As a mature and large-scale form of new energy power generation, wind power has been widely used and developed in countries all over the world. Wind power has the characteristics of uncertainty, and it must be accurately predicted to ensure the normal operation of the power system after grid connection. This paper reviews the traditional methods and new research trends of wind power forecasting. Firstly, the physical methods, time series methods, artificial intelligence methods and combined methods are summarized. Then, the research progress of several important development directions of wind power forecasting:spatial correlation forecasting, cluster forecasting, uncertainty forecasting and ramp forecasting are highlighted. After reviewing the existing wind power forecasting methods, the research direction in this field is further prospected. This work is supported by Science and Technology Project of State Grid Corporation of China (No. 52272217000Z).