Forecasting of wind speed using multiple linear regression and artificial neural networks
- PDF / 449,590 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 13 Downloads / 207 Views
Forecasting of wind speed using multiple linear regression and artificial neural networks Soukaina Barhmi1 · Omkaltoume Elfatni1 · Ismail Belhaj1 Received: 8 February 2018 / Accepted: 11 April 2019 © Springer-Verlag GmbH Germany, part of Springer Nature 2019
Abstract In this paper, two methods are developed for the prediction of wind speed, namely, the Multiple Linear Regression (MLR) and Artificial Neural Networks (ANNs) in north and south regions of Morocco for three years (i.e., 2011–2012–2013). The first method consists of determining the parameters which most significantly influence the wind speed in order to build a regression model between the predictors and the dependent variable. The second proposed approach is ANNs where the neural network chosen is the multilayer perceptron that uses the back-bropagation (BP) as a supervised learning technique for training. The results show that both MLR and ANNs models predict the wind speed at an acceptable correlation coefficient between the actual and predicted wind speed. However, the ANNs perform better in terms of statistical errors notably in terms of mean absolute error, mean absolute percentage error and mean square error. Keywords Wind energy · Wind speed · Prediction · Artificial neural networks · Multiple linear regression
1 Introduction Wind energy has become increasingly popular due to the limitations of the global fossil and nuclear fuel resources. Furthemore, wind energy is considered as a clean resource energy which doesn’t emit dangerous emissions, like those produced by fossil fuel power stations, that cause several human health problems. The wind power is greatly affected by the wind speed, and there exists a strong positive correlation
B
Ismail Belhaj [email protected] Soukaina Barhmi [email protected] Omkaltoume Elfatni [email protected]
1
Faculty of Sciences, Laboratory of High Energy Physics-Modeling and Simulations, Rabat, Morocco
123
S. Barhmi et al.
between them. In this sense, the accurate prediction of wind speed is indispensable in order to protect the security of the running systems, decrease the possibilities of wind power break down and protect the security of the wind power integration. Morocco is a Northern African country located between latitude of 21◦ and 36◦ in north and longitude of 1◦ and 17◦ in the west. It is bordered by the Atlantic Ocean to the west, the Mediterranean Sea to the north, Algeria to the east and Mauritania to the southeast, with an area of 750.810 km2 (289.90 mi2 ) [11]. Along with its key geographical location, Morocco benefits from a great solar and wind energy potential. Two major renewable energy source initiatives (i.e., the Moroccan wind and solar projects) have been launched in order to reach the national target of increasing the share of renewable energy sources in the energy mix up to 42% by 2020 [12]. In recent years, several scientists have focused their research on the wind speed prediction [24]. Debnath et al. [7] reviewed 483 energy planning models (EPMs) and revealed the use o
Data Loading...
