Abstract: To improve the accuracy of the prediction results, a prediction model is designed based on VMD- IWOA-LSSVM(Variational Mode Decomposition-Improved Whale Optimization Algorithm-Least Square Support Vector Machine). The original load data is decomposed into multiple sub-sequences by the variational modal algorithm. The decomposed data is entered respectively least squares support vector machine optimized by an improved whale optimization algorithm through population mutation strategy and neighborhood search extension. The final prediction result is obtained after the prediction results of each sub-sequence are added. Comparing to simulation experiment VMD-WOA-LSSVM and VMD-IWOA-LSSVM, mean Absolute Percentage Error of VMD- IWOA-LSSVM drops 0. 17 and 0. 33 respectively at the April 1 and August 1 than VMD-WOA-LSSVM. It proves that the short-term load prediction model of the VMD-IWOA-LSSVM can effectively improve the accuracy of power load forecasting.

Key words: variational mode decomposition(VMD), least squares support vector machine(LSSVM), improved whale algorithm( IWOA), neighborhood extended search strategy, population mutation strategy, load forecasting