Kusiak.qxd_Outsourcing_book_temp.qxd 26/02/2010 10:14 Page 63
Short-term Prediction of Windfarm Power – A Data-driven Approach
a report by
Andrew Kusiak
1
and Wenyan Li
2
1. Professor, Intelligent Systems Laboratory; 2. PhD Studient, Industrial Engineering, University of Iowa
The wind turbine power generation process is complex and poses Feature (parameter) selection is considered from two perspectives:
challenges in terms of performance optimisation. Predicting wind power domain knowledge and data mining. In terms of domain knowledge,
is a challenge because wind speed varies in response to changing air all the parameters of a wind turbine system can be classified into three
pressure, temperature and terrain topography. Traditional approaches for categories: controllable parameters (e.g. blade pitch angle, generator
the prediction of wind power versed in physics have limitations due to torque), non-controllable parameters (e.g. wind speed) and turbine
factors that have not been previously studied, including the turbine performance parameters (e.g. power output, rotor speed). To obtain
system, with numerous parameters described herein. an accurate prediction model using a data-mining approach,
appropriate predictors need to be selected. Data mining offers various
Recent advances in information technology have allowed the algorithms to perform this task.
3
For example, the boosting tree
continuous collection of data from wind turbines. The collected data algorithm (BTA), as well as the wrapper approach, can be used to
contain information about the nature of the wind and the turbine select the best predictors when integrated with the genetic or the
parameters. The availability of large volumes of data enables the wind best-first search algorithms.
energy transformation process to be modelled using data-mining
algorithms. The literature on data mining in wind energy has primarily Model Construction
focused on estimating and optimising power output. Model building
can be accomplished using a variety of learning algorithms, e.g. neural Direct and Integrated Approaches
networks (NN), support vector machines (SVMs) and so on. The ability of a wind turbine to extract power from the wind is a
function with three main factors: the measured wind speed, the power
This article discusses the short-term prediction of wind power using curve of the turbine and the ability of the machine to handle wind
data-driven approaches. The basic concepts of the data-driven fluctuations.
4
The key parameter determining wind turbine performance
methodology are introduced, including data sampling, parameter is wind speed. Wind speed is normally measured using an anemometer
selection, model construction and evaluation. Next, three industrial placed at the nacelle of a turbine. In some cases, in addition to turbine
case studies are provided, representing different approaches and anemometers, meteorological towers are used to provide additional
applications for the short-term prediction of windfarm power. The measurements of wind speed. However, these additional wind speed
article concludes by looking at future perspectives. measurements are not used directly to control individual turbines;
rather, they are applied for assessment of wind speed.
Data-driven Methodology
Data mining is a broad area and consists of a number of different Two basic ways of predicting short-term windfarm power are available.
strategies and techniques that are inter-related in complex ways. One is to directly use the parameter values measured at sensors on the
Below, two topics in this area – data pre-processing and model wind turbines in the past to predict the future power; the other is to
construction – are discussed. predict the future wind speed first, and then use the predicted wind
speed to predict the windfarm power. The former is the direct
Data Pre-processing approach, and the latter is the integrated approach.
The data used in the research were generated at a windfarm with
about 100 turbines. The data were collected by a supervisory control
Andrew Kusiak is a Professor in the Intelligent
and data acquisition (SCADA) system installed at each wind turbine.
Systems Laboratory at the University of Iowa. He is
Each SCADA system collects data for more than 120 parameters.
the author or co-author of numerous books and
Although the data are sampled at a high frequency, they are normally
technical papers, and his research interests include the
application of computational intelligence in wind and
averaged and stored at a lower frequency, such as 10-minute intervals
combustion energy, manufacturing, product
(referred to as the 10-minute average data).
development and healthcare.
E: andrew-kusiak@uiowa.edu
Data mining offers algorithms for finding patterns and relations in large
Wenyan Li is a PhD student in industrial engineering
data sets using machine-learning algorithms. It is widely recognised
at the University of Iowa, where she is a member of
that data pre-processing is a time-consuming step; for example, 80% the Intelligent Systems Laboratory. Her research
of the time involved may be spent on data sampling, feature selection
interests are data mining, computational intelligence
and process optimisation applied to wind power and
and so on.
1
Methods such as simple random sampling and stratified
the manufacturing industry. She obtained her MS in
sampling
2
can be used. Feature (parameter) selection is also regarded
2009 from the University of Iowa and her BS in 2005.
as an important task in data mining, and some algorithms have proved
to be effective in determining relevant parameters.
© TOUCH BRIEFINGS 2010
63
Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72  |  Page 73  |  Page 74  |  Page 75  |  Page 76  |  Page 77  |  Page 78  |  Page 79  |  Page 80  |  Page 81  |  Page 82  |  Page 83  |  Page 84  |  Page 85  |  Page 86  |  Page 87  |  Page 88  |  Page 89  |  Page 90  |  Page 91  |  Page 92  |  Page 93  |  Page 94  |  Page 95  |  Page 96  |  Page 97  |  Page 98  |  Page 99  |  Page 100  |  Page 101  |  Page 102  |  Page 103  |  Page 104  |  Page 105  |  Page 106  |  Page 107  |  Page 108