A novel wind power prediction approach using multivariate variational mode decomposition and multi-objective crisscross optimization based deep extreme learning machine

With the increasing proportion of wind power, effective wind power prediction plays a vital role in the stable operation and safety management of power systems. Most studies focus only on improving prediction accuracy but ignore prediction stability. To address this issue, a novel hybrid model based on multi-objective crisscross optimization (MOCSO) is proposed to enhance prediction stability. In the data preprocessing stage, the multivariate variational mode decomposition (MVMD) is first employed to simultaneously decompose wind power, meridional wind velocity, and zonal wind velocity, aiming to overcome frequency mismatch among different series and realize synchronous time-frequency analyses of wind velocity and wind power series. In the multi-objective optimization stage, to ensure prediction accuracy and stability, MOCSO is implemented to optimize the key parameters of deep extreme learning machine (DELM) model. Finally, three cases and multiple evaluation criteria are elaborated to comprehensively evaluate the proposed hybrid model. Experimental results show that MOCSO outperforms three state-of-art multi-objective optimization algorithms, and the proposed hybrid model has significant advantages over other models involved in this study.