Short-term Wind Power Forecasting Based on Variational Modes and Improved Multiverse Optimization

In order to improve the wind power forecasting accuracy of wind farms, a combined forecasting method based on variational mode decomposition (VMD) and improved multiverse algorithm (IMVO) optimized extreme learning machine (ELM) was proposed. Firstly, the original load data was decomposed into modal components with the help of VMD algorithm, and divided into high-frequency and low-frequency sequences according to mutual information entropy to simplify the data. Then, based on the traditional multiverse algorithm, it was improved by introducing Tent chaotic map, exponential travel distance rate and elite reverse learning mechanism, and combined with ELM to obtain the IMV0-ELM prediction model. Finally, the prediction results of the high frequency and low frequency components are superimposed to obtain the final prediction result. The experimental results show that the prediction accuracy and convergence speed of the IMVO-ELM model have certain advantages compared with the methods of ELM, MVO-ELM and PSO-ELM. And under the data preprocessing with the help of VMD algorithm, the prediction accuracy is further improved, which verifies the effectiveness of the proposed combined prediction method.