A Single Layer Hermite Neural Network Based Direct Adaptive Control of DC-DC Buck Converter

Abstract

This paper presents a novel control scheme for the output voltage tracking problem of dc-dc buck converter. The proposed control structure integrates a direct adaptive backstepping control approach with a single functional layer Hermite neural network (HNN). Buck converter is a variable structure model with a highly uncertain loading pattern. Hence an estimator based on HNN is employed to estimate the load perturbations influencing the converter. The accurate and timely information about the unanticipated load resistance strengthens the control law in order to attain the desired objective. The online adaptive laws are derived based on Lyapunov stability criterion ensuring the overall stability of the buck converter equipped with the proposed control. The performance of the proposed method is evaluated by subjecting the buck converter to a wide range variations in load resistance, input voltage and reference output voltage. Further, the merits of proposed control are highlighted by comparing the performance with the standard adaptive backstepping control technique under identical conditions of simulation study. Performance indices such as peak overshoot, peak undershoot and settling time have been evaluated, which clearly indicate the outperformance of proposed control.