NEURAL NETWORK BASED DC-DC CONVERTER FOR ELECTRIC VEHICLE APPLICATION WITH PV PANEL

Abstract

Renewable energy sources are becoming popular for all the automobile industries to replace the conventional energy sources. But the stability of the output is the major concern while moving from conventional energy sources to renewable energy sources.  Power electronics plays very important role in the stabilization of the output voltage. Still the stability can be improved by tuning the gate pulses for the power electronics circuits using Machine learning models. Here neural network model has been used to improve the stability of the DC-DC converter. Conversion efficiency can be improved using MPPT techniques. A precise VCO is also used, which can change the switching frequency dynamically based on the input voltage. After being built and simulated on 180nm CMOS technology, the harvester achieves maximum conversion efficiency. The Energy Harvesting (EHFEM) Project for Exercise Machines should respond to changes in real time using an elliptical pedalling action. Exploring the limits of microcontrollers with speeds around or beyond 100 MHz helps to focus future efforts, as the fastest calculation methods are also the most expensive. The viability of adding a neural network into the EHFEM system is investigated in this thesis subject, which compares numerous neural network types. This project uses a feedback control system to regulate a four-switch buck-boost converter utilising an artificial neural network, as well as voltage and current sensing circuits. The impact on system conversion efficiency is also measured in this work.