Short-term Wind Power Predication Based on Feature Engineering and SOA-SVM

Abstract

Contraposing to the issues of strong randomness, large volatility, and high prediction error of wind power, this paper proposes a short-term wind power prediction method based on feature engineering and support vector machine optimized by the seagull optimization algorithm (ISOA-SVM). First, in feature engineering, the abnormal values are preprocessed and utilizing the Pearson correlation coefficient (PCC) method to analyze the correlation between wind power and every feature. On this basis, a set of new features are constructed. Then, in the construction of the combined model, the seagull optimization algorithm is utilized to optimize the parameters of the support vector machine, so as to realize the complementary advantages of the optimization characteristics of the seagull algorithm and the support vector machine in simplifying the high-dimensional space complexity problems. Finally, the short-term wind power prediction model is constructed based on the ISOA-SVM model. The emulation experiment is carried out with the relevant measurement data of a wind power field. The results show that, in relation to other models, the proposed ISOA-SVM model has more excellent precision and dependability of short-term wind power prediction, which provides a certain reference for power dispatching.