基於機器學習的三維地理信息系統天氣預報與可視化;Forecasting and Visualization of Weather in a 3D Geographical Information System based on Machine Learning

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/93216

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/93216

Title:	基於機器學習的三維地理信息系統天氣預報與可視化;Forecasting and Visualization of Weather in a 3D Geographical Information System based on Machine Learning
Authors:	王鴻恩;Wang, Hung-En
Contributors:	資訊工程學系
Keywords:	循環神經網絡;雙向長短期記憶;可視化;天氣預報;分類;數據分析;Recurrent Neural Network(RNN);Bi-directional Long Short-Term Memory(Bi-LSTM);Visualization;Weather Forecasting;Classification;Data Analysis
Date:	2023-07-24
Issue Date:	2024-09-19 16:48:56 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究論文提出了一種利用基於注意力感知 Bi-LSTM 的 3D 地理信息系統 (GIS) 將氣象數據集成到天氣預報和可視化中的綜合解決方案。該研究的主要貢獻包括氣象站數據處理，涉及數據標記、清理和特徵工程，以及利用十年每小時觀測數據分析關鍵天氣預報特徵。對於天氣預報模型訓練，提出了一種基於注意力的感知的 Bi-LSTM。該算法每小時都能提供準確可靠的天氣預報結果，準確率達到83.03%，優於傳統機器學習算法。並且在3D GIS平台的天氣場景設計中，PSNR和SSIM都具有高質量的精度。這些貢獻為氣象相關領域的用戶、研究人員和政策制定者提供了寶貴的工具和資源，同時也啟發了未來的研究和應用。 ;This research paper presents a comprehensive solution for integrating meteorological data into weather forecasting and visualization using attention-aware Bi-LSTM-based 3D geographic information system (GIS). Key contributions of the research include weather station data processing, involving data labeling, cleaning and feature engineering, and analysis of key weather forecast features using ten-year hourly observation data. For weather forecast model training, an attention-based perception-based Bi-LSTM is proposed. The algorithm can provide accurate and reliable weather forecast results every hour, achieving an accuracy rate of 83.03%, which is superior to traditional machine learning algorithms. And in the weather scene design of the 3D GIS platform, both PSNR and SSIM have high-quality accuracy. These contributions provide valuable tools and resources to users, researchers, and policymakers in meteorological-related fields, while also inspiring future research and applications.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	21	View/Open

社群 sharing

Loading...