| Abstract: With the rapid development of the new energy vehicle industry, the intelligent operation of charging infrastructure has become a key link in achieving the balance between power supply and demand and the efficient utilization of energy. To address the problems existing in the current operation of new energy stations, such as multi-source data heterogeneity and insufficient identification of influencing factors, this study proposes a method for identifying key factors in station operation based on data fusion and machine learning. Firstly, the research integrates multi-source data including charging orders, station attributes, user behavior, POI, weather, and holidays, achieves data fusion through attribute alignment and record matching, and constructs a comprehensive indicator system reflecting the operational performance of stations. Secondly, Pearson correlation coefficient and mutual information method are used to quantify the correlation between different types of factors and operational indicators, so as to screen potential key factors. Finally, the LightGBM is employed to evaluate the importance of factors and identify the core driving factors that significantly affect indicators such as order volume, profit, and power consumption. The results show that preferential strategies, equipment utilization rate, and user participation are the main factors influencing the operational effectiveness of stations. This study provides data support and decision-making basis for the refined operation of new energy charging stations and the regulation of power supply and demand. |