Probabilistic wind power forecast using sparse Bayesian learning of unified kernel function

An artificial intelligence method of sparse Bayesian learning (SBL), which is used for probabilistic forecasting of the wind power, is proposed in this paper considering the importance of solving the problem of the wind power forecast. The kernel function in the proposed sparse Bayesian learning in this paper is a kind of unified kernel function which is composed by Gaussian kernel function and other kernel function such as Cauchy kernel function and so on, At last the parameters of the unified kernel function are decided by the algorithm of Particle Swarm Optimization (PSO). A componential forecast strategy is used here to improve the forecast accuracy. According to the experiment results, the wind generation series is decomposed into several more predictable series by discrete wavelet transform (DWT), and then the resulted series are forecasted respectively using SBL algorithm. Through many trials, it is proofed that this wavelet- SBL model can forecast the future generation more efficiently and accurately compared with other methods.