to distinguish between authorised and unauthorised nodes and thereby enhancing security without compromising performance. The core novelty lies in its integrated framework, which employs non-negative matrix factorization (NMF) for efficient feature selection and a Gaussian mixture model (GMM) for identifying complex patterns within the CSI data specifically adapted to the dynamic nature of IoT networks. NMF is utilised to mitigate the high dimensionality and redundancy inherent in raw CSI metrics, reducing processing load, extracting salient features, alleviating overfitting risks, and exhibiting superior resilience to noise. Following NMF, the GMM component is used for data classification, capitalising on its probabilistic and soft clustering attributes to represent intricate distributions and handle heterogeneous CSI data characteristics. This integrated proposed methodology not only exploits the inherent nonlinear and probabilistic characteristics of CSI data but also upholds computational efficiency, making it highly suitable for implementation in resource-constrained IoT wireless networks. The model achieves exceptional classification proficiency, with an accuracy rate of 99.83 percent and a recall of 100 percent, which are crucial for cybersecurity and anomaly detection. Furthermore, the system is designed for efficiency and minimal resource consumption, exhibiting good computational efficiency, reduced training duration, and lower energy consumption compared with more complex, heavily exploited architectures for CSI data processing like CNN and CNN + LSTM, making it particularly suitable for resource-constrained IoT environments. Full article

In resource-constrained intelligent transportation systems (ITSs), raw road sensor data is often affected by non-stationary noise and outliers, which severely reduces the accuracy of traffic flow prediction. To address this, this paper proposes a hybrid prediction framework that integrates Frequency Mode Decomposition (FMD), Enhanced Beluga Whale Optimization (EBWO), and the Logistic-Gaussian Circle-based Bidirectional Gated Recurrent Unit (LGC-BiGRU). The proposed FMD module introduces an adaptive frequency domain segmentation mechanism, which effectively separates noise from the data without introducing modal aliasing. EBWO employs an adaptive balancing factor mechanism to dynamically explore the spatiotemporal characteristics of traffic flow, preventing the framework from getting trapped in local optima. For predictive modeling, the LGC-BiGRU integrates an LGC Optimizer, mapping features to a circular probability space to enhance sensitivity to periodic patterns. Experimental results show that, compared with the best-performing state-of-the-art model, our framework reduces the mean absolute error (MAE) by an average of 30.6% and the root mean square error (RMSE) by 29.9%. These results confirm the framework’s effectiveness and feasibility for traffic flow prediction under complex noise. Full article

Measuring prices in agricultural e-commerce markets is complicated by promotional spikes, rapid product turnover, and specification heterogeneity. The aim of this study is to develop and validate a reproducible framework that couples promotion-robust weekly price measurement with calibrated short-term forecasting for agricultural e-commerce input markets. Drawing on 452,489 de-identified raw transaction records (448,308 retained after quality control) of tomato, pepper, and lettuce seeds from major Chinese e-commerce platforms (July 2020–October 2025), we construct a weekly Paasche-robust price index with a Top-10 SKU basket and winsorized prices, benchmarked against Laspeyres, Fisher, Törnqvist, and GEKS-TPD formulations. A time-dummy hedonic regression is applied for quality adjustment. Short-term forecasts are generated by a two-level ensemble combining SARIMAX, Temporal Fusion Transformer, and LightGBM via inverse-MAPE weighting, with adaptive conformal intervals for uncertainty quantification. The Paasche-robust index attenuates promotional noise while co-moving closely with superlative formulas (MA-12 correlations > 0.80). Machine-learning models reduce the one-week-ahead MAPE by 42–58% relative to SARIMAX, and the weighted ensemble achieves the best four-week-ahead accuracy for two of three crops. Conformal intervals deliver near-nominal coverage at short horizons. The framework offers a reproducible, deployable tool for real-time monitoring of agricultural e-commerce input markets and supports evidence-based decisions by farmers and policymakers. Full article

Financial markets in emerging economies exhibit fractal long-memory dynamics ($H=0.93$ on CBOE VIX; $H=0.81$ on JSE realised volatility) and extreme non-Gaussianity (kurtosis $=51.6$) that together invalidate conventional Gaussian-emission regime models. This paper introduces SARDINE (Stochastic Adaptive Regime Detection via Integrated Neuro-Fuzzy Estimation), a continuous-time framework that embeds Takagi-Sugeno-Kang (TSK) fuzzy membership weights directly inside a stochastic differential equation integrator, governing the geometry of stochastic uncertainty at each integration step. Three formal results underpin the architecture: a Lyapunov-style diffusion stability bound, a regime-adaptive noise attenuation guarantee, and an asymmetric cost advantage condition. Evaluated on 17 JSE Top40 securities ($N=2778$ daily observations, 2015–2026; 537-day held-out test), SARDINE achieves FAR $=0.143$, FNR $=0.195$, and a 1.33-day early-warning lead time, reducing the asymmetric cost by 32% relative to GMM. Against 14 baselines including Neural SDE, PatchTST, and Mamba, the fuzzy-gated diffusion reduces false alarms by 54–63%. A fractional Brownian motion ablation ($H\in \{0.44,0.50,0.93\}$, $B=200$ replicates) reveals that long-memory information should be embedded in the input representation rather than the noise driver; $H=0.50$ is recommended for operational deployment. Full article

Distance estimation plays an important role for path planning and collision avoidance of swarm UAVs. However, the lack of annotated data seriously hinders related studies. In this work, we build and present a UAVDE dataset for UAV distance estimation, in which the distance between two UAVs is obtained by UWB sensors. During experiments, we observe that center-based stereo triangulation becomes unreliable in long-range UAV scenes. We show that this performance degradation is mainly caused by disparity errors introduced by practical UAV imaging conditions and amplified by long-range stereo geometry. To tackle this issue, we propose a novel position correction module, which predicts horizontal compensation offsets for the observed UAV centers under UWB distance supervision and applies them before stereo triangulation. Furthermore, to improve the robustness and generalization ability of position correction, we introduce a causal feature selection module into the position correction process. It adaptively selects correction-relevant causal feature channels and suppresses non-causal components caused by background interference, target appearance variation, and detection noise. We conduct extensive experiments on UAVDE. Our method achieves a significant performance improvement over a strong baseline (by reducing the relative difference from 49.4% to 8.6%), which demonstrates its effectiveness and superiority. Full article

Background/Objectives: Speech disorders caused by neurological, articulatory, cognitive, and behavioral impairments require accurate multimodal analysis frameworks capable of jointly understanding acoustic abnormalities and neurolinguistic inconsistencies for reliable computer-assisted speech disorder assessment. Recent multimodal deep learning approaches have utilized speech signals, linguistic transcripts, attention mechanisms, and contextual fusion strategies to improve disordered speech analysis and intelligent disorder prediction. However, existing methods frequently suffer from insufficient temporal synchronization, ineffective local–global feature learning, multimodal redundancy, poor interpretability, and reduced robustness under heterogeneous speech disorders. To address these challenges, Methods: The proposes NSX-Net (NeuroSpeech Explainable Network), an intelligent multimodal deep learning framework for speech disorder classification using Acoustic Speech Signal Data and Neurolinguistic Text/Linguistic Data. Initially, the Adaptive Speech Refinement Module (ASRM) performs noise removal, silence elimination, signal normalization, spectrogram generation, MFCC extraction, and transcript preprocessing to improve multimodal speech consistency. Subsequently, the Hierarchical Multimodal Feature Learning Unit (HMFLU) extracts discriminative feature representations through the Cross-Domain Representation Encoder (CDRE), Fine-Grained Speech Pattern Analyzer (FGSPA), and Global Sequential Dependency Learner (GSDL) for capturing both local articulation abnormalities and long-range semantic dependencies. Furthermore, the Dual-Path Attention Enhancement Block (DPAEB) emphasizes clinically important disordered speech regions using adaptive local–global attention mechanisms, while the Temporal Resolution Synchronization Module (TRSM) aligns rhythm-level and beat-level multimodal contextual structures to improve temporal consistency and suppress noise disturbances. Results: Experimental evaluation across six publicly available multimodal speech disorder datasets demonstrated that NSX-Net achieved superior performance, with 99.27% Accuracy, 99.11% Precision, 98.97% Recall, 99.02% F1-Score, and 99.41% AUC, significantly outperforming existing state-of-the-art frameworks. Conclusions: Finally, optimized multimodal contextual representations are forwarded into a Softmax classification layer for intelligent multiclass speech disorder prediction with improved interpretability and potential clinical applicability. Full article

To address the challenges of sparse point clouds, incomplete structures, and noise interference in on-orbit perception of non-cooperative satellites, this study develops a reconstruction method that integrates multi-frame point cloud fusion with an energy-driven optimization strategy. A self-developed experimental platform for non-cooperative targets is employed to acquire multi-view point cloud data, which are subsequently fused to improve geometric coverage and structural completeness. Based on this dataset, four representative 3D reconstruction algorithms are systematically evaluated. The results indicate that the Local Delaunay method achieves a favorable balance between geometric accuracy and structural preservation (RMSE = 6.152 mm). However, its performance is limited by reliance on multiple geometric thresholds and fixed-scale constraints, leading to reduced robustness and stability under challenging conditions. To overcome these limitations, an energy-based normal-consistent adaptive reconstruction method (ENALD) is developed within the Local Delaunay framework. The proposed approach incorporates a normal consistency constraint and an adaptive scale mechanism and constructs a multi-feature energy model to guide triangle selection. An energy-ranking strategy is further introduced, retaining the top 50% of candidates to improve reconstruction quality. Experimental results demonstrate that, compared with the conventional Local Delaunay method, ENALD reduces the maximum deviation, average deviation, and RMSE by 21.3%, 10.0%, and 11.9%, respectively, while improving computational efficiency by approximately 45.4%. In addition, the reconstructed surfaces exhibit enhanced structural continuity and better preservation of local details, without introducing noticeable topological artifacts. Full article

This paper presents a self-tuning minimal-rule fuzzy logic controller for high-performance induction motor drives operating under field-oriented control. Unlike conventional full-rule fuzzy controllers and reduced-rule designs with fixed post-design scaling, the proposed method combines a fixed nine-rule Mamdani inference structure with a bounded online output gain adaptation mechanism. The nominal gain and adaptation sensitivity are determined offline using Particle Swarm Optimization, thereby retaining operating-condition responsiveness without requiring online optimization, rule reconstruction, or membership-function retuning. The closed-loop behavior is analyzed using a discrete-time Lyapunov framework derived from the induction motor mechanical dynamics under bounded disturbances. The controller is evaluated through fixed-step simulations incorporating measurement noise, 12-bit signal quantization, and a one-sample computational delay. Comparative results against a conventional PI controller and a classical 49-rule fuzzy controller show that the proposed scheme achieves a rise time of 0.15 s, a settling time of 0.26 s, a post-transient mean absolute tracking error of 4 RPM, negligible overshoot, and a torque ripple of approximately 0.44 Nm. Relative to the classical 49-rule FLC, the proposed design reduces the maximum number of fuzzy-rule evaluations per control update from 49 to 9, corresponding to an 81.6% reduction in structural fuzzy-inference complexity. The results indicate a favorable simulation-level trade-off between dynamic performance, disturbance rejection, and structural algorithmic simplicity. Generated-code SIL, Hardware-in-the-Loop testing, target-processor timing measurements, and experimental implementation remain necessary to establish practical embedded feasibility. Full article

Underwater object detection is crucial for marine ecological monitoring and resource exploration. However, low underwater contrast causes severe aliasing between foreground and background in the spatial domain, and conventional methods struggle to effectively decouple their features. Transformer architectures incur high computational overhead. Conversely, over-compressing these models severely degrades their ability to detect heavily camouflaged or small marine organisms. High-quality underwater samples are limited, and existing time-consuming generative strategies are hard to implement efficiently on edge devices. To address these challenges, this paper proposes a resource-efficient framework for underwater object detection. First, we design a Wavelet-Enhanced Feature Pyramid Network that combines a saliency-focus mechanism and a discrete wavelet transform to overcome background noise in both spatial and frequency domains, extracting features of hidden small objects. Second, a data-dependent dynamic token pruning technique removes redundant tokens, effectively mitigating the computational bottleneck without sacrificing essential semantic capacity. Finally, for extreme sample scarcity, we introduce a Feature Correction Module and a two-stage fine-tuning and feature correction strategy, using a high-precision teacher model to guide a compressed student network in adaptively compensating for optical shifts with few samples. Experiments on URPC2020 and DUO demonstrate that our method improves small object detection accuracy while reducing parameter count and computational overhead, striking a good balance between accuracy and inference efficiency. Full article

Federated learning enables collaborative model training without centralizing private data, yet its deployment on edge devices remains constrained by communication overhead, heterogeneous data distributions, and privacy leakage from shared model updates. Spiking neural networks offer an energy-efficient alternative to conventional artificial neural networks by transmitting sparse binary events rather than dense activations, but existing federated spiking learning methods still suffer from inefficient gradient exchange and insufficient privacy protection under non-independent and identically distributed data. This paper proposes PriSparK, a privacy-preserving and communication-efficient federated spiking neural learning framework that jointly exploits temporal event sparsity, adaptive client-side spike-gradient compression, and privacy-calibrated aggregation. The core of PriSparK is a novel Event-Sparse Differentially Private Federated Spiking Optimization algorithm, which converts local surrogate gradients into spike-saliency-aware sparse updates, dynamically allocates communication budgets across layers and time steps, and injects calibrated Gaussian noise after clipping in a low-dimensional event subspace. To mitigate accuracy degradation caused by aggressive compression and privacy perturbation, PriSparK further introduces a membrane-aware error-feedback mechanism and a heterogeneity-adaptive server aggregation rule that weights client updates according to spike activity stability and local distribution drift. Experiments on neuromorphic and vision benchmarks, including N-MNIST, DVS128 Gesture, CIFAR-10, and Fashion-MNIST, show that PriSparK achieves competitive or superior accuracy compared with federated artificial neural and spiking baselines while substantially reducing uplink communication. Under strong privacy constraints, PriSparK maintains stable convergence and improves the accuracy–communication–privacy trade-off, demonstrating its potential for privacy-sensitive edge intelligence with event-driven neural computation. Full article

Torsional vibrations represent a significant dynamic phenomenon in rotating mechanical systems and are often associated with increased dynamic loading, fatigue damage, noise generation, and reduced operational reliability. Conventional vibration mitigation techniques are generally effective only within a limited frequency range, which restricts their applicability in modern drivetrains operating under variable loading conditions. Consequently, increasing attention has been devoted to nonlinear vibration control concepts based on the principle of targeted energy transfer. This paper presents the development and experimental investigation of a novel TET system with variable torsional stiffness intended for torsional vibration mitigation in rotating mechanical systems. The proposed concept combines the vibration energy redistribution capability of a nonlinear absorber with adaptive stiffness tuning achieved through pneumatic elements. The torsional stiffness of the secondary subsystem can be continuously adjusted by regulating the pressure within air bellows, enabling adaptation of the system dynamics to varying operating conditions. A dedicated experimental test rig with kinematic excitation was developed to investigate the dynamic response of the coupled mechanical system and evaluate the influence of variable stiffness on the TET mechanism. The study focuses on the analysis of vibration energy redistribution, the identification of optimal operating conditions, and the assessment of the potential of variable-stiffness TET systems for wide range torsional vibration control in rotating machinery. Full article

During deep excavation construction, deformation impacts on adjacent structures are inevitably induced, and field monitoring data are often contaminated by noise that degrades prediction accuracy. To address these issues, this study develops a joint denoising strategy combining CEEMDAN, sample entropy-based adaptive IMF screening, and an improved wavelet threshold (IWT) function, followed by a Bayesian optimization-based extreme gradient boosting (BO-XGBoost) model for surface settlement prediction. The developed method automatically identifies high-noise IMF components via sample entropy and processes them using an improved threshold function that overcomes the discontinuity of hard thresholding and the constant bias of soft thresholding, thereby preserving useful information while suppressing noise. Experimental results on a Wuhan metro deep excavation project demonstrate that the CEEMDAN-IWT method improves SNR by up to 4.09% and reduces RMSE by up to 8.00% compared with conventional CEEMDAN-wavelet threshold denoising. The BO-XGBoost model trained on denoised data achieves an RMSE of 0.09 mm and a MAPE of 3.54%, outperforming BP, LSTM, standard XGBoost, GRU, CNN-LSTM, TCN, and simple regression baselines. Feature importance analysis confirms that the denoised data retain physical interpretability consistent with soil deformation continuity. This framework provides a practical solution with promising accuracy for deformation monitoring and early warning in deep excavation engineering. Full article

Diesel generator sets are key frequency-supporting units in islanded microgrids and shipboard power systems, where rapid speed recovery under abrupt load variations is essential for maintaining power quality. However, conventional linear active disturbance rejection control (LADRC) is limited by the disturbance-estimation and noise-amplification trade-off of a single observer, while fixed parameters restrict its adaptability under varying operating conditions. To address these limitations, this paper proposes an RBF neural-network-optimized cascaded LADRC method, termed RBF-CLADRC. A mechanism-based torque balance model is first established, with uncertain mechanical coupling, friction losses, and load variations lumped into the total disturbance. A residual-disturbance cascaded observer is then constructed, in which the first linear extended state observer estimates the total disturbance and the second further reconstructs the residual estimation error. Unlike conventional ML-based ADRC methods that directly tune multiple gains, the proposed RBFNN adjusts only a common controller bandwidth within a prescribed interval, while all observer and feedback gains are generated through predefined analytical relationships. This low-dimensional adaptation preserves coordinated gain variation, reduces online computational complexity, and facilitates real-time implementation. Lyapunov analysis shows that the observer and tracking errors are uniformly ultimately bounded under bounded disturbance rates and converge exponentially for constant disturbances. Finally, comparative simulations in MATLAB/Simulink demonstrate that the proposed method achieves better dynamic response and disturbance-rejection performance than conventional LADRC and other benchmark controllers. Full article

Unauthorized unmanned aerial vehicle (UAV) intrusions in sensitive areas such as airports have made accurate UAV detection and localization a pressing need. Acoustic sensing is passive and weather-independent, but conventional microphone arrays require many elements and a large aperture. This paper proposes an acoustic vector sensor (AVS)-based method, termed Covariance Enhancement and Confidence-guided Tracking for 3D Acoustic Localization (CECT-3DAL). A single AVS measures the sound pressure and three-axis particle velocity at one point. Adaptive diagonal loading improves the robustness of the covariance matrix at a low signal-to-noise ratio (SNR). An exponential spectral enhancement strategy sharpens the spatial spectrum peaks for direction estimation, and an eigenvalue-ratio-based confidence drives confidence-weighted smoothing of the angle sequences. Meanwhile, a dual-sensor geometric model provides a closed-form three-dimensional solution. In simulations, the azimuth and elevation root-mean-square errors (RMSEs) were below 1.5° for SNR above 4 dB. In an anechoic chamber, confidence-weighted smoothing reduced the azimuth and elevation standard deviations from 4.34° and 2.63° to 1.46° and 0.86°. In field experiments, the hovering azimuth stayed within a 90% span of 2–3.5°, with an average horizontal RMSE of 0.209 m against a GPS reference, and trajectories under various flight modes remained continuous and smooth. The proposed method offers a compact, passive, and low-cost solution for counter-UAV acoustic surveillance. Full article