Analysis of wind power generation and prediction using ANN: A case study
Introduction
Wind power is a significant alternate source of energy in these times of energy crisis. Advances in wind turbine technology and identification of rich wind resources in many areas improve prospects for the wind power industry. This motivates researchers for the analysis and prediction of wind power generation [1], [2], [3], [4], [5], [6].
Prediction of wind power generation comprises of several modeling techniques that combine meteorological and historical generation data [4], [7]. In order to achieve the highest possible prediction accuracy, the methods should consider appropriate parameters and data that may indicate future trends. Several studies have emerged which estimate and predict the power produced by wind turbines [8], [9], [10], [11], [12]. In the literature, artificial intelligence techniques such as neural networks, fuzzy logic, etc. are found to be more accurate as compared to traditional statistical models [13], [14].
Wind energy has become a techno-economically viable source of energy in the renewable power sector of India for sustainable power generation. For the present study, data collected from seven wind farms located in Muppandal, Tamil Nadu (India) for a period from April 2002 to March 2005 are used. The artificial neural network (ANN) model constructed using these data shows a good agreement between the estimated and the measured values.
In this paper, the wind power generation is studied as a function of input variables in terms of collected data. Then, the development, training and validation of neural network model for the prediction of energy from wind farms are discussed.
Section snippets
Brief overview of wind generation in Tamil Nadu (India)
India now ranks 4th in the world after Germany, Spain and USA in wind power generation with an installed wind power capacity of 4434.5 MW [15]. The state of Tamil Nadu is in the southern region of India at latitude between 8°5′N and 13°35′N and longitude between 76°15′E and 80°20′E with an area of 1,30,058 km2. The electric power supply grid of this state is connected with the nearby three states (Karnataka, Andhra Pradesh and Kerala) through 400 kV lines. The conventional power plant capacity of
Available wind power
The meteorological factors like, wind speed and air density greatly influence the wind power generation. The available power at the wind turbine in watts is given bywhere ρ is the density of air in kg/m3, A is the swept area of wind turbine in m2, and V is the wind speed in m/s. Of these variables, the wind speed has a major influence on wind turbine power output since the power output varies with cubic value of wind speed. The air density variation during different period of a year
Analysis of input parameters on wind power generation
The data are collected from Muppandal, Tamil Nadu (India) for a period from April 2002 to March 2005. The data covers 137 wind turbines with a total capacity of 37.225 MW from seven wind farms as given in Table 1. The wind speed data are the major parameters for wind energy generation and utmost care is taken in its collection. Two different sources are used to collect data for this locality-one, through the India Meteorology Department, Chennai, India and other from a private wind farm where
Artificial neural network
ANNs are widely accepted as a technology offering an alternative way to tackle complex and ill-defined problems. An ANN is an information processing pattern which works in a way that a human brain processes information. The structure of this information processing system is composed of highly interconnected processing elements, called neurons working in parallel to solve problems. A neural network helps when it is highly complex to formulate an algorithmic solution and also where there is a
Conclusions
The developed neural network model offers a reliable indication of the wind power output from wind farms by using the input parameters—average wind speed, relative humidity and generation hours. Generally, wind speed has direct influence on power generation, but it is seen that wind speeds higher than the rated wind speed of the turbine and high generation hours are important for high power generation. However, seasonal variation and diurnal variation of wind speed are nature dependent. The
References (19)
- et al.
Analysis of wind potential and characteristics in North Aegean, Greece
Renew Energy
(2004) - et al.
A review on the development of wind energy in Turkey
Renew Sustainable Energy Rev
(2004) - et al.
Evaluation of wind energy potential and electricity generation on the coast of Mediterranean Sea in Egypt
Renew Energy
(2006) Wind energy development in Tamilnadu and Andhra Pradesh (India) Institutional dynamics and barriers—a case study
Energy Policy
(2000)- et al.
Forecasting with neural networks
Mar Struct
(2003) - et al.
Application of a control algorithm for wind speed prediction and active power generation
Renew Energy
(2005) Previsional estimation of the energy output of windgenerators
Renew Energy
(2004)- et al.
A classification mechanism for determining average wind speed and power in several regions of Turkey using artificial neural networks
Renew Energy
(2005) - et al.
Using neural networks to estimate wind turbine power generation
IEEE Trans Energy Convers
(2001)
Cited by (272)
An effective hybrid wind power forecasting model based on “decomposition-reconstruction-ensemble” strategy and wind resource matching
2024, Sustainable Energy, Grids and NetworksWind power deviation charge reduction using long short term memory network
2024, e-Prime - Advances in Electrical Engineering, Electronics and EnergyOnshore and offshore wind power generation forecasting using a novel flexible time-varying fractional nonlinear grey model
2023, Energy Conversion and Management
- 1
Tel.: +91 1332 285075; fax: +91 1332 285231.