The invention discloses a wind power forecast method based on an error correction and promotion wavelet combination forecast model. According to the method, wind power history data is processed through utilization of a promotion wavelet decomposition technology, main characteristics of a power data sequence can be improved, frequency components with relatively obvious characteristics can be obtained, and the denoising effect can be achieved, thereby applying to various forecast algorithms. A suitable forecast model is selected according to the characteristics of high and low frequency components, the deficiency of a single forecast method can be removed, and the forecast precision can be greatly improved. According to the method, processing errors are corrected through utilization of an error layered analysis method. Compared with the method of obtaining a next moment error forecast value by directly using the forecast model, the method has the advantage that under the condition that relatively high fluctuation occurs in the errors, the next moment error condition and compensation strength can be analyzed precisely, the errors resulting from an error forecast process can be reduced, and the errors of the integrated forecast method can be reduced.