Abstract: Large-scale connection of wind power into distribution network brings great challenge to the secure and economic operation of the system. To fully characterize the impact of randomness and uncertainty of wind power on distribution system, firstly, by means of principal component analysis the dimensions of massive high-dimensional scenario of large amounts of wind power output were reduced, on this basis a stochastic scenario selection method of wind power, which was based on hierarchical clustering algorithm, was constructed, and the standard to determine the number of optimal clusterings was proposed to effectively divide the categories of wind power scenarios. Secondly, a hierarchical scenarios clustering method-based two-layer stochastic operation optimization system model for distribution network-natural gas network coupling system was built to improve the adaptability of operation scheme to wind power fluctuation from multiple time scales. Meanwhile, the Lagrange factor was led in, and the equivalent equation representation method of Karush–Kuhn–Tucker theory was presented to translate the built stochastic operation optimization model into single layer optimization problem to solve. By means of computing example, the proposed method is compared with existing operation optimization strategy, thereby the effectiveness and superiority of the proposed method are verified.