Research on the application of artificial intelligence (AI) methods in power grid analysis has been widely conducted. The power flow dataset required for the training of AI models is not uniformly distributed, and to improve the quality of the dataset, related research has generally applied tuning methods that tend to have a more uniform sample distribution, resulting in improved performance of AI models. However, the lack of research on methods to evaluate the uniformity of distribution of the dataset poses an obstacle to the judgment of the validity of uniformity adjustment and the analysis of the impact of uniformly distributed datasets on the performance of AI models. This paper proposes a method to evaluate the uniformity of power flow datasets by using the calculation of distances in the highdimensional feature space of the flow datasets and plotting the minimum distance statistics as a way to evaluate the uniformity of the flow datasets. It was also tested on the power flow dataset of 36-node grid and evaluated separately for two uniformity levels, which can clearly represent the uniformity of distribution of the dataset in the high-dimensional feature space.