Forecasting of Short-term Wind Speed by Expanded VMD with Optimized RVFLN

This paper represents a novel hybrid wind speed forecasting approach based on variational mode decomposition (VMD) and Random vector functional link neural network (RVFLN). Initially, VMD is being used to decompose original data of wind speed into a series of intrinsic-mode functions (IMFs). The IMFs are prepared using the RVFLN forecasting model both without and with weight optimization using the Levenberg-Marquardt algorithm (LMA). The performance of the forecasting model is evaluated using ten-minute and one-hour interval wind speed data collected in summer and winter from the Sotavento wind farm located in Spain. To determine the accuracy of the suggested approach the test was performed by using some benchmark techniques. From the result, it is observed that the proposed model is more effective than other benchmark technique.