Robust "All-Clear" Forecast of Solar Proton Events with Machine Learning Using McIntosh and Magnetic Classification of Active Regions

The Increased radiation of Solar Energetic Particles (SEPs) may impact the health of astronauts during space operations and may hinder future space explorations. Solar Proton Events (SPEs) represent a major subclass of SEPs. In this work, the importance of developing "all-clear" prediction of SPEs is considered. The project can be divided into three phases: 1)Creation of database. Data containing SEP events based on active regions of the Sun is retrieved using an Application Programming Interface (API) and a data retrieval tool. Data is then converted into a data frame and missing inputs are extrapolated to produce a whole-Sun input. 2)Application of Artificial Intelligence to produce prediction. Database is divided into a set of training data and testing data. Machine Learning is applied to an Artificial Neural Network designed for whole-Sun input, which is then trained and tested using the respective data sets. 3)Assessing performance. This effort represents an extension of previous "all-clear" forecasting analysis (Sadykov et al. 2021) using data from 2010 to 2020. To produce a more robust "all-clear" forecast, we increase the input data with a new range spanning 1996 to 2020. We will discuss the performance of the developed AI approach against current operational forecasts.