Transient optimal control algorithm for DC-DC converter based on time-series neural network

Abstract:
In this paper, an adaptive predictive correction control method for DC-DC converters based on time series neural network is proposed. The method utilizes the time-series network to identify and predict the output voltage of the converter in the transient process, and compensates and corrects the reference value of the controller, thereby realizing the adaptive control of the converter. In this paper, the NAR model and the NARX model are used to simulate and verify the method. The results show that, compared with the traditional PI control, the method can effectively suppress the voltage fluctuation in the transient process, significantly shorten the adjustment time of the converter, and is expected to be applied in high-performance regulated power supplies.