Introduction to Wind Power Forecasting Using Hybrid VMD–GPR Models with Vedic Time Alignment

This project checks methods in wind power forecasting by comparing Gregorian calendar based on seasonal alignments with the vedic lunisolar calendar parallely. Rather than using timestamps like most forecasting methods, this project seeks to determine whether periodic cycles based on nature’s cosmos could reveal correlational patterns of wind activity surges and enhance accuracy. This study exploits the SOLETE dataset from SYSLAB, Denmark, which consists of 15 months of power generation alongside weather data. The dataset underwent processing with the CleanTS tool (an R package) and it was transformed into Gregorian and Vedic time frameworks. Within both time frameworks, the forecast approaches a hybrid forecasting model integrating “Variational Mode Decomposition (VMD) with Gaussian Process Regression (GPR)” was designed and assessed [11][12 ]. The Vedic forecasting approach is slightly better as it gives RMSE of 2.5519 and MAE of 2.0763, while the Gregorian forecasting approach gives RMSE of 2.6123 and MAE of 2.1424. The MAE correlation analysis over months revealed differing patterns within the two forecasting approaches with vedic giving better correlation than gregorian. This suggests that the Vedic calendar forecasting approach is better than the gregorian calendar system, which is based on natural cycles and is lunisolar, it is more accurate in capturing the chaotic signal of wind patterns than the arbitrary gregorian forecasting approach. This project helps in research, questioning the standard time representation in forecasting models which uses the gregorian timestamps and gives idea that if we put natural cycles through alternative calendar systems will it enhance the accuracy of energy predictions, potentially updating grid integration and operational planning.