A new neural network predictor for digital control DC-DC converter

The purpose of this paper is to present a new neural network based method for digitally controlled dc-dc converters. In the presented method, the neural network predictor is used to modify the reference value of the output voltage in the PID control to improve the transient response. This neural network control operates in coordination with the PID control. At the first, the neural network is repeatedly trained to predict the output voltage using former predicted data for the modification of the reference. After the training, the reference in the PID control is modified by the predictor to improve the transient response. This training process proceeds repeatedly until the enough suppression of the output voltage against the load change is obtained. As a result, the undershoot of the output voltage is considerably suppressed from 3.4% to 2.0% compared with the conventional method. The convergence time is suppressed to 52% compared with conventional method´s one. Therefore, it is confirmed that the proposed method has the superior performance to control dc-dc converters compared to the conventional method.