Artificial neural network based intelligent model for wind power assessment in India

Abstract

Wind resource assessment is essential to evaluate the future wind power generation from a wind farm. As wind power generation depends directly on wind speed, therefore accurate wind speed prediction facilitates wind power generation. In this paper generalized regression neural network is employed for accurate wind speed prediction. The performance of proposed approach is evaluated using publically available dataset of different cities in India. Air temperature, earth temperature, relative humidity, daily solar radiation, elevation, latitude, heating degree days, cooling degree days, longitude and atmospheric pressure are used as input variables. Correlation coefficient of 0.99909 is obtained during training and 0.95143 during testing of GRNN model. The proposed GRNN model is then utilized to find wind speed and power potential of major wind power generating sites of Andhra Pradesh, India. A comparison between the measured and forecasted wind speed and power values validate that generalized regression neural network is an appropriate technique for long term wind speed and power prediction.