This paper presents a control strategy for a DC–DC buck converter that combines a double-hidden-layer recurrent neural network (DHLRNN) with a fractional-order sliding mode controller (FOSMC) and compensates for disturbances through a finite-time disturbance observer (FTDO). The inclusion of the fractional-order term in the SMC is to improve the control accuracy of the DC–DC buck converter. To counteract the adverse effects of system nonlinearities such as disturbances and uncertainties, an FTDO is designed to compensate for the mismatched disturbances. In addition, a DHLRNN is employed to approximate the nonlinear function of the converter system. The effectiveness of this method is verified by experimental results on the converter. Compared with other control methods, this method provides higher voltage tracking accuracy and faster dynamic response under various operating conditions.
Liu et al. (Thu,) studied this question.