融合GAN-LSTM的改进云层运动预测  

Improved Cloud Motion Prediction through the Integration of GAN-LSTM

作　　者：吕雄涛 王施琦 袁硕 张雷 焦小雪 LV Xiongtao;WANG Shiqi;YUAN Shuo;ZHANG Lei;JIAO Xiaoxue(Department of Mathematical Science and Engineering,Hebei University of Engineering,Handan,China,056038;Hebei Computational Optical Imaging and Photoelectric Detection Technology Innovation Center,Handan,China,056038;Hebei International Joint Research Center for Computational Optical Imaging and Intelligent Sensing,Handan,China,056038;Hebei Coal Mine Intelligent Engineering Research Center,Handan,China,056038)

机构地区：[1]河北工程大学数理科学与工程学院,河北邯郸056038 [2]河北省计算光学成像与光电检测技术创新中心,河北邯郸056038 [3]河北省计算光学成像与智能感测国际联合研究中心,河北邯郸056038 [4]河北省煤矿智能化工程研究中心,河北邯郸056038

出　　处：《福建电脑》2025年第4期15-20,共6页Journal of Fujian Computer

基　　金：河北省自然科学基金(No.F2024402016,F2024402025,F2016402115);河北省生态矿山联合基金项目(No.F2021402011);河北省重点研发计划项目(No.20371802D)资助。

摘　　要：为提升云层运动预测的质量和精度,本文提出一个预测云层运动的GAN-UNet-ConvLSTM网络模型。在GAN-LSTM网络模型的基础上,引入UNet和ConvLSTM,以提高网络模型的生成能力和预测能力。实验的结果表明,GAN-UNet-ConvLSTM网络模型可以较准确地预测云层运动和形态变化,同时具有良好的可持续预测能力,为云层运动预测提供了一种可行的网络结构优化方案。To improve the quality and accuracy of cloud motion prediction.This article proposes a GAN UNet ConvLSTM network model for predicting cloud motion.On the basis of the GAN-LSTM network model,UNet and ConvLSTM are introduced to improve the generation and prediction capabilities of the network model.The experimental results show that the GAN UNet ConvLSTM network model can accurately predict cloud motion and morphological changes,while also having good sustainable prediction ability,providing a feasible network structure optimization solution for cloud motion prediction.

关 键 词：生成对抗网络-长短期记忆网络 U形网络 卷积长短时记忆网络 云层运动预测 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]