Search Results (118)

The growing penetration of distributed photovoltaic (PV) systems presents significant operational challenges for power grids, driven by the scarcity of historical data and the high spatiotemporal variability of PV generation. To address these challenges, we propose Generative Reconstruction and Adaptive Identification via Latents (GRAIL), a unified, end-to-end framework that integrates generative modeling with adaptive clustering to discover latent structures and representative scenarios in PV datasets. GRAIL operates through a closed-loop mechanism where clustering feedback guides a cluster-aware data generation process, and the resulting generative augmentation strengthens partitioning in the latent space. Evaluated on a real-world, multi-site PV dataset with a high missing data rate of $45.4\%$, GRAIL consistently outperforms both classical clustering algorithms and deep embedding-based methods. Specifically, GRAIL achieves a Silhouette Score of 0.969, a Calinski–Harabasz index exceeding $4.132\times {10}^6$, and a Davies–Bouldin index of $0.042$, demonstrating superior intra-cluster compactness and inter-cluster separation. The framework also yields a normalized entropy of $0.994$, which indicates highly balanced partitioning. These results underscore that coupling data generation with clustering is a powerful strategy for expressive and robust structure learning in data-sparse environments. Notably, GRAIL achieves significant performance gains over the strongest deep learning baseline that lacks a generative component, securing the highest composite score among all evaluated methods. The framework is also computationally efficient. Its alternating optimization converges rapidly, and clustering and reconstruction metrics stabilize within approximately six iterations. Beyond quantitative performance, GRAIL produces physically interpretable clusters that correspond to distinct weather-driven regimes and capture cross-site dependencies. These clusters serve as compact and robust state descriptors, valuable for downstream applications such as PV forecasting, dispatch optimization, and intelligent energy management in modern power systems. Full article

Maritime safety, environmental protection, and efficient traffic management increasingly rely on data-driven technologies. However, leveraging Automatic Identification System (AIS) data for predictive modelling faces two major challenges: the massive volume of data generated in real-time and growing privacy concerns associated with proprietary vessel information. This paper proposes a novel, privacy-preserving framework for vessel traffic density (VTD) prediction that addresses both challenges. The approach combines the European Maritime Observation and Data Network’s (EMODNet) grid-based VTD calculation method with Convolutional Neural Networks (CNN) to model spatiotemporal traffic patterns and employs Federated Learning to collaboratively build a global predictive model without the need for explicit sharing of proprietary AIS data. Three geographically diverse AIS datasets were harmonized, processed, and used to train local CNN models on hourly VTD matrices. These models were then aggregated via a Federated Learning framework under a lifelong learning scenario. Evaluation using Sparse Mean Squared Error shows that the federated global model achieves promising accuracy in sparse data scenarios and maintains performance parity when compared with local CNN-based models, all while preserving data privacy and minimizing hardware performance needs and data communication overheads. The results highlight the approach’s effectiveness and scalability for real-world maritime applications in traffic forecasting, safety, and operational planning. Full article

Single-phase ground (SPG) faults pose significant challenges in three-phase four-wire distribution systems due to their complex transient characteristics and the presence of multiple influencing factors. To solve the aforementioned issues, a comprehensive fault identification framework is proposed, which uses the TabNet deep learning architecture with hyperparameters optimized by Optuna. Firstly, a 10 kV simulation model is developed in Simulink to generate a diverse fault dataset. For each simulated fault, voltage and current signals from eight channels (L1–L4 voltage and current) are collected. Secondly, multi-domain features are extracted from each channel across time, frequency, waveform, and wavelet perspectives. Then, an attention-based fusion mechanism is employed to capture cross-channel dependencies, followed by L2-norm-based feature selection to enhance generalization. Finally, the optimized TabNet model effectively classifies 24 fault categories, achieving an accuracy of 97.33%, and outperforms baseline methods including Temporal Convolutional Network (TCN), Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM), Capsule Network with Sparse Filtering (CNSF), and Dual-Branch CNN in terms of accuracy, macro-F1 score, and kappa coefficient. It also exhibits strong stability and fast convergence during training. These results demonstrate the robustness and interpretability of the proposed method for SPG fault detection. Full article

Monitoring multimode industrial processes presents significant challenges due to varying operating conditions, nonlinear dynamics, and mode-dependent feature distributions. This paper proposes a novel process monitoring framework that integrates an enhanced Gaussian Mixture Model (GMM) for mode identification with Label Consistent K-SVD (LC-KSVD) for sparse dictionary learning. The improved GMM employs a parallelized Expectation–Maximization algorithm to achieve accurate and scalable mode partitioning in high-dimensional environments. Subsequently, the LC-KSVD then learns label-consistent, discriminative sparse representations, enabling effective monitoring across modes. The proposed method is evaluated through a simulation study and the widely used Continuous Stirred Tank Heater (CSTH) benchmark. Comparative results with traditional techniques such as LNS-PCA and FGMM demonstrate that the proposed method achieves superior fault detection rates (FDRs) and significantly lower false alarm rates (FARs), even under complex mode transitions and mild fault scenarios. Furthermore, the method also provides interpretable fault isolation through reconstruction-error-guided variable contribution analysis. These findings confirm that the proposed LC-KSVD-based scheme offers a reliable solution for fault detection and isolation in multimode process systems. Full article

Foreign object intrusion in railway perimeter areas poses significant risks to train operation safety. To address the limitation of current visual detection technologies that overly focus on target identification while lacking quantitative risk assessment, this paper proposes a railway intrusion risk quantification method integrating track semantic segmentation and spatiotemporal features. An improved BiSeNetV2 network is employed to accurately extract track regions, while physical-constrained risk zones are constructed based on railway structure gauge standards. The lateral spatial distance of intruding objects is precisely calculated using track gauge prior knowledge. A lightweight detection architecture is designed, adopting ShuffleNetV2 as the backbone to reduce computational complexity, with an incorporated Dilated Transformer module to enhance global context awareness and sparse feature extraction, significantly improving detection accuracy for small-scale objects. The comprehensive risk assessment formula integrates object category weights, lateral risk coefficients in intrusion zones, longitudinal distance decay factors, and dynamic velocity compensation. Experimental results demonstrate that the proposed method achieves 84.9% mean average precision (mAP) on our proprietary dataset, outperforming baseline models by 3.3%. By combining lateral distance detection with multidimensional risk indicators, the method enables quantitative intrusion risk assessment and graded early warning, providing data-driven decision support for active train protection systems and substantially enhancing intelligent safety protection capabilities. Full article

Phewa Lake, Nepal’s second-largest natural lake, is under increasing ecological stress due to sedimentation, shoreline encroachment, and water quality decline driven by rapid urban growth, fragile mountainous catchments, and changing climate patterns. This study employs an integrated approach combining sediment yield estimation from its catchment using RUSLE, shoreline encroachment analysis via satellite imagery and historical records, and identification of pollution sources and socio-economic factors through field surveys and community consultations. The results show that steep, sparsely vegetated slopes are the primary sediment sources, with Harpan Khola (a tributary of Phewa Lake) contributing over 80% of the estimated 339,118 tons of annual sediment inflow. From 1962 to 2024, the lake has lost approximately 5.62 sq. km of surface area, primarily due to a combination of sediment deposition and human encroachment. Pollution from untreated sewage, urban runoff, and invasive aquatic weeds further degrades water quality and threatens biodiversity. Based on the findings, this study proposes a way forward to mitigate sedimentation, encroachment, and pollution, along with a sustainable revitalization plan. The approach of this study, along with the proposed sustainability measures, can be replicated in other lake systems within Nepal and in similar watersheds elsewhere. Full article

Maximizing the power output of wind farms is critical for improving the economic viability and grid integration of renewable energy. Active wake control (AWC) strategies, such as yaw-based wake steering, offer significant potential for power generation increase but require predictive models that are both accurate and computationally efficient for real-time implementation. This paper proposes a data-driven observer to rapidly estimate the potential power gain achievable through AWC as a function of the ambient wind direction. The approach is rooted in Koopman operator theory, which allows a linear representation of nonlinear dynamics. Specifically, a model is developed using an Input–Output Extended Dynamic Mode Decomposition framework combined with Sparse Identification (IOEDMDSINDy). This method lifts the low-dimensional wind direction input into a high-dimensional space of observable functions and then employs iterative sparse regression to identify a minimal, interpretable linear model in this lifted space. By training on offline simulation data, the resulting observer serves as an ultra-fast surrogate model, capable of providing instantaneous predictions to inform online control decisions. The methodology is demonstrated and its performance is validated using two case studies: a 9-turbine and a 20-turbine wind farm. The results show that the observer accurately captures the complex, nonlinear relationship between wind direction and power gain, significantly outperforming simpler models. This work provides a key enabling technology for advanced, real-time wind farm control systems. Full article

An important historical and cultural region in southern China, the Xiangjiang River Basin, has formed a unique spatial pattern and regional cultural characteristics in its long-term development. In recent years, the acceleration of urbanization has led to the historical texture and cultural elements of Changsha’s suburban blocks facing deconstruction pressure. How to identify and protect their cultural value at the spatial structure level has become an urgent issue. Taking three typical traditional township blocks in the suburbs of Changsha as the research object, this paper constructs a trinity research framework of “spatial gene identification–diversity analysis–strategy optimization.” It systematically discusses the makeup of the types, quantity, distribution, relative importance ranking, and diversity characteristics of their spatial genes. The results show that (1) the distribution and quantity of spatial genes are affected by multiple driving forces such as historical function, geographic environment, and settlement evolution mechanisms, and that architectural spatial genes have significant advantages in type richness and importance indicators; (2) spatial gene diversity shows the structural characteristics of “enriched artificial space and sparse natural space,” and different blocks show clear differences in node space and boundary space; (3) spatial genetic diversity not only reflects the complexity of the spatial evolution of a block but is also directly related to its cultural inheritance and the feasibility of renewal strategies. Based on this, this paper proposes strategies such as building a spatial gene database, improving the diversity evaluation system, and implementing differentiated protection mechanisms. These strategies provide theoretical support and methods for the protection and sustainable development of cultural heritage in traditional blocks. Full article

Harmful algal blooms (HABs) are one of the major environmental concerns, as they have various negative effects on public and environmental health, recreational services, and economics. HAB modeling is challenging due to inconsistent and insufficient data, as well as the nonlinear nature of algae formation data. However, it is crucial for attaining sustainable development goals related to clean water and sanitation. From this point of view, we employed the sparse identification nonlinear dynamics (SINDy) technique to model microcystin, an algal toxin, utilizing dissolved oxygen as a water quality metric and evaporation as a meteorological parameter. SINDy is a novel approach that combines a sparse regression and machine learning method to reconstruct the analytical representation of a dynamical system. The model results indicate that MAPE values of approximately 2% were achieved in three out of four lakes, while the MAPE value of the remaining lake is 11%. Moreover, a model-driven and web-based interactive tool was created to develop environmental education, raise public awareness on HAB events, and produce more effective solutions to HAB problems through what-if scenarios. This interactive and user-friendly web platform allows tracking the status of HABs in lakes and observing the impact of specific parameters on harmful algae formation. Full article

This study focuses on the systematic conservation of historical architectural heritage in Heilongjiang Province, particularly addressing the challenges of point-based protection and spatial fragmentation. It explores the construction of a connected and conductive heritage corridor network, using historical building clusters across the province as empirical cases. A comprehensive analytical framework is established by integrating the nearest neighbor index, kernel density estimation, minimum cumulative resistance (MCR) model, entropy weighting, circuit theory, and network structure metrics. Kernel density analysis reveals a distinct spatial aggregation pattern, characterized by “one core, multiple zones.” Seven resistance factors—including elevation, slope, land use, road networks, and service accessibility—are constructed, with weights assigned through an entropy-based method to generate an integrated resistance surface and suitability map. Circuit theory is employed to simulate cultural “current” flows, identifying 401 potential corridors at the provincial, municipal, and district levels. A hierarchical station system is further developed based on current density, forming a coordinated structure of primary trunks, secondary branches, and complementary nodes. The corridor network’s connectivity is evaluated using graph-theoretic indices (α, β, and γ), which indicate high levels of closure, structural complexity, and accessibility. The results yield the following key findings: (1) Historical architectural resources in Heilongjiang demonstrate significant coupling with the Chinese Eastern Railway and multi-ethnic cultural corridors, forming a “one horizontal, three vertical” spatial configuration. The horizontal axis (Qiqihar–Harbin–Mudanjiang) aligns with the core cultural route of the railway, while the three vertical axes (Qiqihar–Heihe, Harbin–Heihe, and Mudanjiang–Luobei) correspond to ethnic cultural pathways. This forms a framework of “railway as backbone, ethnicity as wings.” (2) Comparative analysis of corridor paths, railways, and highways reveals structural mismatches in certain regions, including absent high-speed connections along northern trunk lines, insufficient feeder lines in secondary corridors, sparse terminal links, and missing ecological stations near regional boundaries. To address these gaps, a three-tier transportation coordination strategy is recommended: it comprises provincial corridors linked to high-speed rail, municipal corridors aligned with conventional rail, and district corridors connected via highway systems. Key enhancement zones include Yichun–Heihe, Youyi–Hulin, and Hegang–Wuying, where targeted infrastructure upgrades and integrated station hubs are proposed. Based on these findings, this study proposes a comprehensive governance paradigm for heritage corridors that balances multi-level coordination (provincial–municipal–district) with ecological planning. A closed-loop strategy of “identification–analysis–optimization” is developed, featuring tiered collaboration, cultural–ecological synergy, and multi-agent dynamic evaluation. The framework provides a replicable methodology for integrated protection and spatial sustainability of historical architecture in Heilongjiang and other cold-region contexts. Full article

Accurate localization of handheld ground-penetrating radar (HH-GPR) systems is critical for high-quality subsurface imaging and precise geospatial mapping of detected buried objects. In our previous works, we demonstrated that a UWB positioning system with an extended Kalman filter (EKF) employing a proprietary pendulum (PND) dynamics model yielded highly accurate results. Building on that foundation, we present a factor-graph-based estimation algorithm to further enhance the accuracy of HH-GPR antenna trajectory estimation. The system was modeled under realistic conditions, and both the EKF and various factor-graph algorithms were implemented. Comparative evaluation indicates that the factor-graph approach achieves an improvement in localization accuracy from over 30 to almost 50 percent compared to the EKF PND. The sparse matrix representation inherent in the factor graph enabled an efficient iterative solution of the underlying linearized system. This enhanced positioning accuracy is expected to facilitate the generation of clearer, more distinct underground images, thereby supporting the potential for more reliable identification and classification of buried objects and infrastructure. Full article

Self-driving technology aims to minimize human error and improve safety, efficiency, and mobility through advanced autonomous driving algorithms. Among these, Model Predictive Control (MPC) is highly valued for its optimization capabilities and ability to manage constraints. However, its effectiveness depends on an accurate system model, as modeling errors and disturbances can degrade performance, making uncertainty management crucial. Learning-based MPC addresses this challenge by adapting the predictive model to changing and unmodeled conditions. However, existing approaches often involve trade-offs: robust methods tend to be overly conservative, stochastic methods struggle with real-time feasibility, and deep learning lacks interpretability. Sparse regression techniques provide an alternative by identifying compact models that retain essential dynamics while eliminating unnecessary complexity. In this context, the Sparse Identification of Nonlinear Dynamics (SINDy) algorithm is particularly appealing, as it derives governing equations directly from data, balancing accuracy and computational efficiency. This work investigates the use of SINDy for learning and adapting vehicle dynamics models within an MPC framework. The methodology consists of three key phases. First, in offline identification, SINDy estimates the parameters of a three-degree-of-freedom single-track model using simulation data, capturing tire nonlinearities to create a fully tunable vehicle model. This is then validated in a high-fidelity CarMaker simulation to assess its accuracy in complex scenarios. Finally, in the online phase, MPC starts with an incorrect predictive model, which SINDy continuously updates in real time, improving performance by reducing lap time and ensuring a smoother trajectory. Additionally, a constrained version of SINDy is implemented to avoid obtaining physically meaningless parameters while aiming for an accurate approximation of the effects of unmodeled states. Simulation results demonstrate that the proposed framework enables an adaptive and efficient representation of vehicle dynamics, with potential applications to other control strategies and dynamical systems. Full article

Autonomous navigation is key to improving efficiency and addressing labor shortages in the fruit industry. Semi-structured orchards, with straight tree rows, dense weeds, thick canopies, and varying light conditions, pose challenges for tree identification and navigation line extraction. Traditional 3D lidars suffer from a narrow vertical FoV, sparse point clouds, and high costs. Furthermore, most lidar-based tree-row-detection algorithms struggle to extract high-quality navigation lines in scenarios with thin trunks and dense foliage occlusion. To address these challenges, we developed a 3D perception system using a servo motor to control the rolling motion of a 2D lidar, constructing 3D point clouds with a wide vertical FoV and high resolution. In addition, a method for trunk feature point extraction and tree row line detection for autonomous navigation has been proposed, based on trunk geometric features and RANSAC. Outdoor tests demonstrate the system’s effectiveness. At speeds of 0.2 m/s and 0.5 m/s, the average distance errors are 0.023 m and 0.016 m, respectively, while the average angular errors are 0.272° and 0.146°. This low-cost solution overcomes traditional lidar-based navigation method limitations, making it promising for autonomous navigation in semi-structured orchards. Full article

Compressive Sensing (CS) is a transformative signal processing framework that enables sparse signal acquisition at rates below the Nyquist limit, offering substantial advantages in data efficiency and reconstruction accuracy. This survey explores the theoretical foundations of CS, including sensing matrices, sparse bases, and recovery algorithms, with a focus on its applications in power engineering. CS has demonstrated significant potential in enhancing key areas such as state estimation (SE), fault detection, fault localization, outage identification, harmonic source identification (HSI), Power Quality Detection condition monitoring, and so on. Furthermore, CS addresses challenges in data compression, real-time grid monitoring, and efficient resource utilization. A case study on smart meter data recovery demonstrates the practical application of CS in real-world power systems. By bridging CS theory and its application, this survey underscores its potential to drive innovation, efficiency, and sustainability in power engineering and beyond. Full article

Rolling element bearings are key components for reducing friction and supporting rotors. Harsh working conditions contribute to the wear of bearings and consequent breakdown of machines, which leads to economic losses and even catastrophic accidents. Faulty impulses from bearings can excite resonance behavior in a system and produce modulation phenomena. Fault characteristics in modulated signals can be extracted using demodulation analysis methods, significantly improving the reliability and effectiveness of the fault diagnosis of rolling bearings. Optimal demodulation frequency band selection is a primary step for the demodulation-analysis-based fault diagnosis of bearing faults. To exploit the resonant modulation mechanism in the fault diagnosis of rolling element bearings, resonant frequencies identified through stochastic subspace identification are employed to guide the impulsive sparsity measures of an Autogram for bearing fault diagnosis, which combines physical modulation dynamics and data characteristics. The frequency band that not only matches the natural frequencies but also shows highly sparse impulsive characteristics is selected as the optimal demodulation frequency band for bearing fault diagnosis. The results of simulations and experimental data validate the advantages of the proposed method, which exploits physics-guided data processing for optimal demodulation frequency band determination. Full article