-
摘要:
电离层总电子含量(Total Electron Content，TEC)的监测与预报是空间环境研究的重要内容，对卫星通讯和导航定位等有重要意义.TEC值影响因素较多，很难确定精确物理模型来对其进行预测.本文设计了基于注意力机制的LSTM模型(Att-LSTM)，采用过去24小时TEC观测数据对未来TEC进行预测.选择北半球东经100°上，每2.5°纬度选择一个位置，共计36个位置来验证本文提出模型的性能，并与主流的深度学习模型如DNN、RNN、LSTM进行对比实验.取得了如下成果：(1)在选定的36个地区未来2小时单点预测上，基于本文的Att-LSTM模型的TEC预测性能明显优于其他对比模型；(2)讨论了纬度对Att-LSTM预测未来2小时TEC值时性能的影响，发现在北纬0°到60°之间，Att-LSTM预测性能随着纬度的升高而略有降低，在北纬62.5°~87.5°之间，模型预测性能出现扰动，预测效果略差；(3)讨论了磁暴期和磁静期模型的预测性能，发现无论是磁暴期还是磁静期，本文模型预测性能均较好；(4)还讨论了对未来多时点预测效果，实验结果表明，本文所提出的模型对未来2、4个小时的预测拟合度R-Square均超过0.95，预测结果比较可靠，对未来6、8、10个小时预测拟合度最高为0.7934，预测拟合度R-Square下降迅速，预测结果不可靠.
Abstract:The monitoring and prediction of total electron content (TEC) in the ionosphere is an important part of space environment research, which is of great significance for satellite communications, navigation and positioning. TEC values are affected by many factors, and it is difficult to determine an accurate physical model to predict them. In this paper, the LSTM model based on attention mechanism (Att LSTM) is designed to predict the future TEC using the TEC observation data of the past 24 hours. In this paper, we choose one location every 2.5 latitudes at 100° east longitude of the Northern Hemisphere, a total of 36 locations, to verify the performance of our proposed model. And compare with the mainstream deep learning models such as DNN, RNN, LSTM. The following achievements have been made in this paper: (1) On the single point prediction of the next two hours in the selected 36 regions, the TEC prediction performance based on the Att LSTM model in this paper is obviously superior to other comparison models; (2) The paper discusses the influence of latitude on the performance of Att LSTM in predicting the TEC value in the next two hours. It is found that between 0° and 60° north latitude, the prediction performance of Att LSTM decreases slightly with the increase of latitude. Between 62.5° and 87.5° north latitude, the prediction performance of the model is disturbed, and the prediction effect is slightly poor; (3) This paper discusses the prediction performance of the models of magnetic storm period and magnetic quiescent period, and finds that the models in this paper have good prediction performance for the next 2 hours, no matter in the magnetic storm period or the magnetic quiescent period; (4) The paper also discusses the prediction effect of future multiple time points. The experimental results show that the prediction fitting degree R-Square of the proposed model for the next 2 and 4 hours exceeds 0.95, and the prediction results are relatively reliable. The prediction fitting degree for the next 6, 8 and 10 hours is the highest 0.7934. The prediction fitting degree R-Square decreases rapidly, and the prediction results are unreliable.
-
Key words:
- Attention mechanism /
- Long short-term memory /
- Ionosphere /
- Total electron content
-
-
图 8
在0°—87.5°N纬度区域的Att-LSTM模型的TEC预测值与TEC真实值的绝对误差分布图
Figure 8.
Distribution of absolute errors between predicted and true TEC values for the Att-LSTM model in the latitude region 0° to 87.5°N
图 9
Att-LSTM模型对未来2小时TEC预测效果
Figure 9.
Effectiveness of the Att-LSTM model for predicting TEC in the future two hours
图 10
磁静期和磁暴期绝对误差分布直方图
Figure 10.
Histograms of absolute error distribution during magnetostatic and magnetic storm periods
图 11
磁静期(a)和磁暴期(b)Att-LSTM模型的TEC预测性能对比(Att-LSTM代表Att-LSTM模型的TEC预测值，CODE代表原始TEC观测值)
Figure 11.
Comparison of TEC prediction performance of Att-LSTM models during magnetostatic (a) and magnetic storm (b) periods (Att-LSTM represents the TEC prediction value of the Att-LSTM model, and CODE represents the original TEC observations)
图 12
Att-LSTM模型对A11—A19地区未来多个时间段TEC的预测效果
Figure 12.
Effectiveness of the Att-LSTM model in predicting TEC for multiple future time periods in the A11—A19 regions
图 13
Att-LSTM模型对A11—A19地区未来多个时间段TEC真实值与观测值的绝对误差分布直方图
Figure 13.
Absolute error distribution histograms of the Att-LSTM model for the true and observed TEC values for multiple future time periods in the A11—A19 regions
表 1
本文实验选取的所有地区
Table 1.
Description of all the locations in this paper
Lat. 0°N 2.5°N 5°N 7.5°N 10°N 12.5°N 15°N 17.5°N 20°N Lon.=100°E A1 A2 A3 A4 A5 A6 A7 A8 A9 Lat. 22.5°N 25°N 27.5°N 30°N 32.5°N 35°N 37.5°N 40°N 42.5°N Lon.=100°E A10 A11 A12 A13 A14 A15 A16 A17 A18 Lat. 45°N 47.5°N 50°N 52.5°N 55°N 57.5°N 60°N 62.5°N 65°N Lon.=100°E A19 A20 A21 A22 A23 A24 A25 A26 A27 Lat. 67.5°N 70°N 72.5°N 75°N 77.5°N 80°N 82.5°N 85°N 87.5°N Lon.=100°E A28 A29 A30 A31 A32 A33 A34 A35 A36 
下载: 导出CSV
表 2
Att-LSTM参数设置
Table 2.
Parameters of Att-LSTM
实验参数 设置值 input_dim 13 optimizer Adagrad loss “RMSE”和“R-Square” activation “Relu” learning rate 0.001 batch_size 64 num of hidden units 64 output_dim 1
下载: 导出CSV
表 3
四种模型在表 1所有地区TEC预测性能对比
Table 3.
Comparison of TEC prediction performance of the four models for all regions in Table 1
模型 评价指标 均值 最小值 最大值 DNN RMSE(TECU) 6.0177 3.1699 10.5349 RNN 4.5991 2.4711 7.5964 LSTM 2.0225 0.8610 4.8151 Att-LSTM 1.4007 0.0495 4.7205 DNN R-Square 0.6492 0.3747 0.8994 RNN 0.8097 0.6496 0.8962 LSTM 0.9702 0.9483 0.9838 Att-LSTM 0.9869 0.9493 0.9999
下载: 导出CSV
表 4
Att-LSTM模型对A11—A19地区未来多个时间段TEC的预测性能(加黑部分为最优结果)
Table 4.
Prediction performance of Att-LSTM model for TEC in the A11—A19 regions over multiple future time periods
预测未来时段 评价指标 A11 A12 A13 A14 A15 A16 A17 A18 A19 4时 RMSE(TECU) 0.9178 1.0276 1.2818 1.5852 1.1040 1.1913 1.2264 0.9083 1.3231 6时 3.3574 3.4730 3.8501 3.8688 3.9683 3.9564 3.8970 3.9012 4.0875 8时 3.8854 4.1572 4.4190 4.5560 4.6336 4.7068 4.7201 4.7746 4.8816 10时 4.1589 4.4089 4.5281 4.5679 4.6940 4.7451 4.7292 4.6996 4.9585 4时 R-Square 0.9818 0.9784 0.9684 0.9545 0.9787 0.9754 0.9733 0.9850 0.9699 6时 0.7920 0.7934 0.7667 0.7763 0.7749 0.7799 0.7839 0.7820 0.7743 8时 0.7516 0.7367 0.7232 0.7228 0.7246 0.7205 0.7150 0.7065 0.7051 10时 0.7198 0.7082 0.7071 0.7120 0.7142 0.7105 0.7059 0.7038 0.6912
下载: 导出CSV
-
Akhoondzadeh M. 2013. A MLP neural network as an investigator of TEC time series to detect seismo-ionospheric anomalies. Advances in Space Research, 51(11): 2048-2057, doi: 10.1016/j.asr.2013.01.012.
Cander L R, Milosavljevic M M, Stankovic S S, et al. 1998. Ionospheric forecasting technique by artificial neural network. Electronics Letters, 34(16): 1573-1574. doi: 10.1049/el:19981113
Galassi A, Lippi M, Torroni P. 2021. Attention in natural language processing. IEEE Transactions on Neural Networks and Learning Systems, 32(10): 4291-4308, doi: 10.1109/TNNLS.2020.3019893.
Graves A. 2012. Long short-term memory. //Supervised Sequence Labelling with Recurrent Neural Networks. Berlin, Heidelberg: Springer, 37-45, doi:
10.1007/978-3-642-24797-2_4 .Habarulema J B, McKinnell L A, Cilliers P J. 2007. Prediction of global positioning system total electron content using neural networks over South Africa. Journal of Atmospheric and Solar-Terrestrial Physics, 69(15): 1842-1850, doi: 10.1016/j.jastp.2007.09.002.
Habarulema J B, McKinnell L A, Cilliers P J, et al. 2009. Application of neural networks to South African GPS TEC modelling. Advances in Space Research, 43(11): 1711-1720. doi: 10.1016/j.asr.2008.08.020
Hochreiter S, Schmidhuber J. 1997. Long short-term memory. Neural Computation, 9(8): 1735-1780, doi: 10.1162/neco.1997.9.8.1735.
Hu J, Ruan H B, Huang F Q, et al. 2024. Assessment of International GNSS Service Global Ionosphere Map products over China region based on measurements from the Crustal Movement Observation Network of China. Earth Planet. Phys., 8(2): 1-8, doi: 10.26464/epp2024004.
Huang Z, Yuan H. 2014. Ionospheric single-station TEC short-term forecast using RBF neural network. Radio Science, 49(4): 283-292, doi: 10.1002/2013RS005247.
Inyurt S, Sekertekin A. 2019. Modeling and predicting seasonal ionospheric variations in Turkey using artificial neural network (ANN). Astrophysics and Space Science, 364(4): 62, doi: 10.1007/s10509-019-3545-9.
Jiang H, Liu J B, Wang Z M, et al. 2019. Assessment of spatial and temporal TEC variations derived from ionospheric models over the Polar Regions. Journal of Geodesy, 93(4): 455-471, doi: 10.1007/s00190-018-1175-6.
Karpov I V, Karpov M I, Borchevkina O P, et al. 2019. Spatial and temporal variations of the ionosphere during meteorological disturbances in December 2010. Russian Journal of Physical Chemistry B, 13(4): 714-719, doi: 10.1134/S1990793119040067.
Kaselimi M, Voulodimos A, Doulamis N, et al. 2020. A causal long short-term memory sequence to sequence model for TEC prediction using GNSS observations. Remote Sensing, 12(9): 1354, doi: 10.3390/rs12091354.
Kong Q Y, Feng J, Liu W, et al. 2010. Development on short-term prediction of iononspheric parameters. Progress in Geophysics (in Chinese), 25(6): 1968-1976, doi: 10.3969/j.issn.1004-2903.2010.06.011.
Le X A, Wan W X, Liu L B, et al. 2010. Development of an ionospheric numerical assimilation nowcast and forecast system based on Gauss-Markov Kalman filter—An observation system simulation experiment taking example for China and its surrounding area. Chinese Journal of Geophysics (in Chinese), 53(4): 787-795, doi: 10.3969/j.issn.0001-5733.2010.04.003.
Li X L, Yuan A H, Lu X Q. 2021. Vision-to-language tasks based on attributes and attention mechanism. IEEE Transactions on Cybernetics, 51(2): 913-926, doi: 10.1109/TCYB.2019.2914351.
Li Z, Yang B Y, Huang J P, et al. 2022. Analysis of pre-earthquake space electric field disturbance observed by CSES. Atmosphere, 13(6): 934, doi: 10.3390/atmos13060934.
Liu F, Zhou X S, Cao J L, et al. 2022. Anomaly detection in quasi-periodic time series based on automatic data segmentation and attentional LSTM-CNN. IEEE Transactions on Knowledge and Data Engineering, 34(6): 2626-2640, doi: 10.1109/TKDE.2020.3014806.
Liu X D, Song L J, Yang X H, et al. 2010. Predicting shortdated ionospheric TEC based on wavelet neural network. Hydrographic Surveying and Charting (in Chinese), 30(5): 49-51, 55.
Qiao J D, Liu Y, Fan Z Q, et al. 2021. Ionospheric TEC data assimilation based on Gauss-Markov Kalman filter. Advances in Space Research, 68(10): 4189-4204, doi: 10.1016/j.asr.2021.08.004.
Ratnam D V, Dinesh B V, Tejaswi B, et al. 2012. TEC prediction model using neural networks over a low latitude GPS station. International Journal of Soft Computing and Engineering, 2(2): 517-521.
Ren Q B, Li M C, Li H, et al. 2021. A novel deep learning prediction model for concrete dam displacements using interpretable mixed attention mechanism. Advanced Engineering Informatics, 50: 101407, doi: 10.1016/j.aei.2021.101407.
Ruwali A, Kumar A J S, Prakash K B, et al. 2021. Implementation of hybrid deep learning model (LSTM-CNN) for ionospheric TEC forecasting using GPS data. IEEE Geoscience and Remote Sensing Letters, 18(6): 1004-1008, doi: 10.1109/LGRS.2020.2992633.
Sharma G, Champati Ray P K, Mohanty S, et al. 2017. Ionospheric TEC modelling for earthquakes precursors from GNSS data. Quaternary International, 462: 65-74, doi: 10.1016/j.quaint.2017.05.007.
Sivavaraprasad G, Deepika V S, SreenivasaRao D, et al. 2020. Performance evaluation of neural network TEC forecasting models over equatorial low-latitude Indian GNSS station. Geodesy and Geodynamics, 11(3): 192-201, doi: 10.1016/j.geog.2019.11.002.
Sun W Q, Xu L, Huang X, et al. 2017. Forecasting of ionospheric vertical total electron content (TEC) using LSTM networks. //2017 International Conference on Machine Learning and Cybernetics (ICMLC). Ningbo, China: IEEE, 340-344.
Tulunay E, Senalp E T, Radicella S M, et al. 2006. Forecasting total electron content maps by neural network technique. Radio Science, 41(04): 1-12.
Tan S S, Zhou B, Guo S T, et al. 2011. Research on COMPASS navigation signals of China. Chinese Space Science and Technology (in Chinese), 31(4): 9-14, 29.
Tang J, Li Y J, Yang D P, et al. 2022. An approach for predicting global ionospheric TEC using machine learning. Remote Sensing, 14(7): 1585, doi: doi.org/10.3390/rs14071585.
Tang R X, Zeng F T, Chen Z, et al. 2020. The comparison of predicting storm-time ionospheric TEC by three methods: ARIMA, LSTM, and Seq2Seq. Atmosphere, 11(4): 316, doi: 10.3390/atmos11040316.
Unnikrishnan K, Haridas S, Choudhary R K, et al. 2018. Neural Network Model for the prediction of TEC variabilities over Indian equatorial sector. Indian J. Sci. Res., 18(1): 56-58.
Wang F K, Mamo T, Cheng X B. 2020. Bi-directional long short-term memory recurrent neural network with attention for stack voltage degradation from proton exchange membrane fuel cells. Journal of Power Sources, 461: 228170, doi: 10.1016/j.jpowsour.2020.228170.
Watthanasangmechai K, Supnithi P, Lerkvaranyu S, et al. 2012. TEC prediction with neural network for equatorial latitude station in Thailand. Earth, Planets and Space, 64(6): 473-483, doi: 10.5047/eps.2011.05.025.
Wen Y Z, Tao D, Wang G X, et al. 2022. Ionospheric TEC and plasma anomalies possibly associated with the 14 July 2019 MW7.2 Indonesia Laiwui earthquake, from analysis of GPS and CSES data. Earth Planet. Phys., 6(4): 313-328, doi: 10.26464/epp2022028.
Weng L B, Fang H X, Miao Z Q, et al. 2012. Forecasting of ionospheric TEC one hour in advance by artificial neural network. Chin. J. Space Sci. (in Chinese), 32(2): 204-208.
Xiong B, Li X L, Wang Y Q, et al. 2022. Prediction of ionospheric TEC over China based on long and short-term memory neural network. Chinese Journal of Geophysics (in Chinese), 65(7): 2365-2377, doi: 10.6038/cjg2022P0557.
Yuan T J, Chen Y H, Liu S Q, et al. 2018. Prediction model for ionospheric total electron content based on deep learning recurrent neural networkormalsize. Chin. J. Space Sci., 38(1): 48-57, doi: 10.11728/cjss2018.01.048.
Yue X A, Wan W X, Liu L B, et al. 2008. Development of a middle and low latitude theoretical ionospheric model and an observation system data assimilation experiment. Chinese Science Bulletin, 53(1): 94-101, doi: 10.1007/s11434-007-0462-z.
孔庆颜, 冯静, 柳文等. 2010. 电离层参数短期预报研究进展. 地球物理学进展, 25(6): 1968-1976, doi: 10.3969/j.issn.1004-2903.2010.06.011.
乐新安, 万卫星, 刘立波等. 2010. 基于Gauss-Markov卡尔曼滤波的电离层数值同化现报预报系统的构建——以中国及周边地区为例的观测系统模拟试验. 地球物理学报, 53(4): 787-795, doi: 10.3969/j.issn.0001-5733.2010.04.003.
刘先冬, 宋力杰, 杨晓晖等. 2010. 基于小波神经网络的电离层TEC短期预报. 海洋测绘, 30(5): 49-51, 55.
谭述森, 周兵, 郭盛桃等. 2011. 中国全球卫星导航信号基本框架设计. 中国空间科学技术, 31(4): 9-14, 29.
翁利斌, 方涵先, 缪子青等. 2012. 利用人工神经网络提前1 h预报电离层TEC. 空间科学学报, 32(2): 204-208.
熊波, 李肖霖, 王宇晴等. 2022. 基于长短时记忆神经网络的中国地区电离层TEC预测. 地球物理学报, 65(7): 2365-2377. http://www.geophy.cn/article/doi/10.6038/cjg2022P0557
袁天娇, 陈艳红, 刘四清等. 2018. 基于深度学习递归神经网络的电离层总电子含量经验预报模型. 空间科学学报, 38(1): 48-57, doi: 10.11728/cjss2018.01.048.
-
图(13)
表(4)
计量
- 文章访问数:
- PDF下载数:
- 施引文献: 0