| 摘要: |
| 针对电离层闪烁预测困难和预测模型精度低的问题，提出基于XGBoost机器学习模型（XGB）并结合混沌映射改进蜣螂优化算法（Tent Dung Beetle Optimizer,TDBO）的TDBO-XGB电离层闪烁预测模型，利用2020年1月1日至2024年3月21日香港HKOH站日落前5 h与电离层闪烁相关的背景电离层参数，对日落后3 h电离层是否发生闪烁事件进行预测建模，并利用模型对6组不同组合的背景电离层参数进行预测分析。结果表明，TDBO-XGB模型预测电离层闪烁发生的准确率达93.72%，比单一默认参数的XGB模型提高1.94%；在利用不同组合参数作为输入数据的闪烁预测中，表征太阳活动的参数对电离层闪烁预测模型的预测结果起提升作用，且使用站点东侧南北两半球沿经度线的TEC变化数据对电离层闪烁进行预测的效果提升显著。 |
| 关键词: 电离层闪烁预测 机器学习 优化算法 混沌映射 |
| DOI：10.20079/j.issn.1001-893x.240703002 |
|
| 基金项目: |
|
| Ionospheric Scintillation Prediction Based on TDBO-XGB Model |
| LI Feia,b,b,JI Yuanfaa,b |
| (1a.Guangxi Key Laboratory of Precision Navigation Technology and Application;1b.School of Information and Communicaiton;1c.International Joint Research Laboratory of Spatio-temporal Information and Intelligent Location Services,Guilin University of Electronic Technology,Guilin 541004,China;2.Guangxi Industrial Research Institute of Time and Space Information Technology Research Institute Co.,Nanning 530031，China;3.College of Electronic Information and Automation，Guilin University of Aerospace Technology,Guilin 541004,China) |
| Abstract: |
| Responding to the difficulty of predicting ionospheric scintillation and the low accuracy of prediction models,a TDBO-XGB ionospheric scintillation prediction model based on XGBoost machine learning model(XGB) combined with the Tent Dung Beetle Optimizer(TDBO) of the chaos mapping is proposed.The background ionospheric parameters associated with ionospheric scintillation five hours before sunset at HKOH station in Hong Kong from January 1,2020 to March 21,2024 are used to model the prediction of whether an ionospheric scintillation event occurs three hours after sunset,and the model is used to predict and analyze six different combinations of background ionospheric parameters.The results show that the accuracy of the TDBO-XGB model in predicting the occurrence of ionospheric scintillation reaches 93.72%,which is 1.94% higher than that of the XGB model with a single default parameter;in the prediction of scintillation using different combinations of parameters as the input data,the parameter that characterizes the solar activity enhances the prediction results of the ionospheric scintillation prediction model,and the effect of using the TEC data along the longitude line in the north and south hemispheres east of the site is significant for the prediction of ionospheric scintillation. |
| Key words: ionospheric scintillation prediction of ionospheric scintillation machine learning optimization algorithm chaos mapping |
