Genetic algorithm-piecewise support vector machine model for short term wind power prediction

Short term wind power prediction is one of the effective ways to cope with the operational problems caused by the variability of the wind energy resource when a large penetration of wind power is integrated into electric power systems. In this paper, a model combining a Genetic Algorithm with a Piecewise Support Vector Machine (GA-PSVM) is developed that improves the precision of short term wind power prediction systems based on power curves of the wind turbine generator systems. A Genetic Algorithm (GA) is used to search automatically for the parameters of a Piecewise Support Vector Machine (PSVM) model. The resulting GA-PSVM model can be used to predict wind power generation from one to six hours ahead. Operational data from a wind farm in North China are used to evaluate the proposed model. The results show that the mean relative errors (MRE) of GA-PSVM model are 2.03% lower than that of the standard SVM model applied to the same data set.