Professional Documents
Culture Documents
Donald C. Wunsch
Network Models for Wind Turbine
Department of Electrical and Computer
Engineering, Power Curve Estimation
University of Missouri-Rolla,
Rolla MO 65409 This paper examines and compares regression and artificial neural network models used
for the estimation of wind turbine power curves. First, characteristics of wind turbine
power generation are investigated. Then, models for turbine power curve estimation using
Edgar OHair both regression and neural network methods are presented and compared. The parameter
estimates for the regression model and training of the neural network are completed with
Michael G. Giesselmann the wind farm data, and the performances of the two models are studied. The regression
model is shown to be function dependent, and the neural network model obtains its power
Department of Electrical Engineering, curve estimation through learning. The neural network model is found to possess better
Texas Tech University, performance than the regression model for turbine power curve estimation under compli-
Lubbock TX 79409 cated influence factors. DOI: 10.1115/1.1413216
1 Introduction hub height and the velocity profile are known, and that the veloc-
During the last decade, models have emerged 15 to estimate ity profile does not change. The velocity profile is defined as the
and predict the power produced by wind farms. To maximize the difference in velocity as a function of height from the bottom to
use of wind generated electricity when connected to the electric the top of the turbine blade and is another factor which can influ-
grid, it is necessary to be able to estimate and predict the power ence turbine power production 8,9. The hub height velocity and
production of a wind turbine. However, this power production can the velocity profile are measured by the met. towers. However,
be influenced by many factors and usually fluctuates rapidly, im- due to the limited number of the met. towers, the variable terrain,
posing considerable difficulties on the management of combined the turbines distributed over a wide range on the wind farm, and
electric power systems. Several different techniques have been wind dynamics, the actual wind velocity and profile for each tur-
presented to estimate and predict the highly variable energy pro- bine are usually quite different from those obtained from the met.
duction. Two typical models are regression models 1,2 and arti-
towers. These are some of the main reasons that the measured
ficial neural networks 35. In this paper, we compare regression
and artificial neural network models for wind turbine power curve turbine power production versus meteorological tower wind speed
estimation using data from the Central and South West wind farm does not fall on the manufacturers power curve as shown by Fig.
near Fort Davis, Texas. 2. In the figure, the line represents the manufacturers warranted
estimated output power curve for the wind turbine with 500 kW
2 The Wind Farm and the Wind Power Generation rated power, which has been adjusted especially for the Fort Davis
wind farm to account for the difference in attitude with respect to
The Fort Davis wind farm consists of 12 turbines and two me-
sea level. The dots represent the measured wind turbine power
teorological towers met. tower Fig. 1. Data received from the
wind farm can be divided into two categories. The first contains output for a met. tower 10-minute average wind velocity. Which
data from the two met. towers such as wind velocities and direc- met. tower velocity to use is selected based on which direction the
tions measured at three elevations 10, 30, and 40m. The second wind comes from, i.e., if the wind comes from the east, the mea-
contains information about turbine power generation, such as av- sured wind speed from the east tower is used; if the wind comes
erage power outputs, voltages, and currents. The two met. towers, from the west, the measured wind from the west tower is chosen.
indicated with , are the sites for measurement of wind speed and In Fig. 2, the large difference of turbine power production at the
direction. Each dotted circle is the location of a wind turbine. same wind speed, as well as high power productions at low wind
Turbine power production depends on the energy contained in speeds and low power productions at high wind speeds, implies
the wind. The basic measuring unit of the energy contained in the that the wind at the turbine can be quite different from the wind at
wind is wind power density 6, or power per unit of area normal
the met. towers.
to the wind azimuth, calculated as Eq. 1, where PW is wind
power density W/m2, is air density kg/m3, and V is horizon- The air density in Eq. 1 also influences the energy contained
tal component of the mean free-stream wind velocity m/s. in the wind and therefore turbine power production. However,
has less influence on turbine power production than the wind
Pw0.5 V3 (1) speed because the dynamic range of is usually small and wind
However, both wind velocity and air density are generally not power is proportional to the cube of wind speed. In addition to the
constant. The hub height of the turbine is 40m above the ground. above factors, wind power production is also affected by other
The industry standard is to relate the power to the hub height wind factors such as seasons of a year, time of day 1, and wind fluc-
velocity 7. Such a relationship implies that the velocity at the tuation within a certain time period. In the following comparison
between neural networks and regression models, the only factors
Contributed by the Solar Energy Division of THE AMERICAN SOCIETY OF ME- considered are the 40m wind speeds and wind directions from the
CHANICAL ENGINEERS for publication in the ASME JOURNAL OF SOLAR ENERGY
ENGINEERING. Manuscript received by the ASME Solar Energy Division, January, two met. towers. Introducing other factors would make the speci-
2001; final revision, July, 2001. Associate Editor: D. Berg. fication of a function for a regression model quite difficult.
Journal of Solar Energy Engineering Copyright 2001 by ASME NOVEMBER 2001, Vol. 123 327
S y f x ,
i1
i i
2
(3)
3 Regression Model for Wind Turbine Power Estimation
p
3.1 Prediction by Regression Model. Regression models
quantitatively describe the variability among the observations by
y i x
j0
ji j i x 0i 1 (4)
partitioning an observation into two parts 10. The first part of
this decomposition is the predicted portion having the character- YX (5)
istic that can be ascribed to all the observations considered as a
group in a parametric framework. The remaining portion, called XTX XTY (6)
the residual, is the difference between the observed and the pre-
dicted values and must be ascribed to unknown sources. This can p
f xi ,
be expressed as f xi , f xi , 0 j0 j (7)
j0 j 0
y i f xi , i i1,2, . . . ,n (2)
where n is the number of the observations, y i is ith observation, 3.2 Regression Model for Wind Turbine Power Estimation.
xi (x 1i ,x 2i , . . . ,x ki ) is the predictor variable vector related to As stated in Section 2, there are many factors which can affect
observation y i , ( 0 , 1 , . . . , p ) is the parameter vector, and wind turbine power generation. However, we will only include
i is the error associated with ith observation. The function f is 40m wind velocity and direction measurements from the two met.
estimated by fitting a polynomial or other type of function. Fitting towers in our modeling and comparative analysis.
refers to calculating values of the parameters from a set of data. The following rules have been used for the regression model:
Usually, the estimate , a least squares estimate of , tries to a A polynomial f i () is chosen as the main representation of the
minimize the error sum of squares shown by Eq. 3. regression model; b Wind speeds are taken as the most important
5 Comparative Analysis of Regression Model to network model gets better performance for both low and high
wind speeds even for test data from April 1996 as shown by
Neural Network Model
Fig. 6.
The parameter estimates of for the regression and the weight
adjustments for the neural network training try to minimize the
6 Comparisons of Wind Turbine Power Estimation
error sum of squares. However, the regression model requires the
form of the function f in Eq. 2 to be predefined explicitly, mak- using Regression and Neural Network Models
ing the model function dependent but easy to evaluate. On the For wind turbine power estimation, data for 40m wind speeds
other hand, the neural network achieves the ability of function and wind directions from both of the met. towers are used. For
approximation through learning from data, making it data each turbine, 1500 data sets of 10-minute averages, which can
dependent. represent typical wind turbine power productions, are selected
There are also differences in obtaining parameter estimates of from March 1996 and are then used to do both the parameter
for the regression model and weight adjustments for the neural estimates for the regression model and the training of the neural
network. For the regression model, a matrix representation of Eq. network model. The selection of the training data is mainly based
5 is usually obtained, and then one function solving step is on the considerations of: 1 trying to make training data distribute
needed to get the least squares estimate section 3.1. For the equally in different wind speed sections, and 2 trying to make
neural network, the network weights are usually adjusted many the training data cover, equally, the wind coming from different
times perhaps thousands of times until a stop criterion is directions.
reached. The regression model for the power estimation is based on Eq.
Some evident relationships can also be seen between regression 8 which contains a polynomial with two wind speed variables
models and neural networks when the sigmoid nonlinear activa- and two products of wind speeds with transformed wind direc-
tion functions of neurons are reduced to linear functions. A gen- tions. Figure 7 shows the comparison among the 2nd, 3rd, and 4th
eral first-order neural network will be equivalent to a multi- order polynomial regression models for turbine power estimation
variable linear regression model, and a high-order neural network for test data for April 1996. It can be seen that the 3rd order
will be equivalent to a multi-variable polynomial regression regression model provides the best prediction results, and that
model. Thus, the nonlinear activation function is an important either the lower or the higher order will degenerate the results.
factor in a neural network for complicated function approximation This indicates that the low order polynomial is not suitable for this
problems. application; and the high order polynomial can introduce noise at
Figures 5 and 6 compare some of the properties between regres- high wind speed even though the parameters for high order items
sion and neural network models. This example uses measured are very small. The products of wind speeds with transformed
wind speeds from the west met. tower to predict the wind speeds wind directions have been found to improve the prediction accu-
of the east met. tower using regression and neural network mod- racy. By introducing the products, the estimation of turbine power
els. A data set of 1500 patterns in March 1996 is used for both the production can be adjusted for wind coming from different
parameter estimates of the regression model and the training of directions.
the neural network model. For the regression model, the relation The neural network is built and trained for each turbine. Each
obtained between input and output predicted value is a straight network has three layers. The nodes chosen in the three layers are
line when a linear function is predefined Fig. 5, reflecting the 4, 8, and 1 excluding bias. The input patterns are preprocessed
function dependent characteristics of the model. However, the measured wind velocities and directions, and the desired network
neural network achieves the input-output relationship through outputs correspond to the normalized measured turbine power 5.
learning. From the regression point of view, this relationship or The performance of the trained network is compared with that
equivalent function is more complicated Fig. 5. of the 3rd order regression model for the same test data for April
But the neural network model is more data dependent. In Fig. 5, 1996 in Fig. 8. Table 1 gives the absolute average power differ-
the parameter estimates and the training of the network weights ence in kW of turbine No. 6 between the measured and estimated
are based on the first 1500 patterns of data from March 1996. This power of various models for four months of test data. Both the
selection of patterns has much higher data density in low wind figure and the table show that the neural network has better per-
than in high wind, making the neural network bias its weights formance. The comparison does not show much difference be-
more toward the low wind data 16. However, when training data tween the high-order and general first-order neural networks.
are selected equally from different wind speed sections, the neural Even though all the items in Eq. 8 are taken as inputs to the
Fig. 8 Comparison of wind turbine power prediction by 3rd order regression and neural network models for
test data turbine No. 6, April 1996