The embodiment of the invention provides a method and a device for establishing a wind power forecasting model and a wind power forecasting method and device, which relate to the field of wind power forecasting of a power system. The method and device normalize the acquired historical wind power parameters and the corresponding historical wind power to obtain training samples and output samples, and take a pre-stored Gaussian function as a radial basis function and a linear activation function as an output function to establish a radial basis function neural network, and then adopt a K. The mean clustering algorithm trains the radial basis function neural network based on the training samples to determine the center of the radial basis function, At the same time, the least square recursionmethod is used to determine the weight of the radial basis function based on the training samples and output samples, so as to establish a wind power prediction model, which can effectively avoid theuncertainty of wind power and provide accurate forecasting information for dispatchers.