The meteorological service texts is a window for the organizing committee, judges, and staff of the teams to obtain meteorological information about their related work, providing the necessary meteorological protection for the smooth holding of the event and the activities during the event. The existing text production requires manual writing and review, which is not efficient, while the fully automated text generation mainly relies on templates and has a fixed form. To address these problems, a knowledge graph-based texts generation method of Winter Olympics event meteorological services was proposed in combination with natural language processing technology. Focusing on content analysis and feature extraction from historical event meteorological service texts, the knowledge graph of alpine skiing events was constructed using meteorological data and historical event information. The method generated meteorological description texts based on real-time meteorological data and manuscript templates, and then obtained event impact results and generates corresponding texts based on knowledge graph query inference techniques. The experimental results show that the automatic generation of meteorological service text have good accuracy and readability, which helps the smooth promotion of the Winter Olympic events, and the text generation method has good application prospects for the special area.