Improvement of current balanced parallel soft-start operation for dc-dc converters with current prediction

This paper proposes a current balance control in parallel operation using neural network control in addition to the conventional digital soft-start control for dc-dc converters. The neural network predictor is used in the average current balance control to improve the voltage drop when the current balance operation starts. In the proposed method, the neural network is trained to predict the output current flowing through the load of the converter from its former sensed data. The neural network control operates to reduce the difference between the output current and the desired one in parallel operation and the voltage drop is compensated. From evaluation results, it is confirmed that the proposed method has a superior soft-start characteristic compared to conventional method.