Abstract: Aiming at the problem of photovoltaic data missing caused by the equipment failure or the human interference, a reconstruction method for the missing data in photovoltaics is proposed based on the generative adversarial network and the crisscross particle swarm optimization algorithm. Firstly, the Wasserstein divergence generation antagonism network (WGAN-div) is used to learn the timing law and the coupling relationship of the photovoltaic data. Secondly, the reconstruction constraints are designed to optimize the noise input of the generator, so that the reconstructed samples are close to the actual samples to the maximum extent. As for the problem of optimizing the high-dimensional variables, the vertical and horizontal crossover algorithm is used to catalyze the optimization process of particle swarm optimization to prevent the premature problems during the optimization. Experimental results show that the proposed method has high reconstruction accuracy when the photovoltaic data contains a large number of missing values. This method is also suitable for the missing value reconstruction of similar data in the power system, which has a good application prospect.