MEDIUM TERM WIND SPEED FORECASTING USING COMBINATION OF LINEAR AND NONLINEAR MODELS

Abstract

A multistage technique has been proposed for improved medium-term forecasting of windelectric power generation of Wind Power Plant (WPP). This technique is based on an amalgamation of both
linear and non-linear methods i.e. Linear Regression, Artificial Neural Networks (ANN), and Support
Vector Machine (SVM) respectively. In the first stage, the input data of wind speed obtained from the
Numerical Weather Prediction (NWP) model has been extracted to localize wind data to the given WPP site.
For this purpose, the best grid is selected in the zone of WPP to train the ANN, SVM, and Regression
Models. In the second stage, ANN, SVM, and Regression Models are applied separately to the chosen NWP
data. Different 72-hours ahead forecasts of wind-electric speed are then combined at the third stage by using
suitable weights. In the final stage, this forecasted data is re-clustered to give an accurate forecast.
Moreover, the performance of the proposed multistage model is compared with actual wind-electric speed
generation data. Furthermore, the proposed ensemble model achieves 0.13 percent of mean absolute
percentage error (MAPE) and 0.19 percent of root mean square error (RMSE) at 72nd hour. It has been
shown that the proposed model performs better than other medium-term forecast prediction models.
Consequently, this work will also allow bringing balance in electricity in supply and demand in the future.