Wind Speed Prediction using LevenbergMarqardt Back Propagation Neural Network

Authors: Arpita Yadav, Kapil Sahu

Certificate: View Certificate

Abstract

Several fields of science and technology are adopting Artificial Intelligence as an effective tool in complex and overwhelmingly large data analysis. One such field is prediction problems where statistical prediction prove to be too complex to handle or are not highly accurate. In this paper, we devise a model for wind speed prediction based on the use of Artificial Neural Networks (ANN). Wind speed prediction plays an extremely critical role in generation of renewable generation of power and reducing the dependence on fossil fuels. Here the LevenbergMarquardt algorithm is used which employs back propagation and hence attains lesser time of convergence and overall average error. The performance metrics chosen are Mean Square Error (MSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE).

Introduction

Neural Networks have started affecting numerous areas of science and technology where human intervention has not been competent enough to process data and predict outcomes with high accuracy or within constrained time limits or both. One such area has been the prediction of wind speed which is a crucial factor in deciding or predicting the amount of wind power that can be generated by wind power stations.Prediction of wind power is challenging yet instrumental at the same time. This is because wind speed changes continuously with time due various natural parameters, leading to uncertainty in availability of amount of wind power that can be generated using it. If we integrate this wind power system directly to the existing power system, it will lead to a number of issues in terms of attaining good power quality, power system stability, frequency of generated power, rated terminal voltage, optimizing spinning reserve capacity, uncertainty in wind power in to unit commitment and reducing power dispatching issues in the grid power balance, and, then we have planning and economic problems including, economic load dispatching. Therefore present researchers are focussing on accurate prediction models using artificial neural networks.

Conclusion

From the previous discussions, it can be concluded that the proposed technique efficiently employs the Levenberg-Marquardt (LM) algorithm to predict wind speed value. The back propagation mechanism attains reduction in error in minimalistic epochs and predicts with an average MAPE of 14% (approx.) The MAE attained is 2.02km/hr and the MSE obtained is 6.391. The regression curve also augments the obtained results. As a standard convention, 70% data is utilized for training, 15% is utilized for testing and rest 15% is utilized for validation. Finally the overall regression is also analysed.