Pluto

Awards & Nominations

Pluto has received the following awards and nominations. Way to go!

Global Nominee

Hyper Space Predictor | No More Suprises

High-Level Project Summary

We created an AI Model which takes in the Solar Wind Data from NASA DSCOVR Satellite to predict the possibility of another Carrington Event

Link to Final Project

https://github.com/abdulbasit108/PlutoSpaceApp.git

Link to Project "Demo"

https://www.youtube.com/watch?v=3468CI5ye_w

Detailed Project Description

Our project is based on an Unsupervised Anomaly Detection Algorithm which flags any anomalous data. It works by fitting a Gaussian Distribution Curve on the Solar Wind training data and then estimating the mean and variance of the dataset. Then, when we enter any new data input, the program calculates its probability using the trained mean and variance. This probability is then compared to a preset threshold called epsilon. If its less then that, it is classified as an Anomaly.

Our Dataset included 6 features for every reading of Solar Wind including Thermal Speed, Density, Temperature, Velocity X-Component, Velocity Y-Component and Velocity Z-Component

The benefit this algorithm provides us is that it will flag not only a Carrington-like event but also all other kinds of anomalies that might occur. The system provides us with an early warning which can then be further investigated by NASA to make sure everything is smooth.

We used Python's NumPy Library to create a 2D array of our dataset and then created functions to fit a Gaussian Curve to our data. All work was done in Jupyter Notebooks and VSCODE

Limitations:

Due to lack of target labels on even some subset of data, the epsilon value is just an estimate and can be further improved.
Interface could not be created due to lack of time
Model can be improved to produce more precise results by adding data from different satellites and for a longer duration of time.

Space Agency Data

We used DSCOVR Plasma and Solar Winds Data provided by NASA at SPDF - Coordinated Data Analysis Web (CDAWeb) (nasa.gov)

Hackathon Journey

Our experience at Space Apps 2022 was remarkable and we got to learn so much. The main highlight was getting to experience how a real-world problem is solved. We were always fascinated by space and had a keen interest in AI & Machine Learning and what it can do. So as soon as NASA arrived with the Space Apps Challenge we were onboard from Day 1. Our first step was to theoretically think of all the different Machine Learning Algorithms that could help us solve this problem and then select one as soon as we got access to the Space Data. The main setback was that we could not access the data from the links provided and had to search for the specific required data elsewhere.

Pluto

Karachi | Save the Earth from Another Carrington Event!