Sunspot Activity and Cycle Prediction Study Based on Time Series Algorithm and CNN-LSTM Modeling

This study investigates the relationship between sunspot activity and solar cycle variations, utilizing time series algorithms and the CNN-LSTM model for prediction. Initial preprocessing of sunspot number and area data involved outlier elimination and linear interpolation. The ARIMA time series algorithm was chosen for its ability to integrate convolutional neural networks and long short-term memory networks, enhancing prediction accuracy. The model forecasts the current solar cycle from December 2019 to February 2030 and the next from February 2030 to June 2040. Additionally, the study predicts the onset and duration of the next solar maximum and calculates sunspot numbers and areas for the current and upcoming cycles. The research effectively combines temporal and spatial data analysis, offering a comprehensive approach to predicting solar cycle dynamics.