A swarm based double Q-learning for optimal PV array reconfiguration with a coordinated control of hydrogen energy storage system

Abstract

The general PV array reconfiguration attempts to maximize the power output by weakening the mismatch loss under partial shading conditions (PSC). This easily causes a large power fluctuation and an increasing power regulation cost. To resolve this conflict, this work presents an optimal PV array reconfiguration to balance the power fluctuation via coordinating with a hydrogen energy storage system. Consequently, the power regulation cost can be effectively reduced, while the hydrogen selling and total profits can be increased. Besides, a swarm based double Q-learning (SDQ) is proposed to find a high-quality optimum of PV array reconfiguration, thus the total profit of PV array can be dramatically improved. Case studies are implemented to evaluate the performance of SDQ under different PSC. Simulation results for a 10 × 10 PV array show that the power regulation cost by SDQ can be reduced about 89.22% against to that without optimization under a discrete varying PSC, while the total profit can be improved from a negative value to a large positive value. Besides, the total profit of SDQ is 5.41% larger than that of Q-learning under a continuous PSC, while it is also 9.06% larger than that of butterfly optimization algorithm under a discrete varying PSC.