Search Results (282)

Multi-agent inverse reinforcement learning (MA-IRL) infers the underlying reward functions or objectives of multiple agents by observing their behavioral data, thereby providing insights into collaboration, competition, or mixed interaction strategies among agents, and addressing the symmetrical ambiguity problem where multiple rewards may correspond to the same strategy. However, most existing algorithms mainly focus on solving cooperative and non-cooperative tasks among homogeneous multi-agent systems, making it difficult to adapt to the dynamic topologies and heterogeneous behavioral strategies of multi-agent systems in real-world applications. This makes it difficult for the algorithm to adapt to scenarios with locally sparse interactions and dynamic heterogeneity, such as autonomous driving, drone swarms, and robot clusters. To address this problem, this study proposes a dynamic heterogeneous multi-agent inverse reinforcement learning framework (GAMF-DHIRL) based on a graph attention mean field (GAMF) to infer the potential reward functions of agents. In GAMF-DHIRL, we introduce a graph attention mean field theory based on adversarial maximum entropy inverse reinforcement learning to dynamically model dependencies between agents and adaptively adjust the influence weights of neighboring nodes through attention mechanisms. Specifically, the GAMF module uses a dynamic adjacency matrix to capture the time-varying characteristics of the interactions among agents. Meanwhile, the typed mean-field approximation reduces computational complexity. Experiments demonstrate that the proposed method can efficiently recover reward functions of heterogeneous agents in collaborative tasks and adversarial environments, and it outperforms traditional MA-IRL methods. Full article

A clustering ensemble, which leverages multiple base clusterings to obtain a reliable consensus result, is a critical challenging task for Earth observation in remote sensing applications. With the development of multi-source remote sensing data, exploring the underlying graph-structured patterns has become increasingly important. However, existing clustering ensemble methods mostly employ shallow clustering in the base clustering generation stage, which fails to utilize the structural information. Moreover, the high dimensionality inherent in data further increases the difficulty of clustering. To address these problems, we propose a novel method termed Global Self-Attention-driven Graph Clustering Ensemble (GSAGCE). Specifically, GSAGCE firstly adopts basic autoencoders and global self-attention graph autoencoders (GSAGAEs) to extract node attribute information and structural information, respectively. GSAGAEs not only enhance structural information in the embedding but also have the capability to capture long-range vertex dependencies. Next, we employ a fusion strategy to adaptively fuse this dual information by considering the importance of nodes through an attention mechanism. Furthermore, we design a self-supervised strategy to adjust the clustering distribution, which integrates the attribute and structural embeddings as more reliable guidance to produce base clusterings. In the ensemble strategy, we devise a double-weighted graph partitioning consensus function that simultaneously considers both global and local diversity within the base clusterings to enhance the consensus performance. Extensive experiments on benchmark datasets demonstrate the superiority of GSAGCE compared to other state-of-the-art methods. Full article

This study presents a comparative analysis of two bio-inspired optimization techniques: the Dragonfly Algorithm (DA) and Cuckoo Search (CS). The DA models the collective behavior of dragonflies, replicating dynamic processes such as foraging, evasion, and synchronized movement to effectively explore and exploit the solution space. In contrast, the CS algorithm draws inspiration from the brood parasitism strategy observed in certain Cuckoo species, where eggs are laid in the nests of other birds, thereby leveraging randomization and selection mechanisms for optimization. To enhance the performance of both algorithms, Type-2 fuzzy logic systems were integrated into their structures. Specifically, the DA was fine-tuned through the adjustment of its inertia weight (W) and attraction coefficient (Beta), while the CS algorithm was optimized by calibrating the Lévy flight distribution parameter. A comprehensive set of benchmark functions, F1 through F10, was employed to evaluate and compare the effectiveness and convergence behavior of each method under fuzzy-enhanced configurations. Results indicate that the fuzzy-based adaptations consistently improved convergence stability and accuracy, demonstrating the advantage of integrating Type-2 fuzzy parameter control into swarm-based optimization frameworks. Full article

Accurate vehicle localization remains a critical challenge due to the frequent loss or degradation of sensor data, such as from visual, inertial, and GPS sources. In this study, we present a novel localization algorithm that dynamically fuses data from heterogeneous sensors to achieve stable and precise positioning. The proposed algorithm integrates a deep learning-based visual-inertial odometry (VIO) module with a Kalman filter for global data fusion. A key innovation of the method is its adaptive fusion strategy, which adjusts feature weights based on sensor reliability, thereby ensuring optimal data utilization. Extensive experiments across varied scenarios demonstrate the algorithm’s superior performance, consistently achieving lower RMSE values and reducing position errors by 79–91% compared to four state-of-the-art baselines—even under adverse conditions such as sensor failures or missing data. This work lays the foundation for deploying robust localization systems in real-world applications, including autonomous vehicles, robotics, and navigation technologies. Full article

Virtual synchronous generator (VSG) simulates the dynamic characteristics of synchronous generator, offering significant advantages in flexibly adjusting virtual inertia and damping parameters. However, their dynamic stability is susceptible to constraints such as control parameter design, grid disturbances, and the intermittent nature of distributed power sources. This study addresses the degradation of transient performance in traditional sliding mode control for VSG, caused by insufficient multi-parameter cooperative adaptation. It proposes an adaptive sliding mode control strategy based on radial basis function (RBF) neural networks. Through theoretical analysis of the influence mechanism of virtual inertia and damping coefficient perturbations on system stability, the RBF neural network achieves dynamic parameter decoupling and nonlinear mapping. Combined with an integral-type sliding surface to design a weight-adaptive convergence law, it effectively avoids local optima and ensures global stability. This strategy not only enables multi-parameter cooperative adaptive regulation of frequency fluctuations but also significantly enhances the system’s robustness under parameter perturbations. Simulation results demonstrate that compared to traditional control methods, the proposed strategy exhibits significant advantages in dynamic response speed and overshoot suppression. Full article

Simultaneous Localization and Mapping (SLAM) utilizes sensor data to concurrently construct environmental maps and estimate its own position, finding wide application in scenarios like robotic navigation and augmented reality. SLAM systems based on 3D Gaussian Splatting (3DGS) have garnered significant attention due to their real-time, high-fidelity rendering capabilities. However, in real-world environments containing dynamic objects, existing 3DGS-SLAM methods often suffer from mapping errors and tracking drift due to dynamic interference. To address this challenge, this paper proposes BDGS-SLAM—a Bayesian Dynamic Gaussian Splatting SLAM framework specifically designed for dynamic environments. During the tracking phase, the system integrates semantic detection results from YOLOv5 to build a dynamic prior probability model based on Bayesian filtering, enabling accurate identification of dynamic Gaussians. In the mapping phase, a multi-view probabilistic update mechanism is employed, which aggregates historical observation information from co-visible keyframes. By introducing an exponential decay factor to dynamically adjust weights, this mechanism effectively restores static Gaussians that were mistakenly culled. Furthermore, an adaptive dynamic Gaussian optimization strategy is proposed. This strategy applies penalizing constraints to suppress the negative impact of dynamic Gaussians on rendering while avoiding the erroneous removal of static Gaussians and ensuring the integrity of critical scene information. Experimental results demonstrate that, compared to baseline methods, BDGS-SLAM achieves comparable tracking accuracy while generating fewer artifacts in rendered results and realizing higher-fidelity scene reconstruction. Full article

Unmanned Aerial Vehicle (UAV)-based aerial object detection has been widely applied in various fields, including logistics, public security, disaster response, and smart agriculture. However, numerous small objects in UAV aerial images are often overwhelmed by large-scale complex backgrounds, making their appearance difficult to distinguish and thereby prone to being missed by detectors. To tackle these issues, we propose a novel Contextual-Semantic Interactive Perception Network (CSIPN) for small object detection in UAV aerial scenarios, which enhances detection performance through scene interaction modeling, dynamic context modeling, and dynamic feature fusion. The core components of the CSIPN include the Scene Interaction Modeling Module (SIMM), the Dynamic Context Modeling Module (DCMM), and the Semantic-Context Dynamic Fusion Module (SCDFM). Specifically, the SIMM introduces a lightweight self-attention mechanism to generate a global scene semantic embedding vector, which then interacts with shallow spatial descriptors to explicitly depict the latent relationships between small objects and complex background, thereby selectively activating key spatial responses. The DCMM employs two dynamically adjustable receptive-field branches to adaptively model contextual cues and effectively supplement the contextual information required for detecting various small objects. The SCDFM utilizes a dual-weighting strategy to dynamically fuse deep semantic information with shallow contextual details, highlighting features relevant to small object detection while suppressing irrelevant background. Our method achieves mAPs of 37.2%, 93.4%, 50.8%, and 48.3% on the TinyPerson dataset, the WAID dataset, the VisDrone-DET dataset, and our self-built WildDrone dataset, respectively, while using only 25.3M parameters, surpassing existing state-of-the-art detectors and demonstrating its superiority and robustness. Full article

With the continuous growth of user reviews, identifying underlying sentiment across multi-source texts efficiently and accurately has become a significant challenge in NLP. Traditional single models in cross-domain sentiment analysis often exhibit insufficient stability, limited generalization capabilities, and sensitivity to class imbalance. Existing ensemble methods predominantly rely on static weighting or voting strategies among homogeneous models, failing to fully leverage the complementary advantages between models. To address these issues, this study proposes a heterogeneous ensemble sentiment classification model integrating multi-view features and dynamic weighting. At the feature learning layer, the model constructs three complementary base learners, a RoBERTa-FC for extracting global semantic features, a BERT-BiGRU for capturing temporal dependencies, and a TextCNN-Attention for focusing on local semantic features, thereby achieving multi-level text representation. At the decision layer, a meta-learner is used to fuse multi-view features, and dynamic uncertainty weighting and attention weighting strategies are employed to adaptively adjust outputs from different base learners. Experimental results across multiple domains demonstrate that the proposed model consistently outperforms single learners and comparison methods in terms of Accuracy, Precision, Recall, F1 Score, and Macro-AUC. On average, the ensemble model achieves a Macro-AUC of 0.9582 ± 0.023 across five datasets, with an Accuracy of 0.9423, an F1 Score of 0.9590, and a Macro-AUC of 0.9797 on the AlY_ds dataset. Moreover, in cross-dataset ranking evaluation based on equally weighted metrics, the model consistently ranks within the top two, confirming its superior cross-domain adaptability and robustness. These findings highlight the effectiveness of the proposed framework in enhancing sentiment classification performance and provide valuable insights for future research on lightweight dynamic ensembles, multilingual, and multimodal applications. Full article

Accurate prediction of fracturing pressure is critical for operational safety and fracturing efficiency in unconventional reservoirs. Traditional physics-based models and existing deep learning architectures often struggle to capture the intense fluctuations and complex temporal dependencies observed in actual fracturing operations. To address these challenges, this paper proposes a multi-source data-driven fracturing pressure prediction model, a model integrating TCN-BiLSTM-Attention Mechanism (Temporal Convolutional Network, Bidirectional Long Short-Term Memory, Attention Mechanism), and introduces a feature selection mechanism for fracture pressure prediction. This model employs TCN to extract multi-scale local fluctuation features, BiLSTM to capture long-term dependencies, and Attention to adaptively adjust feature weights. A two-stage feature selection strategy combining correlation analysis and ablation experiments effectively eliminates redundant features and enhances model robustness. Field data from the Sichuan Basin were used for model validation. Results demonstrate that our method significantly outperforms baseline models (LSTM, BiLSTM, and TCN-BiLSTM) in mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R²), particularly under high-fluctuation conditions. When integrated with slope reversal analysis, it achieves sand blockage warnings up to 41 s in advance, offering substantial potential for real-time decision support in fracturing operations. Full article

Urban water supply networks are prone to complex noise interference, which significantly degrades the performance of data-driven forecasting models. Conventional denoising techniques, such as standard Variational Mode Decomposition (VMD), often rely on empirical parameter selection or optimize only a subset of parameters, lacking a robust mechanism for identifying noise-dominant components post-decomposition. To address these issues, this paper proposed a novel denoising framework termed Dynamic Hyperbolic Tangent PSO-optimized VMD (DHTPSO-VMD). The DHTPSO algorithm adaptively adjusts inertia weights and cognitive/social learning factors during iteration, mitigating the local optima convergence typical of traditional PSO and enabling automated VMD parameter selection. Furthermore, a dual-criteria screening strategy based on Variance Contribution Rate (VCR) and Correlation Coefficient Metric (CCM) is employed to accurately identify and eliminate noise-related Intrinsic Mode Functions (IMFs). Validation using pressure data from District A in Zhejiang Province, China, demonstrated that the proposed DHTPSO-VMD method significantly outperforms benchmark approaches (PSO-VMD, EMD, SABO-VMD, GWO-VMD) in terms of Signal-to-Noise Ratio (SNR), Mean Absolute Error (MAE), and Mean Square Error (MSE). Subsequent forecasting experiments using an Informer model showed that signals preprocessed with DHTPSO-VMD achieved superior prediction accuracy (R² = 0.948924), underscoring its practical utility for smart water supply management. Full article

Building clustering is a key challenge in cartographic generalization, where the goal is to group spatially related buildings into semantically coherent clusters while preserving the true distribution patterns of urban structures. Existing methods often rely on either spatial distance or building feature similarity alone, leading to clusters that sacrifice either accuracy or spatial continuity. Moreover, most deep learning-based approaches, including graph attention networks (GATs), fail to explicitly incorporate spatial distance constraints and typically restrict message passing to first-order neighborhoods, limiting their ability to capture long-range structural dependencies. To address these issues, this paper proposes LA-GATs, a multi-feature constrained and spatially adaptive building clustering network. First, a Delaunay triangulation is constructed based on nearest-neighbor distances to represent spatial topology, and a heterogeneous feature matrix is built by integrating architectural spatial features, including compactness, orientation, color, and height. Then, a spatial distance-constrained attention mechanism is designed, where attention weights are adjusted using a distance decay function to enhance local spatial correlation. A second-order neighborhood aggregation strategy is further introduced to extend message propagation and mitigate the impact of triangulation errors. Finally, spectral clustering is performed on the learned similarity matrix. Comprehensive experimental validation on real-world datasets from Xi’an and Beijing, showing that LA-GATs outperforms existing clustering methods in both compactness, silhouette coefficient and adjusted rand index, with up to about 21% improvement in residential clustering accuracy. Full article

Enhancing ultra-capacitor (UC) utilization and mitigating battery stress are pivotal for improving the energy management efficiency and service life of hybrid energy storage systems (HESSs). Conventional energy management strategies (EMSs), however, rely on fixed parameters and therefore struggle to allocate power flexibly or reduce battery degradation. This paper proposes a VMD-LSTM-based EMS that incorporates auto-tuning weight and constraint to address these limitations. First, a VMD-LSTM predictor was proposed to improve the velocity and road gradient prediction accuracy, thus leading an accurate power demand for EMS and enabling real-time parameter adaptation, especially in the nonlinear area. Second, the model predictive controller (MPC) was adopted to construct the EMS by solving a multi-objective problem using quadratic programming. Third, a combination of rule-based and fuzzy logic-based strategies was introduced to adjust the weights and constraints, optimizing UC utilization while alleviating the burden on batteries. Simulation results show that the proposed scheme boosts UC utilization by 10.98% and extends battery life by 19.75% compared to traditional MPC. These gains underscore the practical viability of intelligent, optimizing EMSs for HESSs. Full article

Physics-Informed Neural Networks (PINNs) have emerged as a promising paradigm for solving partial differential equations (PDEs) by embedding physical laws into the learning process. However, standard PINNs often suffer from training instabilities and unbalanced optimization when handling multi-term loss functions, especially in problems involving singular perturbations, fractional operators, or multi-scale behaviors. To address these limitations, we propose a novel gradient variance weighting physics-informed neural network (GVW-PINN), which adaptively adjusts the loss weights based on the variance of gradient magnitudes during training. This mechanism balances the optimization dynamics across different loss terms, thereby enhancing both convergence stability and solution accuracy. We evaluate GVW-PINN on three representative PDE models and numerical experiments demonstrate that GVW-PINN consistently outperforms the conventional PINN in terms of training efficiency, loss convergence, and predictive accuracy. In particular, GVW-PINN achieves smoother and faster loss reduction, reduces relative errors by one to two orders of magnitude, and exhibits superior generalization to unseen domains. The proposed framework provides a robust and flexible strategy for applying PINNs to a wide range of integer- and fractional-order PDEs, highlighting its potential for advancing data-driven scientific computing in complex physical systems. Full article

The reconfiguration of distribution networks plays a crucial role in minimizing active power losses and enhancing reliability, but the problem becomes increasingly complex with the integration of distributed generation (DG). Traditional optimization methods and even earlier hybrid metaheuristics often suffer from premature convergence or require problem reformulations that compromise feasibility. To overcome these limitations, this paper proposes a novel hybrid algorithm that couples Particle Swarm Optimization (PSO) with Simulated Annealing (SA) through an adaptive inertia weight mechanism derived from the Lundy–Mees cooling schedule. Unlike prior hybrid approaches, our method directly addresses the original non-convex, combinatorial nature of the Distribution Network Reconfiguration (DNR) problem without convexification or post-processing adjustments. The main contributions of this study are fourfold: (i) proposing a PSO-SA hybridization strategy that enhances global exploration and avoids stagnation; (ii) introducing an adaptive inertia weight rule tuned by SA, more effective than traditional schemes; (iii) applying a stagnation-based stopping criterion to speed up convergence and reduce computational cost; and (iv) validating the approach on 5-, 33-, and 69-bus systems, with and without DG, showing robustness, recurrence rates above 80%, and low variability compared to conventional PSO. Simulation results confirm that the proposed PSO-SA algorithm achieves superior performance in both loss minimization and solution stability, positioning it as a competitive and scalable alternative for modern active distribution systems. Full article

Fault prediction and diagnosis are critical for enhancing the maintenance and reliability of power system equipment, reducing operational costs, and preventing potential failures. In power transformers, periodic oil sampling and gas ratio analysis provide valuable insights for predictive maintenance and life-cycle assessment. Machine learning methods, such as XGBoost, have proven to deliver more accurate results, especially when historical data is limited. However, the performance of XGBoost is highly dependent on the optimization of its hyperparameters. To address this, this paper proposes an improved Particle Swarm Optimization (IPSO) method to optimize the hyperparameters of XGBoost for transformer fault diagnosis. The PSO algorithm is enhanced by introducing topology optimization, adaptively adjusting the acceleration factor, dividing the swarm into master–slave particle groups to strengthen search capability, and dynamically adjusting inertia weights using a linear adaptive strategy. IPSO is applied to optimize key hyperparameters of the XGBoost model, improving both its diagnostic accuracy and generalization ability. Experimental results confirm the effectiveness of the proposed model in enhancing fault prediction and diagnosis in power systems. Full article