Artificial Neural Network Based on a Predictive Current Control in a DC-DC Buck Converter

Predictive control is a modern control strategy used in power converters that include switching devices in its topologies; is simple to understand and easy to be implemented, however, if the converter has to many operating modes the procedure may demand high computational requirements for high switching frequencies. This paper presents the inclusion of an artificial neural network in the controller, the predictive controller is used during the training phase and once the neural network is fine-tuned it can operate without the predictive control algorithm, minimizing the computational cost. The algorithm is validated by simulation results in Matlab-Simulink in a current control for a Buck converter.