Multi Step Ahead Wind Speed Forecasting Using Long Short Term Memory Recurrent Neural Network
Firuz Ahamed Nahid1*, Jubaer Alam 1, Khadiza Akter1.
1Department of EEE, IUBAT-International University of Business Agriculture and Technology, Dhaka, Bangladesh
*Corresponding author: E-mail: fanahid@iubat.edu
ABSTRACT: This paper proposes a multi-step ahead wind speed forecasting approach utilizing Long Short Term Memory Recurrent Neural Network – LSTMRNN (a deep learning technique). Accurate wind speed forecasting is the prerequisite to harvest maximum power from windmills. Proper forecasting of wind speed directly impacts the generation, demand management and unit commitment of windmills. In this research, one year’s historical dataset consisting of wind speed, relative humidity and temperature of thirty minutes interval has been considered to train the proposed LSTMRNN model. The model’s performance has been evaluated by Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The performance of the proposed model has also been compared with another machine learning model called Convolutional Neural Network – CNN. The results show that the proposed model can forecast wind speed with much better accuracy in comparison with CNN.
KEYWORDS: Machine Learning, Wind Speed Forecast, LSTMRNN, RMSE, MAE, MAPE, CNN, Deep learning and windmill.
Introduction
To achieve the goals of sustainable development, it is a must to develop and extract energy from renewable resources rather than conventional energy sources like fossils (oil, gas and coal) and so on, which are declining day by day. There are also issues like price hike of fuels, CDM (Clean Development Mechanism), change in climate, insufficient and unreliable supply of power in developing countries etc. (Hu & Chen, 2018) To keep pace with the development on this 21stcentury, the demand of energy is increasing locally as well as globally; which is forcing us towards a crisis of global energy. To maintain the development in a sustainable way, we must be concerned about the environmental issue.
