Instantaneous Adaptive Controller Based on Neural Network for Buck Converter

Abstract

The proposal for this work is a technique for the control of a Buck DC-DC converter, called Instantaneous Adaptative Neural Network Controller (IANNC). The proposed control method uses a Perceptron neural network that finds to keep the converter's output voltage stable. Through the IANNC it is possible to establish the appropriate duty cycle for the pulse width modulation (PWM). The proposed method was compared with a classic Proportional Integral Derivative (PID) controller to validate the results obtained. The results were obtained through the numerical simulator PSIM and physical prototype, where the stability of the converter's output voltage against a resistive load was considered.