Design of a Neural Network for Controlling the Output Voltage of a Step-Down DC/DC Converter

This article deals with the design of an artificial neural network (ANN) capable of mitigating fundamental error mechanisms in power systems, such as step changes in input voltage and load resistance. The primary objective is to create an ANN tailored for a DC/DC buck converter. The exploration includes the selection of an appropriate ANN type and its structural configuration. Additionally, a crucial aspect involves the identification and creation of training input-output datasets, ensuring the ANN effectively serves its purpose. In this context, the aim is to substitute the conventional PID controller with this ANN, tasking it with generating the necessary duty cycle.