Prediction of the Next Solar Cycle and Solar Maximum Based on Machine Learning

Xuanyi Xiang; Xinyue Zhang; Qingyi Huang; Jingyi Liu

doi:10.62051/wkvjsd04

Authors

Xuanyi Xiang
Xinyue Zhang
Qingyi Huang
Jingyi Liu

DOI:

https://doi.org/10.62051/wkvjsd04

Keywords:

Solar Cycle Time, KNN Regression Algorithm, XGBOOST Regression Model.

Abstract

The intensity of solar activity varies with the solar cycle of about 11 years, reaching a peak during the solar maximum. Solar activity can affect space weather in ionospheric states and conditions related to shortwave radio propagation or satellite communications, which brings challenges to space exploration, communications, and weather forecasting. Therefore, accurate prediction of the start time and duration of the solar maximum becomes pivotal. In order to solve this problem, this research first applies the K Nearest Neighbors classification algorithm to predict that the beginning of the next solar cycle will be around June 2030 until about July 2040. Then, by establishing the Extreme Gradient Boosting model, this study forecasts that the solar maximum in the next solar cycle will start in August 2034 and last for 8 months. With this accurate prediction of the next solar maximum, aerospace, communications, meteorology, and other industries can reduce the interference of solar activities.

Downloads

Download data is not yet available.

References

[1] SUN Wei, YU XiaoXiao, LIU FuYao, ET al.2024. Numerical simulation and prediction of the sunspot magnetic field and polarity index [J]. Progress in Geophysics, 2024, 39(3): 885-895.

[2] ZHOU Meilin, ZHONG Libo. Sunspot Data Collection and Experimental Validation for McIntosh Classification [J]. Astronomical Research and Technology, 2023, 20 (04): 353-368.

[3] CHENG Shu, SHI Yaolin, ZHANG Huai. Prediction of sunspot changes based on neural network [J]. Journal of University of Chinese Academy of Sciences, 2022, 39 (05): 615-626.

[4] Lockwood M, Owens M J, Barnard L A, Application of historic datasets to understanding open solar flux and the 20th-century grand solar maximum. 1. Geomagnetic, ionospheric, and sunspot observations [J]. Frontiers in Astronomy and Space Sciences, 2022, 9:960775.

[5] Amrita P, Soumya R, Arindam S, et al. Prediction of solar cycle 25 using deep learning based long short-term memory forecasting technique [J]. Advances in Space Research, 2022, 69(01):798-813.

[6] Uddin, S, Haque, I, Lu, H. et al. Comparative performance analysis of K-nearest neighbour (KNN) algorithm and its different variants for disease prediction[J]. Scientific Reports, 2022, 12:6256.

[7] Zhang Shichao, Challenges in KNN Classification [J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 34(10): 4663-4675.

[8] MU Meihong, ZHAO Gang, HE Bingshun, Qing Li, et al. XGBoost-based method for flash flood risk assessment [J]. Journal of Hydrology, 2021, 598: 126382.

[9] Mu Chengyu. Research on credit risk assessment model of small, medium and micro enterprises based on Xgboost regression [J]. Business News,2022(15):119-122

[10] Petrovay, K. Solar cycle prediction [J]. Living Reviews in Solar Physics, 2020, 17(2):1-93.

[11] Davide C, J M W, Giuseppe J. The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation. [J]. Peer J. Computer science, 2021, 7 e623-e623.