Forecasting distribution grid flows using smart meter data

We are extremely proud of Jules Truong, who finished his Master Thesis under Utiligize’s and Professor Pierre Pinson’s supervision. He applied machine learning to forecast production and consumption on Evonet’s (now merged into N1) network – from high to low voltage. Using graph theory, he was able to determine flows every hour of the day across thousands of cables and transformers. Jules’ work was done in Python, running on Google’s high-performance Bigquery database and high-end servers. It is now a part of our asset management platform. Read his thesis below:

Jules MSc thesis