Awards & Nominations
SeekSolutions has received the following awards and nominations. Way to go!
Global Nominee
A convolutional long-short term memory based neural network pipeline to predict solar wind behaviour
High-Level Project Summary
The project aims to deliver the framework for a convolutional long-short term memory neural network in order to accurately and efficiently detect solar wind velocities, temperatures and densities in order to streamline the current manifold pipeline on magnetic data from spacecrafts Wind and DISCOVR
Link to Final Project
Link to Project "Demo"
Detailed Project Description
Seek solutions developed a machine learning algorithm which takes in magnetic data from the DSCOVR and Wind spacecrafts in order to predict solar wind velocity, temperature and density. This predictive model aims to replace the current manifold pipeline which is time-consuming and inefficient. The input magnetic data is initially pre-processed and then undergone a dynamic time warping process in order to temporally align magnetic data to the same particles in space time. Thereafter the data is passed to a trained convolutional long short term memory neural network in order to accurately predict the solar wind characteristics. The model was developed in python using tensorflow.
Space Agency Data
Hackathon Journey
The world’s largest hackathon certainly challenged our own perception of teamwork, collaboration and unconventional problem solving. Although the mental stimulation was exhilarating, the friendships made along the way left the deepest mark on this experience. The community really adopted the “There is always room for one more” narrative!
The contributions made by teammates, other teams, subject matter experts and the Discord channel was unparalleled. It was so reassuring to know others were facing the same challenges as you and by working together solved it for not just us, but for the others doing the same challenge.
References
Owens, Mathew J., and Jonathan D. Nichols. "Using in situ solar-wind observations to generate inner-boundary conditions to outer-heliosphere simulations–I. Dynamic time warping applied to synthetic observations." Monthly Notices of the Royal Astronomical Society 508.2 (2021): 2575-2582.
The Wind Mission’s Magnetic Field Data Sets, BW(t)
The DSCOVR Magnetic Field Data Sets, BD(t)
The Wind Mission’s Ion Parameters
DSCOVR Instrumentation Capabilities and Calibration Test Plan
Tags
#neuralnetwork #dscovr #solarwind #carringtonevent