High-Level Project Summary
Our solution is deep learning model that will be able to predict the solar storms.• Deep learning model (Time series model) that uses previous available data that will link some of the phenomena that occur on the sun to the release of large amounts of plasma to the earth.• We present it through a website to the public to be able to know the impact of the solar wind on communications and rationalize electricity consumption when it increases to reduce damage. • This model will be provided with continuous data on what is happening on the sun, so it analyses that data directly and concludes when a harmful solar storm will occur on Earth.
Link to Final Project
Link to Project "Demo"
Detailed Project Description
Our project is a deep learning model which analysis data and predict when the solar storms could happen as so show it on website to be available for public and to be able to know the impact of the
solar wind on communications and rationalize electricity consumption when it
increases to reduce damage.
We did that with many steps as below:
1- Data collection : By searching for all available data.
2- Data analysis and Eda : Analyzing specific data which causes solar storms.
3- Data preparation : Simple and clean data to use it in model.
4- Feature engineering : Select important columns to make verdict model.
5- Choosing the model : There was many models but we chose LSTM.
6- Time series forecast with LSTM model.
Space Agency Data
We have a great use of NASA's satellites (https://omniweb.gsfc.nasa.gov/form/dx1.html) and (https://www.swpc.noaa.gov/products/ace-real-time-solar-wind) as they provided us with the necessary data to be used in the LSTM model.
Hackathon Journey
As first-time at the Hackathon, we started it full of doubts, but above all curious to see how far our capabilities can take us. We worked in parallel each one of us knew what to do and each have his own task.
We faced a lot of obstacles , we tried to solve all of them as a team, the hardest thing was starting working because we needed to understand many things about that phenomena to start solving our challenge.
References
- https://omniweb.gsfc.nasa.gov/form/dx1.html
- https://cdaweb.gsfc.nasa.gov/pub/data/wind/mfi/mfi_h2/2022/
- https://www.swpc.noaa.gov/products/ace-real-time-solar-wind
Resources:
- DR Dalia Elfiky, Head of Structural, Thermal Control and Space Environment Dept.
- DR Eman Raslan, External instructor at ITI, TA at AUC.
Tools:
- Excel
- Jupyter
- Pycharm
- Google colab
- VS code
Tags
#SpaceApps #deeplearning #LSTM_model #data_science #Satellites #AI #ML