Leveraging AI for Superior PV Energy Predictions: A User-Friendly Approach

In this talk, Eaton demonstrates the complete workflow for training machine learning models for photovoltaic (PV) energy predictions using MATLAB^® app-based tools for AI. This project aimed to streamline the process of importing, preprocessing, and analyzing data, and ultimately deploying predictive models in an interactive, user-friendly manner.

Accurately predicting PV energy output is crucial for optimizing the performance and reliability of solar power systems. Traditional methods of data handling and AI model training can be time-consuming and complex. Eaton addressed these challenges by leveraging MATLAB low-code tools for AI, enabling efficient and effective model development and deployment. A new app was also developed with MATLAB App Designer, which provides a user-friendly option to train a model as well as predict PV energy with new data sets. You will also hear about edge deployment and AI compression/TinyML techniques.