Effect of time-duration of neural network based control on transient response of DC-DC converter

Abstract

In this paper, we study a neural network based digital control method for dc-dc converters. In particular, we consider the time duration effect of the neural network control term on the improvement of the transient response of dc-dc converter. To obtain a fast response in the transient state, we focus on the utilization of the neural network predictor in addition to the conventional PID control. In the presented method, the neural network is trained to improve the transient response of output voltage of converters by the modification of the reference value in a conventional PID control. The training process of neural network proceeds repeatedly until the enough suppression of the output voltage against the load change is obtained. To realize optimal control of the neural network control, we also investigate the optimal duration of the reference modification by the neural network control, simultaneously. As a result, the undershoot of the output voltage is considerably suppressed from 3.3% to 1.7% compared with the conventional PID method. The convergence time is suppressed to 48% compared with conventional method's one. Therefore, it is confirmed that the presented method has the superior performance to control dc-dc converters compared to the conventional method.