In order to improve the forecasting ability of WRF mode against the gust wind speed, the probability density function matching method (PDF) is used to correct the wind speed errors of wind speed forecasting data from WRF model. The results show that: (1) The gust wind force forecasting based on the PDF method is significantly better than the WRF output. When the observed daily maximum wind speed force ≤ level 5, two forecasting results are both consistent with the observation. When observed daily maximum wind speed force ≥ level 6, comparing with the observed and the wind speed forecasting based on the PDF method, the output from WRF is weaker. (2) By comparing two forecasting results above different topographic conditions, it is found that the effects of the WRF and PDF methods on gust wind in the plain area where the observed wind force is weak are both good. However, in the mountain and valley areas where the observed wind force is strong, the effect of WRF forecast is obviously poor but the effect of the PDF method is improved compared with WRF. (3) By testing forecasting effect on the daily maximum wind speed of 81 national stations in Anhui Province in 2017, it is showed that the forecast error is basically the same as that of the past 5 years, which shows that the probability density distribution function based on the historical data from 2012 to 2016 can represent the joint distribution characteristics of the observed and WRF simulated maximum wind speed from 81 national stations in Anhui Province for many years. So it is reliable by using the PDF method to forecast daily maximum wind speed.