Abstract:Based on the monthly runoff data of 9 hydrological stations in the Huaihe River Basin from 1956 to 2016, Pettitt nonparametric test, GAMLSS model and flood frequency analysis model were used to reveal the variation of flood frequency in the middle and upper reaches of the Huaihe River. Sexual and non-stationary conditions under the conditions of flood intensity and floods caused by the impact. The results showed that:Huangchuan, Hengpaitou and Bengbu Stations did not change significantly, the other six stations of the mean or variance variation, variation time is mainly concentrated in 2000 or so. The optimal fitting distribution function of Huaihe River Basin is Weibull, followed by Lognormal distribution. The non-stationary model of Bantai, Jiangjiaji and Hengpaitou Stations were selected, and the other six stations choose the stationary model. While the optimal fitting function based on GAMLSS model has good fitting effect. The divergence curves of the Hengpaitou Station and the Jiangjiaji Station in the nonstationary station show a fluctuating trend with the time series. The site distribution of the station is 25% and 75%, and the actual distribution is not very reasonable. The rest of the site stability of the overall distribution of the site is more reasonable. The design value of the Pearson type Ⅲ distribution design under the condition of 10 years and 20 years of design flow rate and stationary conditions is not very different, but in 30 years, 50 years and 100 years the difference between the design flow of a gradual increase. County, Wangjiaba, Lutaizi and Bengbu Stations design flow value of more than 10000 m3/s. flooding station and Bengbu Station flood amplification factor increases with time and is greater than 1, a hundred years in the event of less than 80 years. Huangchuan Station, Wangjiaba Station and Fuyang Station flood amplification factor are less than 1, and the 100-year re-emergence period will be more than 300 years. The maximum annual peak flow of each station is highly correlated with the Huaihe River Basin and the flood area in Anhui Province, and basically has passed the 95% or 99% significance test.