Real-Time Modelling and Controlling of Boost Converter Using Supervised Learning

The paper proposes a methodology based on a supervised machine learning technique to model boost converter. Instead of mathematical modelling, data-driven modelling has been done. The input voltage, output voltage, time, reference and error data has been collected using MATLAB Simulink model. The supervised machine learning modelling has been done using support vector machine (SVM) and K-Nearest Neighbour (KNN) algorithm. The model with better accuracy has been used to predict the PI value as per reference and error to operate DC-DC-converter. The model has been analysed using confusion matrix, parallel co-ordinates plots, receiver operator characteristics (ROC) curve and scattering plot. The accuracy of data-driven modelling has been compared with mathematical modelling. The proposed algorithm has been tested on hardware proto type where data has been collected using KG045 voltage sensor and ACS712 current sensor.