Presents:
Chosen Challenge:
Our Solution
(Demo)
Carrington event of 1859
Considered the most powerful solar storm recorded in history.
Can we prevent a similar event?
Obtain magnetic field information from the satellite windy dscovr, which will serve as training data for the neural network
1
SOLUTION
2
To do this, a neural network will be implemented that, given certain input values, manages to calculate the values that were lost.
3
To solve the problem, data for 3 months was collected from both the DSCOVR satellite and the WIND satellite. Considering the following variables:
Magnetic field magnitude
Proton density
Thermal Speed
Solar wind speed
A synchronization is performed on the values obtained so that it is possible to work them together
4
Data normalization continues, which serves to train and validate the neural network.
5
Once the data is divided, the neural network is built. This works by taking data on the magnetic field of both satellites, and the WINDS’ ion parameters: velocity, the proton density and the thermal velocity. .
6
By creating the respective learning layers, the missing parameters of DSCVR FC can be obtained.
7
Managing to complete the missing data of the DSCOVR satellite as well as being possible to make future predictions
Demo and User Experience:
Finally, all the relevant information is presented in an interactive format to facilitate the visualization of results.
Carrington's algorithm will contribute to more accurate data analysis and records for scientists and stakeholders to provide timely warnings.
Look into the future:
We live in an era where it is possible to obtain large amounts of data of all kinds.
Since its inception, neural networks have had great potential to process data, since it is very difficult to find direct correlations between different data sets.
With that idea in mind, it was decided to use neural networks to handle the large amount of data from satellites destined for sun analysis experiments.