Review of Satellite Remote Sensing of Carbon Dioxide Inversion and Assimilation

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(18), P. 3394 - 3394

Published: Sept. 12, 2024

With the rapid development of satellite remote sensing technology, carbon-cycle research, as a key focus global climate change, has also been widely developed in terms carbon source/sink-research methods. The internationally recognized “top-down” approach, which is based on observations, an important means to verify greenhouse gas-emission inventories. This article reviews principles, categories, and detection payloads for gases introduces inversion algorithms datasets XCO2. It emphasizes methods machine learning assimilation algorithms. Additionally, it presents technology achievements carbon-assimilation systems used estimate fluxes. Finally, summarizes prospects future improve accuracy estimating monitoring Earth’s processes.

Language: Английский

---

OSNet: An Edge Enhancement Network for a Joint Application of SAR and Optical Images

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(3), P. 505 - 505

Published: Jan. 31, 2025

The combined use of synthetic aperture radar (SAR) and optical images for surface observation is gaining increasing attention. Optical images, with their distinct edge features, can accurately classify different objects, while SAR reveal deeper internal variations. To address the challenge differing feature distributions in multi-source we propose an enhancement network, OSNet (network images), designed to jointly extract features from enhance representation. consists three core modules: a dual-branch backbone, synergistic attention integration module, global-guided local fusion module. These modules, respectively, handle modality-independent extraction, sharing, global-local fusion. In backbone introduce differentiable Lee filter Laplacian detection operator branch suppress noise features. Additionally, module facilitate cross-modal information exchange between two branches. We validated OSNet’s performance on segmentation tasks (WHU-OPT-SAR) regression (SNOW-OPT-SAR). results show that improved PA MIoU by 2.31% 2.58%, task, reduced MAE RMSE 3.14% 4.22%, task.

Language: Английский

---

Multi-agent reinforcement learning for energy management in microgrids with shared hydrogen storage

International Journal of Hydrogen Energy, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 1, 2025

Language: Английский

---

Autonomous Decision-Making Method for Uav Formations with Dynamically Switchable Targets for Open Environments

Published: Jan. 1, 2025

Language: Английский

---

Speech Emotion Recognition Based on CNN-Transformer with Different Loss Function

Journal of Computer and Communications, Journal Year: 2025, Volume and Issue: 13(03), P. 103 - 115

Published: Jan. 1, 2025

Language: Английский

---

Reliability assessment of multi-agent reinforcement learning algorithms for hybrid local electricity market simulation

Applied Energy, Journal Year: 2025, Volume and Issue: 389, P. 125789 - 125789

Published: March 30, 2025

Language: Английский

---

A Survey on Multi-UAV Path Planning: Classification, Algorithms, Open Research Problems, and Future Directions

Drones, Journal Year: 2025, Volume and Issue: 9(4), P. 263 - 263

Published: March 31, 2025

Multi-UAV path planning algorithms are crucial for the successful design and operation of unmanned aerial vehicle (UAV) networks. While many network researchers have proposed UAV to improve system performance, an in-depth review multi-UAV has not been fully explored yet. The purpose this study is survey, classify, compare existing in literature over last eight years various scenarios. After detailing classification, we based on time consumption, computational cost, complexity, convergence speed, adaptability. We also examine approaches, including metaheuristic, classical, heuristic, machine learning, hybrid methods. Finally, identify several open research problems further investigation. More required smart that can re-plan pathways fly real complex Therefore, aims provide insight into engineers contribute next-generation systems.

Language: Английский

---

Application of the Capsule Network With Attention-Mechanism-Based Dynamic Routing in the Analysis of Utopian Online Literature

International Journal of Knowledge Management, Journal Year: 2025, Volume and Issue: 21(1), P. 1 - 23

Published: April 3, 2025

This study applies the dynamic routing algorithm based on attention mechanism to capsule network model, aiming address problem of multi - label entity relation extraction in utopian online literature. The research constructs an improved model and elaborates relevant theories, technical foundations, construction process. Through experiments, by comparing with models such as convolutional neural average pooling (Avg pooling+CNN), recurrent (RNN), (Att+RNN), results show that Att+capsule performs better terms indicators precision, recall, F1 score, is also more stable during training case analysis further verifies effectiveness extraction, theme mining, sentiment analysis. Future can be carried out directions expanding dataset, optimizing structure, application fields.

Language: Английский

---

Detection and grading of oxidation for copper–water heat pipe wicks based on the machine learning methods

Applied Thermal Engineering, Journal Year: 2025, Volume and Issue: unknown, P. 126437 - 126437

Published: April 1, 2025

Language: Английский

---

Predicting e-commerce product prices through the integration of variational mode decomposition and deep neural networks

PeerJ Computer Science, Journal Year: 2024, Volume and Issue: 10, P. e2353 - e2353

Published: Oct. 8, 2024

Product prices frequently manifest nonlinear and nonstationary time-series attributes, indicating potential variations in their behavioral patterns over time. Conventional linear models may fall short adequately capturing these intricate properties. In addressing this, the present study leverages adaptive non-recursive attributes of Variational Mode Decomposition (VMD) methodology. It employs VMD to dissect time series into multiple Intrinsic Functions (IMF). Subsequently, a method rooted minimum fuzzy entropy criterion is introduced for determining optimal modal number (K) decomposition process. This effectively mitigates issues related confusion endpoint effects, thereby enhancing efficacy VMD. subsequent phase, deep neural networks (DNN) are harnessed forecast identified modes, with cumulative predictions yielding ultimate e-commerce product price prognostications. The predictive proposed Decomposition-deep network (VMD-DNN) model assessed on three public datasets, wherein mean absolute percentage error (MAPE) E-commerce Price Prediction Dataset Online Retail notably low at 0.6578 0.5414, respectively. corresponds remarkable reduction rate 66.5% 70.4%. Moreover, VMD-DNN excels predicting through DNN, amplifying capability by 4%. attains superior results terms directional symmetry, boasting highest Directional Symmetry (DS) score 86.25. Notably, forecasted trends across diverse ranges closely mirror actual trends.

Language: Английский

---