Data Generation for AI Prediction Tool for Savonius Rotor

 We are pleased to give an in-depth status report on our ambitious project to create an AI prediction tool for Savonius turbine power output as part of our continuous efforts to harness the power of artificial intelligence for renewable energy applications. This tool has the potential to completely change the way we forecast turbine performance by doing away with the need for time-consuming practical testing and expensive simulations. Let's examine our accomplishments to date, including the technical details of generating data with MATLAB.

Overview

Our goal is to develop an AI-powered approach that predicts Savonius turbine power output under various entrance angles using actual experimental data. The goal here is to increase the efficiency and accessibility of renewable energy sources, rather than just to advance technology.

 Block Diagram

Data Collection and Preprocessing

We started by gathering unprocessed experimental data from Savonius rotor trials to build the framework for our AI model. Important characteristics including wind speed, rotor speed, torque, and power output are included in this data. However, raw data is insufficient; it also needs to be checked and cleaned. We have thoroughly cleaned the data, addressing any missing or incorrect data points. Then, in order to simplify our model, we extracted the relevant features and did feature selection.

Summary

We've made significant advances in our quest to create an AI prediction tool for Savonius Rotor. We are getting closer to our objective of offering precise and effective predictions for turbine performance thanks to the data generation step in MATLAB. We expect new difficulties and fascinating chances for creativity as we advance. As we continue our mission to develop renewable energy using the power of AI, keep checking back for more updates.

Author: Kritin Singh

Date: 19/09/23

Data Normalization and PCA Analysis:

Data normalization was one of our project's key turning points. To make our data more consistent and analysis-ready, we leveraged the potent powers of MATLAB. After that, we used Principal Component Analysis (PCA) to understand the structure of the data. Which elements had the most effects on the performance of the turbine? That information was disclosed by the Pareto chart.

 

Diving into the Core of the Project:

We investigated the Savonius rotor, which is the project's central component, as our voyage progressed. To identify the discrepancies, PCA was performed on both adjusted and raw data. The outcomes gave us useful information that would help us optimize the design of the turbine for optimal effectiveness.

 

The Magic of Biplot:

For those who are unfamiliar with the idea, a biplot functions as a magic lens that displays the relationships between various variables and the principal components. It assisted us in our study by allowing us to see the complex connections between the numerous aspects that influence turbine performance. This stage was very important in helping us improve the Rotor's design.

 

Preliminary Model Building:

But the data analysis wasn't where our excitement ended. To predict rotor performance, we dared to construct preliminary regression models. We found our closest buddy in the MATLAB Regression Learner software, which allowed us to improve models and make predictions. In a surprising development, we forecast data for a blade entry angle of 15 degrees, a critical angle that can have a big impact on energy capture.

 

Author: Kritin Singh

Date: 12/10/23