The invention discloses an automatic classifying method of weather types. The automatic classifying method comprises the steps: selecting weather data; extracting a wind speed sequence Wpre, converting the wind speed sequence to wind-like energy sequence WEpre, and setting different wind process interval threshold values T1 and T2; filtering the wind-like energy sequence WEpre, filtering a high-frequency interference signal to obtain a low-frequency trend signal WA; acquiring a local extreme value sequence Ex; acquiring a starting index of each weather process; acquiring various weather processes Vi. By adopting the method, results of different error states in a future result can be predicted, an estimation interval with large uncertainty state is enlarged, an estimation interval with small uncertainty state is narrowed, the probability prediction result is optimized, the probability prediction precision is improved, the operation cost of a power grid can be effectively reduced, and the economical efficiency of the operation of the power grid can be improved.