Modified Multi-kernel Extreme Learning Machine-based Electric Load Prediction
This work made a preliminary attempt to combine the multi-kernel extreme learning machine with variational modal decomposition and PSO-GA for electric load prediction.The model was designed to decompose the original electric load sequence into sub-sequences with different frequencies,and to integrate the sub-sequences with the modified multi-kernel extreme learning machine,thereby achieving reconstruction of the sub-sequences and obtaining the final prediction result.The adoption of the multi-kernel extreme learning machine could enhance the global search ability of the model for different features when coping with multi-feature load data,which were difficult to characterize using uni-kernel extreme machine.In a comparative test using load data of an actual region,the predictive model established in this work achieved more accurate prediction result than those from the selected reference short-term predictive models.
NETL
NSTL
万方数据
