Design of neural network fractional-order backstepping controller for MPPT of PV systems using fractional-order boost converter

The main objective of this article is to apply the fractional calculus for establishing a novel design of photovoltaic (PV) system. In order to enhance the efficiency and robustness of the maximum power point tracking (MPPT) approach, a fractional-order (FO) DC-DC boost converter is proposed for a PV system. Due to the nonlinearity of the PV module, an artificial neural network (ANN) loop has been used to consistently generate an optimal reference voltage. Using FO control, an incommensurate FO backstepping controller (FO-BSC) has been ultimately integrated for tracking the maximum power point in the presence of tremendously atmospheric conditions and load changes. In this context, the asymptotic stability is guaranteed via fractional Lyapunov function. According to Grunwald-Letnikov fractional definition, the FO dynamic equations of the proposed converter have been derived using the principle of the average method, and the FO components of the converter are successfully approximated via Oustaloup rational approximation. MATLAB/Simulink has been used to confirm the validity and the accuracy of the constructed model, which is simulated by changing orders of the fractional components. In addition, the performance of the proposed boost converter has been verified under an open-loop test and a closed-loop test. The results have proved that the efficiency of the extracted maximum power has been improved with an average up to 99% in comparison with the conventional model based on a BSC.