Considering a regional grid with several wind farms integrated， the total wind power has a better regularity comparing to that of a single wind farm. An ultra-short-term wind power forecasting method is proposed based on the concepts of total wind power and distribution factor. The least-square support vector machine（LS-SVM） and Kalman filter are adopted respectively to forecast the total wind power and distribution factor recursively， so that the good regularity of total wind power can be restored. Case studies show that the method not only improves the forecasting accuracy but also reduces the distribution range of the forecasting errors.