Performance prediction of Savonius Wind Turbine using adaptive neuro-based fuzzy inference system (ANFIS)

Abstract

Savonius wind turbine has got the upper hand in comparison to other wind turbine in terms of simplicity in construction and better opening torque at low wind speeds. In this background, models based on adaptive neuro-based fuzzy inference system (ANFIS) have been prepared in order to predict the output performance parameters like tip speed ratio and actual torque of Savonius wind turbine in response to the input parameters like number of blades in the turbine and wind speed. The current work utilizes the experimental data of Savonius wind turbine which has been mentioned in the literature. In the literature, Savonius wind turbine with 2, 3, and 4 blades are tested at different wind speed using wind tunnel to determine the tip speed ratio and actual torque delivered by them. The results predicted from the ANFIS models are substantial close to the experimental results. Moreover, the statistical pointers like R2, RMSE and MAPE are found to be 0.90, 0.066 and 18.26 for prediction of tip speed ratio and 0.97, 0.004 and 14.23 for prediction of actual torque, which highlight the precision of the models. Hence, it is finally realized that the developed ANFIS models are capable of finding the output parameter like tip speed ratio and actual torque of Savonius wind turbine with 2, 3, and 4 blades.