HoloNet: Enhancing Predictive Modeling of Space Phenomena Using Holomorphic Neural Networks in CubeSats

Abstract

The provision of precise space weather forecasts necessities the development of enhanced methodologies for simulating space events. This paper then presents the HoloNet, a holomorphic neural network for improving the predictive modeling of space events based on CubeSat-acquired data. Herein, HoloNet leverages key aspects of CubeSats to address the challenges of managing multi-source data, such as geomagnetic indices, satellite position, solar wind parameters, and sunspot activity. The uniqueness of the approach remains in applying holomorphic neural network for this purpose, which provide a higher dimensionality of features under consideration and thus have the potential to enhance the predictive capability of the model regarding solar-terrestrial interactions. The present’s analysis involved merging data from four different sources, undergoing data preprocessing, feature extraction, and deep learning. The obtained result demonstrates a high degree of generalization, a reduction in prediction error, and a crucial features analysis that identifies the primary consumers of changes in solar activity. This work will demonstrate how HoloNet a significantly after the use of CubeSat technologies for space weather monitoring by enhancing data exploitation and predictive capabilities. In conclusion, this work lays a solid foundation for the use of holomorphic neural networks in CubeSats for precise space Phenomena.