-
-
-
-
Section of our websites where users can input data to predict energy production
-
Example of our model generating accurate energy results based on user data
-
-
-
This is the structure of our neural network.
Inspiration
We noticed that weather conditions highly affect renewable energy production, and often, bad planning leads to money wasted on renewable energy. Thus, we wanted to utilize AI so that energy departments can more informed budgeting decisions regarding energy production depending on the weather.
What it does
Our program inputs the weather forecast data and uses it to predict the amount of renewable energy generated in megawatts, as well as the price to generate this energy in pounds per megawatt, for that hour. This allows people to determine how the weather will affect energy production and costs based on current weather conditions.
How we built it
We built the Neural Network using the Python TensorFlow library. We converted our neural network into a JSON file using Tensorflow.js. We built our web application using HTML, CSS, and JavaScript.
Challenges we ran into
One challenge we ran into was that the dataset we were using had a few errors, and while we were building our Neural Network, we ran into issues because of missing data values. We fixed this challenge by spending time cleaning up the data in order to make sure it could be used to train the neural network. Additionally, we also ran into some challenges as we tried connecting the Neural Network to the web application, but overcame this by utilizing the Tensorflow.js library to convert our neural network into a JSON file so that it could be implemented into the front-end portion of our project.
Accomplishments that we're proud of
We are proud of being able to clean up the data as necessary and set up all of it in order to plug into the Neural Network. We are also proud of giving the website a unique aesthetic with our parallax scrolling effects and coding it effectively. Last but not least, we are proud of our high accuracy rates for our model.
What we learned
We learned some techniques to clean up data, utilize divs and CSS grids effectively, and to remember to take a step back and look at the big picture when something wasn't working.
What's next for Energy Forecast
Next, we plan on improving our dataset and collecting data from more locations around the world, and add a geological factor to our algorithm.
Log in or sign up for Devpost to join the conversation.