Optiturbino

Inspiration

We took a look at Kaggle datasets and manufacturer websites like NordEx to understand the relevant key features.

What it does

Predict whether the model is in need of maintenance and if so when intervention is needed

How we built it

Classification model to predict if the turbine needs maintenance or not, where the output will be fed into the LSTM model along with turbine specific data to determine the remaining time estimated until a problem can occur

Challenges we ran into

Lack of data that provides insights about time evolution of turbine performance according to which turbine type and manufacturer it is

Accomplishments that we're proud of

Finding correlations linking the manufacturers to the different turbine specs. Created synthetic data to compensate for the lack of timestamps per turbine type.

What we learned

Data in the real world is quite messy and we have to be creative how to draw insights from it to make rational decisions

What's next for Optiturbino

1. Finetune our model with more features
2. Provide insights about optimal turbine installation, potential downtimes, realtime monitoring when cooperating with manufacturers to get sensor data

Built With

• classification
• dataspell
• lstm
• polymerresearch
• python
• pytorch
• sagemaker