Wind Power Day-ahead Prediction Based on LSSVM With Fruit Fly Optimization Algorithm

摘要

The accuracy of least squares support vector machine (LSSVM) for wind power prediction is greatly affected by its parameters. To solve the problem of the man-made choice of the parameter values, a model for day-ahead wind power prediction based on fruit fly optimization algorithm (FOA) is proposed in the paper. For day-ahead prediction, numerical weather prediction (NWP) including wind speed, wind direction, temperature and atmospheric pressure has great influence on wind power. LSSVM is adopted to model the non-linear relationship in the study. FOA is employed to search for the optimal parameters of LSSVM. The simulation show that the new method based on FOA has better prediction properties than the model based on particle swarm optimization.