Elsevier

Applied Energy

Volume 88, Issue 11, November 2011, Pages 4024-4032
Applied Energy

Estimation of wind speed profile using adaptive neuro-fuzzy inference system (ANFIS)

https://doi.org/10.1016/j.apenergy.2011.04.015Get rights and content

Abstract

Wind energy has become a major competitor of traditional fossil fuel energy, particularly with the successful operation of multi-megawatt sized wind turbines. However, wind with reasonable speed is not adequately sustainable everywhere to build an economical wind farm. The potential site has to be thoroughly investigated at least with respect to wind speed profile and air density. Wind speed increases with height, thus an increase of the height of turbine rotor leads to more generated power. Therefore, it is imperative to have a precise knowledge of wind speed profiles in order to assess the potential for a wind farm site. This paper proposes a clustering algorithm based neuro-fuzzy method to find wind speed profile up to height of 100 m based on knowledge of wind speed at heights 10, 20, 30, 40 m. The model estimated wind speed at 40 m based on measured data at 10, 20, and 30 m has 3% mean absolute percent error when compared with measured wind speed at height 40 m. This close agreement between estimated and measured wind speed at 40 m indicates the viability of the proposed method. The comparison with the 1/7th law and experimental wind shear method further proofs the suitability of the proposed method for generating wind speed profile based on knowledge of wind speed at lower heights.

Highlights

► We model wind profile up to height 100 m using ANFIS ► The model uses measured wind speed at heights 10–40 m to estimate wind speed up to 100 m ► We compare estimated values with 1/7th law and experimental wind shear method ► Results indicate viability of the developed system

Introduction

Wind, as an energy resource has gained significant focus around the world. The power of wind is generally regarded as one of the very important sources of renewable, inexhaustible, and clean energy. Wind energy has become a competitor of traditional fossil fuel power plants with the successful operation of multi-megawatt sized wind turbines. However, this absolutely free energy source is not adequately available everywhere to build an economical wind farm. The potential site has to be thoroughly investigated at least with respect to wind speed profile and air density. The increased height of turbine rotor leads to more generated power because wind speed increases with height. In order to assess the potential of a wind farm site, it is imperative to have precise knowledge of wind speed profiles. This study attempts to estimate the wind profile using neuro-fuzzy method.

Renewable energy has been considered as one of the strong contenders to improve plight of 2 billion people who are not having access to modern forms of energy [1]. At the same time, at least another half a billion people living in the regions where the population is growing most rapidly, have limited or unreliable access to energy [2]. An accurate wind resource assessment plays a vital role for harnessing the power of the wind [3]. Wind assessment techniques are used to create wind maps on a local scale and micrositing of wind turbines, estimate vertical wind speed variations and long-term wind resource at a given site [4]. These techniques made it possible to conduct wind assessment studies all over the world like Qatar [5] and Nigeria [6], a few sites in Jordan [7] and Kuwait [8], a wind atlas of Quebec, Canada [9], offshore California [10], the Red Sea coast and in Egypt [11].

Potts et al. [12] used GIS based software (WindMap) to perform the wind resources assessment of Western and Central Massachusetts. Brower [13] developed maps for monthly and annual mean wind speeds in Iowa using wind speed data from 21 locations, which would be beneficial for interested wind farms developers to obtain reasonably accurate estimates of potential wind energy production in Iowa. Raichle and Carson [14] reported that ridges in the Southern Appalachian Mountain region are suitable for utility-scale wind development. Muzathik et al. [15] studied the daily, monthly and annual wind speed along with their prevailing direction at a site in Terengganu, Malaysia, and concluded that small (kW range) wind machines could be used to provide power during the northeast monsoon season.

In Saudi Arabia, a wind atlas was developed which presented the monthly mean wind speed contours and frequency distribution for all the months of the year [16]. Rehman and Halawani [17] presented the statistical characteristics of wind speed and its diurnal variation for several sites in Saudi Arabia. Rehman et al. [18] adopted net present value approach to compute the cost of energy generation at 20 locations. Rehman and Al-Abbadi [19] presented the calculated values of wind shear coefficients (WSE) using measured values of wind speed at 20, 30, and 40 m above ground level, for Dhahran. Al-Abbadi and Rehman [20] presented the wind speed characteristics including wind statistics, local values of WSE, Weibull distribution parameters, turbulence intensity (TI), and wind energy yield using wind speed measurements made at different heights for Gassim city.

Due to increasing emphasis on wind power utilization and accurate wind power assessment and better hands on information in future time domain, long-term wind speed prediction is very important for siting and sizing of wind power applications [21], [22]. On the other hand, the short-term forecasting of wind speed is important for improving the efficiency of a wind power generation systems [23] as well as for the integration of wind energy into the power system [24]. Furthermore, the forecasts in the range of days are related to the maintenance and resources planning of the wind power plants [25]. There are several methods in the literature used for the prediction of wind speed for different time durations and include different physical models, statistical methods, hybrid physical-statistical methods, artificial intelligence and neurofuzzy, and other modern methods [26], [27]. Recently, Li and Shi [28] presented a comparative study on the application of 3 types of neural networks (adaptive linear element, back propagation, and radial basis function) for the prediction of wind speed 1-h ahead. Their results showed that no single neural network model outperforms the others universally in terms of all evaluation metrics.

In this paper a neuro-fuzzy model is used to estimate wind speed at high altitudes based on measurements at lower heights in Juaymah city of Saudi Arabia. The neuro-fuzzy model exploits the capability of both neural network and fuzzy logic systems. Fuzzy logic theory allows better representation of a given system behavior using a set of simple rules although it is unable to tackle knowledge stored in the form of numerical data [29]. On the other hand, artificial neural network (ANN) has been shown to be capable of learning virtually any smooth nonlinear function with a high degree of accuracy through a learning process. However, it shows limited capability in handing systems represented by linguistic information. Jang [30] proposed a neuro-fuzzy system known as adaptive neuro-fuzzy inference system (ANFIS). The ANFIS architecture has been successfully used to model nonlinear functions, identify nonlinear components in a control system, and predict a chaotic time series [31]. ANFIS is used in this paper for the estimation of wind speed profile.

Section 2 of this paper discusses briefly the ANFIS system while section 3 describes its application for wind profile estimation. Section 4 describes the obtained results. Section 5 concludes the paper.

Section snippets

ANFIS based methodology

ANFIS was developed to serve as a basis for constructing fuzzy inference system (FIS), and its architecture is obtained by embedding the fuzzy inference system into a framework of ANN [30]. Fig. 1 shows the architecture of ANFIS model for 2 inputs (x and y). The functions of each layer are described below.

Implementation

The wind speed profile is estimated using the above mentioned ANFIS based methodology. As the Gaussian membership functions are the most common, they are considered in this paper. Different values of the fuzzy exponent were investigated and the value of 2 was found to a system that performs reasonably well. Moreover, several numbers of membership functions were considered to find optimum number. Based on many experiments, the number of membership functions (MFs) was selected as 10 for all the

Results and discussion

Wind fluctuates significantly with time and height as shown in Fig. 4. This makes the estimation of wind speed at higher hub heights a challenging task. Usually, wind speed measurements are made at 10 m or so above the ground level. The wind profile at higher heights is needed to assess the wind energy potential. The estimation of wind speed at higher height based on 1/7th wind power law is a matter of concern and raises questions about its accuracy. The other practice is to install wind masts

Conclusion

This paper demonstrated the viability of ANFIS method for the estimation of wind speed at a higher heights based on wind speed knowledge at lower heights. Using wind speed at 10, 20, 30, and 40 m, we were able to generate the wind profile up to 100 m. The system was validated using wind speed at heights 10, 20, and 30 m to estimate the wind speed at height 40 m. In this case the mean absolute percent error between the estimated value and the measured value is about 3%. Moreover, the generated wind

Acknowledgment

The authors acknowledge the support of King Fahd University of Petroleum and Minerals.

References (33)

Cited by (187)

  • An overview of deterministic and probabilistic forecasting methods of wind energy

    2023, iScience
    Citation Excerpt :

    Monfared et al. (2009)175 applied fuzzy logic to ANN for the prediction of wind speed. Mohandes et al. (2011)176 and Osório et al. (2015)177 adopted an adaptive neuro-fuzzy inference system (ANFIS) for wind speed and wind power prediction. Ma et al. (2017)178 implemented wind speed prediction using the generalized dynamic fuzzy neural network.

View all citing articles on Scopus
View full text