Fast neuro-adaptive control of DC-DC buck converters: Design and implementation

A fast neuro-adaptive compensation based control scheme is proposed for dc-dc buck converters. The compensation of uncertainties is obtained by design and development of a single functional layer Hermite neural network. The estimates are further utilized for subsequent compensation in the online adaptation process using backstepping control. The stability of over all closed loop converter equipped with the proposed control is proved to be asymptotically stable under Lyapunov stability criterion. Extensive simulation and experiments are conducted to evaluate the response of buck converter under the action of proposed control at wide operating points. Further, the results are compared with recently published relevant control method. Finally, the performances indices suggests a significant improvement in the transient performance by yielding lesser settling time and lower peak overshoot/undershoot under proposed control, thereby confirming the validity proposed scheme.