With the global climate change, extreme precipitation events are frequently occurring, causing serious impacts on countries and societies. In order to analyze the distribution characteristics of extreme precipitation in the southwestern region of China, this study takes the annual precipitation data of Cengong County as the research object. Firstly, in this paper, the general modeling process of Generalized Extreme Value model (GEV) was introduced, and the rainfall data from 1960 to 2022 in Cengong County were grouped, and the maximum value of each group was extracted to form a new sequence, thus obtaining the research data for this paper. Secondly, the estimation values of the GEV model parameters are obtained using Maximum Likelihood Estimation (MLE) and profile likelihood estimation. After model diagnosis and comparison, the GEV model is selected as the modeling model. Then, the return levels corresponding to return periods of 10, 50, and 100 years were analyzed, illustrating that the GEV model possesses a notable extrapolation capability and can effectively compute return levels for extended return periods. Finally, the fitting effect of the GEV model under different starting years is analyzed, demonstrating that the GEV model is insensitive to the starting year and consistent in trend.