Partial discharge, the breakdown of a small portion of a solid or fluid electrical insulation system under high voltage stress, can lead to a destructive phenomenon in electric transmission lines. Without early detection, partial discharges can damage the power grids to complete failure. Therefore, it is important to detect partial discharges so that repairs can be made before any lasting harm occurs, ensure supply reliability and long-term operational sustainability.
What it does
The program will breakdown the signal into their components waves based on the frequency and amplitude of the electric grid. It will have boolean values to check whether a section of a grid will have partial discharge at a specific voltage pressure.
How we built it
Python and its machine learning libraries with Google Cloud storage. We also grabbed some Signal Processing background knowledge from IBM over the past 24 hours, learning to deconstruct a signal, and finally, run our model on Google Cloud DataLab. We also did a website for demonstration with some front-end HTML and CSS programming.
Challenges we ran into
We had problems manipulating our data on the Cloud DataLab, and had a hard time choosing between Principle Component Analysis and K-means Clustering as our machine learning tool.
What's next for GridTecT
We look forward to building a model with high accuracy and easy to reimplement to ensure supply reliability and prevent loss operational expenses across different companies and institutions.