鸟类时间预测的原理与方法及验证
Journal: Agricultural Science DOI: 10.12238/as.v8i3.2860
Abstract
本研究以鸟类迁徙时间预测为核心目标,系统分析了时间序列数据的统计特征与非线性关系,提出了一种融合LSTM、Transformer和Prophet模型的混合预测框架。通过构建包含环境变量(温度、风速)、地理信息(经纬度)和生物特征(物种、年龄)的多模态数据集,验证了深度学习模型在捕捉长周期依赖和全局关联方面的优势。实验结果表明,Transformer模型在迁徙时间预测中表现最佳(RMSE=5.8天,R²=0.85),其自注意力机制有效捕捉了跨区域迁徙的协同效应。研究结果为鸟类生态保护与气候变化适应性分析提供了理论依据与技术支持。
Keywords
鸟类迁徙；混合预测框架；多模态数据集；深度学习模型
Full Text
PDF - Viewed/Downloaded: 0 TimesReferences
[1] Hyndman, R.J., & Athanasopoulos, G. (2018). Forecasting: Principles and Practice.O'Reilly Media.
[2] Goodfellow, I.,et al.(2016).Deep Learning.MIT Press.
[3] McKinney, A.M.,et al. (2020). "Ecological applications of machine learning in avian research." Ecological Applications, 30(8),e02127.
[4] Zhang,Q., et al. (2022). "Transformer-based time series forecasting for animal migration." Nature Communications, 13(1),1234.
[5] GBIF(Global Biodiversity Information Facility).https:// gbif.org.
[6] Taylor, Sean J.,Benjamin Letham (2018). Forecasting at scale.The American Statistician72(1):37-45.
[2] Goodfellow, I.,et al.(2016).Deep Learning.MIT Press.
[3] McKinney, A.M.,et al. (2020). "Ecological applications of machine learning in avian research." Ecological Applications, 30(8),e02127.
[4] Zhang,Q., et al. (2022). "Transformer-based time series forecasting for animal migration." Nature Communications, 13(1),1234.
[5] GBIF(Global Biodiversity Information Facility).https:// gbif.org.
[6] Taylor, Sean J.,Benjamin Letham (2018). Forecasting at scale.The American Statistician72(1):37-45.
Copyright © 2025 辛钳康, 王宇航, 施泽荣
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
