This article focuses on designing a neural network to control the output voltage of a step-down DC/DC converter. Previous attempts, where the neural network was trained based on voltage regulation ratio, revealed limitations as it failed to update the duty cycle when the load changed. The network structure was modified to address this issue with an additional input parameter, effectively eliminating the deficiency. Simulations were made to verify the network’s ability to recalculate the required duty cycle for various scenarios, including input voltage and load changes.
Šimko et al. (Thu,) studied this question.