Abstract: In order to realize local real-time control strategy, a distributed photovoltaic power generation control method based on affine strategy parameter robust optimization is proposed. A function which can transfer linear term of voltage and polynomial term of sunlight observation was introduced to effectively describe system uncertainty. A two-stage data-driven robust optimization model was used to co-optimize droop control and affine strategy parameters, so as to generate multi-sided data-driven uncertainty set with global information and solve control tuning problem. Compared with other distributed control strategies, experimental results show that the proposed method can realize real-time robust optimal control.