Summary
Solterra uses machine learning in order to accurately predict solar and wind harvests for each user. These personalized forecasts allow for optimal management of a user’s renewable resources.
We take in weather and location and use our deep learning model to forecast Irradiance, temperature, and wind every hour of the day for the next seven days. This data is then used by hardware models of solar and wind systems for a power generation forecast. We have completed the Deep Learning Model and, on paper, know what the solar and wind hardware models should look like.
We were able to achieve good performance using just a single time step for a 7-day forecast as you can see in our training results and mean absolute error.
Model Performance:
- Global Horizontal Irradiance: ± 18.13 W/m^2
- Wind Speed: ± 0.22 m/s
- Temperature: ± 0.56 °C
Through the use of Solterra, power forecasts created from wind and solar trends allow users to maximize their grid efficiency and save on costs. Solterra allows communities to primarily depend on renewable energy instead of fossil fuels, creating a multi positive impact in several scopes.
Households:
- Find the best time to sell back power
- Optimize storage based on consumption patterns
Grid Operators:
- Finer resolution power planning for grid operations
- Make dependable transition to solar and wind energy
Developing Markets
- Enable Community Agriculture
- Validate international climatology and remote sensing research
Built With
- python
- tensorflow
Log in or sign up for Devpost to join the conversation.