A Novel Hybrid Short-Term Wind Power Prediction Framework Based on Singular Spectrum Analysis and Deep Belief Network Utilized Improved Adaptive Genetic Algorithm

Abstract

A machine learning based framework involving data-mining method was proposed in this paper. To begin with, a powerful signal decomposition technique (singular spectrum analysis, SSA) was used to divide the original wind sequence into several sub-series to form a potential feature set. Then, the optimal sub-series is screened as the input feature set based on a novel swarm intelligence optimization algorithm (adaptive genetic algorithm based on improved harmony search algorithm, IAGA). Finally, a more appropriate sub-feature set together with the corresponding machine learning model (deep belief network, DBN) were established. A series of simulations is conducted by utilizing actual dataset to validate the proposed method. Comparison results represent that the proposed SSA-IAGA-DBN method achieves high prediction accuracy and robustness in short term wind power prediction tasks.