High Efficient Online Double Q Learning Method Based Distributed MPPT in PV Systems under Varying Meteorological Conditions

In photovoltaic systems, which are one of the renewable energy production tools, continuous production of power at maximum levels and with the highest possible efficiency is one of the most important objectives. For this purpose, different maximum power point tracking (MPPT) algorithms are used in photovoltaics (PV) module connections. In the array and string structures, the power generation of PV panels decreases due to mismatch reasons such as partial shading, temperature variations, irregular panel surface pollution, aging, and production differences of PV panels. Since solar panels are made up of series-connected solar cells, if some of the cells produce low current, it reduces the current of all the series-connected cells, causing the entire panel to operate at lower performance. The study proposes tabular online double Q learning with unique reinitialization algorithm for maximization of extraction power from PV panels, achieving fast convergence to the maximum power point and maintaining in transient and/or steady-state conditions of irradiation. Simulation results show that the proposed DQL algorithm achieved the highest power extraction in all five irradiance scenarios, delivering 439 W, 575 W, 778 W, 577 W, and 325 W. Under uniform irradiance, DQL reached a tracking efficiency of 99.9%, outperforming the P&O (89.2%) and INC (96.3%) algorithms. The proposed method was compared with algorithms such as particle perturb & observe (PO) and incremental conductance (INC) under rapidly changing meteorological irradiation conditions, and the success and validity of the proposed method were demonstrated with graphs.