This project helps optimize household electricity consumption to reduce costs and carbon emissions. This fits the social good and sustainability theme.
Inspiration: Electricty consumption has risen dramatically with emerging technologies having higher power requirements than ever and firms investing heavily into tech intensive infrastructure like data centers. Hence to reduce the Carbon footprint of such large levels of consumptions as well as help larger groups of households offset cost, we made WattSense
What it does: The software predicts as well as optimizes future electricity consumptions according to the market's peak-off peak demand periods. It also spreads out electricity consumption over 24 hrs to reduce larger CO2 emissions. It uses a CSV dataset as a source to perform all features. Also indicates how much CO2 emissions were reduced and the financial gains of households over a day, month or year.
How we built it: Using python to represent electricity consumption of 50000 US households we trained an LSTM model to predict and optimise future electricity consumption in order to offset costs as well as reduce CO2 emissions.
Challenges we ran into: It was hard to find a dataset with equal time intervals and enough data to train the LSTM model. We also struggled in training the model and had to try different ways of fitting it to the dataset. There was also an issue where it wouldn't predict more data points and output the same value for the entire graph. Connecting the front end to backend using an api fetching the model from the backend.
Log in or sign up for Devpost to join the conversation.