A novel hybrid system based on multi-objective optimization for wind speed forecasting

Abstract

Wind power has demonstrated high-efficiency utilization in electricity system, accordingly, accurate and stable forecasting of wind speed is of vital significance in power grid security management and market economics. However, most former studies only consider either the accuracy or stability, with difficulty achieving the two targets simultaneously, which is insufficient for an effective forecasting method. This paper proposes a novel hybrid forecasting system that includes an effective data decomposition technique, a multi-objective optimization algorithm, a forecasting algorithm, and a set of comprehensive evaluation methods. In this system, the complete ensemble empirical mode decomposition (CEEMD) divides the original wind speed sequence into a set of intrinsic mode functions and then extreme learning machine (ELM) optimized by the multi-objective grey wolf optimization (MOGWO) is applied to achieve excellent forecasting performance. To validate the forecasting performance of the developed forecasting system, wind speed data at 10-min interval collected from Shandong Peninsula, China is considered as case study and comprehensive evaluations are introduced. The results demonstrate that the proposed hybrid system transcends the other compared single and traditional models and simultaneously realizes high accuracy and strong stability. Thus, the proposed CEEMD-MOGWO-ELM system can be effectively and satisfactorily used for smart-grid operation and management.