Journal
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS
Volume 58, Issue 6, Pages 4917-4927Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TAES.2022.3188741
Keywords
Predictive models; Indexes; Ionosphere; Magnetic fields; Machine learning; Magnetometers; Urban areas; Feature selection; ionospheric scintillation; machine learning; predictive models
Ask authors/readers for more resources
This article aims to develop short-term predictive models for amplitude ionospheric scintillation through machine learning techniques. A dataset considering various factors was used, and six models were applied for prediction, showing satisfactory predictive capacity.
Ionospheric scintillation is a phenomenon that can compromise and even make the operation of some space-based systems unfeasible. In this context, it is important to develop tools capable of predicting its occurrence. However, modeling this phenomenon is quite complex due to the influence of several other aspects, such as the geomagnetic and solar activities, the seasons, and the geographic location. Therefore, the main objective of this article was to develop short-term predictive models about amplitude ionospheric scintillation through machine learning techniques. The dataset used was built considering information related to geomagnetic, solar, and interplanetary activities, the phenomenon's temporal and geographic dependence, and the ionosphere's state. To predict the value of the scintillation index $S_{4}$ 30 min in advance, six models were used, based on three algorithms, the artificial neural network, the extreme gradient boosting, and the random forest. The results indicated a very satisfactory prediction capacity since a coefficient of determination of 0.87 was achieved by the lower performance model. Additionally, the results demonstrated the usefulness of the considered dataset and the feature selection approach in the model's development phase, which led to better models in accord with some statistical tests performed.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available