This study presents an application of clustering volatility patterns in high-resolution wind power data. Using 343 days of 4-s sampling rate data from a wind power plant, we segment 32,877 15-min windows and extract eight statistical features. After removing zero-power segments, 30,373 valid segments are analyzed through a two-stage clustering approach: first grouping by mean power into ten categories, then applying K-means clustering based on volatility features. The optimal cluster numbers are determined using the elbow method and validated with silhouette scores. Results identify the fluctuation patterns reasonably, demonstrate through PCA visualizations. Via scaling, the analysis of a 1,600 MW wind power plant is conducted, the wind power is decomposed to different fluctuating components corresponding energy storage power with different power response rates. This approach effectively smooth power output while meeting grid dispatch requirements, providing valuable insights for renewable energy integration and hybrid storage system operation.