Buildings contribute to over one-third of global energy-related carbon emissions. Among these, university buildings, especially dormitories, have high energy consumption and density, making them a key focus for energy conservation and emission reduction. Considering that individual energy use behavior significantly impacts building energy consumption, it is crucial to effectively manage and regulate dormitory energy loads based on students' usage habits to reduce energy consumption. This study targets universities in southwestern China and proposes a two-stage energy optimization scheduling strategy based on data mining and flexible load control, tailored to students' energy use behavior. A survey was conducted to collect data on dormitory energy use behavior among university students. Typical energy use characteristics were extracted, and key influencing factors were identified. Subsequently, a two-step clustering analysis method was employed to categorize student dormitories into four major types, analyzing their main features. The type with the highest energy-saving potential was selected to establish a typical energy use model for student dormitories, producing typical energy load curves for summer and winter seasons. Regulatory demands were identified, and flexible load control was utilized for optimization. The research results indicate that employing demand-side management methods, with the scheduling of flexible loads, can smooth dormitory energy usage curves while significantly reducing energy consumption. Additionally, energy-saving potential can be further explored through building design optimization and the incorporation of renewable energy sources.