Multi-step depth model predictive control for photovoltaic power systems based on maximum power point tracking techniques

Abstract

To increase the PV power generation efficiency, this paper proposes a multi-step depth model predictive control algorithm based on maximum power point tracking techniques. The multi-step depth model predictive control algorithm combines deep neural networks and model predictive control to improve control performances. Deep neural networks can improve prediction accuracy and can reduce steady-state oscillations. The model predictive control technique can improve the tracking speed and dynamic performance of photovoltaic power systems. Compared with a perturbation and observation algorithm, an incremental conductance algorithm, and a model predictive control algorithm, the simulation results verify that the multi-step depth model predictive control algorithm can effectively obtain the fastest tracking speed (0.001 s), the smallest steady-state oscillation (6 W), and the highest power generation efficiency (99.06%) under the step, ramp, sine, and real stringent dynamic irradiance and temperature cases. A hardware-in-the-loop experimental test verifies the validity of the multi-step depth model predictive control algorithm.