-
Notifications
You must be signed in to change notification settings - Fork 0
/
thesis.tex
286 lines (224 loc) · 12.8 KB
/
thesis.tex
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
\documentclass[twoside,11pt,openright]{report}
\usepackage[latin1]{inputenc}
\usepackage[american]{babel}
\usepackage{a4}
\usepackage{latexsym}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{epsfig}
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage[labeled]{multibib}
\usepackage{color}
\usepackage{datetime}
\usepackage{epstopdf}
\usepackage{graphicx}
\usepackage{subcaption}
\usepackage{hyperref}
\usepackage{enumitem} %Used to make lists smaller
\usepackage{float}
\usepackage{placeins}
\usepackage[table]{xcolor}
\usepackage{todonotes}
\usepackage{menukeys}
\usepackage{longtable}
\usepackage{rotating}
%\usepackage{amssymb}% http://ctan.org/pkg/amssymb
%\usepackage{pifont}% http://ctan.org/pkg/pifont
\newcommand{\x}{\scalebox{0.6}{x}}%
\newcommand{\m}{\scalebox{0.6}{(m)}}%
\renewcommand*\ttdefault{txtt}
\newcommand\fnurl[2]{%
\href{#2}{#1}\footnote{\url{#2}}%
}
%\newcommand{\todo}[1]{{\color[rgb]{.5,0,0}\textbf{$\blacktriangleright$#1$\blacktriangleleft$}}}
% \newcites{A,B}{Primary Bibliography,Secondary Bibliography}
% see http://imf.au.dk/system/latex/bog/
\begin{document}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\pagestyle{empty}
\pagenumbering{roman}
\vspace*{\fill}\noindent{\rule{\linewidth}{1mm}\\[4ex]
{\Huge\sf Wind Power and Electricity Price Prediction Using Artificial Neural Networks to Support Decision Making in the Danish Electricity Market}\\[2ex]
{\huge\sf Brian Bak Laursen, 20071275}\\[2ex]
{\huge\sf Kristian Barrett, 20073457}\\[2ex]
\noindent\rule{\linewidth}{1mm}\\[4ex]
\noindent{\Large\sf Master's Thesis, Department of Computer Science, ICT Product Development\\[1ex]
\monthname\ \the\year \\[1ex] Advisor: Niels Olof Bouvin\\[15ex]}\\[\fill]}
\epsfig{file=logo.eps}\clearpage
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\pagestyle{plain}
\chapter*{Abstract}
\addcontentsline{toc}{chapter}{Abstract}
\definecolor{light-gray}{gray}{0.70}
In this thesis, we describe the foundation for modelling Artificial Neural Networks for prediction in the area of wind power and electricity pricing for the Danish electricity market. The thesis is divided into three areas of concern: (a) analysing the influential factors of wind power and electricity price; (b) constructing experiments for verification of the analysis and establishing the best Artificial Neural Network models for prediction; (c) determining feasibility of the approach by discussing experimental results as well as the potential for its practical use.
The main concepts of Artificial Neural Networks and time-series forecasting are presented. They include Machine Learning, Feedforward Neural Network, Supervised Learning with Backpropagation and Decision Support. The thesis highlights the characteristics of wind power and the electricity price based on concepts from existing prediction systems.
We facilitate the Feedforward Neural Networks with Resilient Backpropagation towards the prediction models. The models learn from relevant data that consists of influential historical data of either wind power or the electricity price.
The thesis illustrates the feasibility of Artificial Neural Network in the context of wind power and electricity price prediction through the experimental results and discussions thereof.
\chapter*{Resum\'e}
\addcontentsline{toc}{chapter}{Resum\'e}
Vi beskriver i dette speciale fundamentet for at modellere Artificial Neural Networks til forudsigelse af vindenergi og elektricitets priser i det danske elektricitets-marked. Vi inddeler problemstillingen i tre omr\aa der: (a) analyse af de betydningsfulde faktorer for vindenergi og elektricitets priser; (b) opbygning af eksperimenter der har til form\aa l at validere analysen og etablere de bedste Artificial Neural Network modeller; (c) bestemme anvendeligheden af vores tilgang med udgangspunkt i en diskussion af de eksperimentelle resultater og potentialet for brug i praksis.
Vi har pr\ae senteret de koncepter, der omgiver Artificial Neural Networks og time series forudsigelser. Dette inkluderer Machine Learning, Feedforward Neural Network, Supervised Learning med Backpropagation og Decision Support. Vi fremh\ae ver karakteristika for vindenergi og elektricitets priser baseret p\aa ~koncepter fra eksisterende systemer.
Vi anvender Feedforward Neural Networks med Resilient Backpropagation til forudsigelse. Modellerne bliver tr\ae net med relevant data, der best\aa r af de influerende faktorer for enten vindenergi eller elektricitets prisen.
Specialet illustrerer anvendelsen af det Artificial Neural Network til forudsigelse af vindenergi og elektricitets priser gennem de eksperimentelle resultater og en diskussion deraf.
\chapter*{Acknowledgements}
\addcontentsline{toc}{chapter}{Acknowledgments}
We are grateful for the guidance provided by our supervisor Niels Olof Bouvin throughout the entire process of writing our thesis. Time has never been an issue and the feedback has always been insightful which we very much appreciate.
We are also thankful to Jacob Styrup Bang for providing us with the computational power for achieving our experiments within a timely manner. At the same time we apologize for putting down the server.
Finally, we would like to thank our family and friends for always giving moral support and proofreading this thesis. Special thanks to the IT-guys for always joining us and listening to "boring" stuff about Artificial Neural Networks in the cafeteria at lunchtime. It will be dearly missed when entering real life.
\vspace{2ex}
\begin{flushright}
\emph{Brian Bak Laursen,}\\
\emph{Kristian Barrett,}\\
\emph{Aarhus, \today.}
\end{flushright}
\tableofcontents
\listoffigures
\newpage
\pagenumbering{arabic}
\setcounter{secnumdepth}{2}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Introduction}
\label{ch:intro}
\section{The Problem}
\input{sections/Introduction/TheProblem.tex}
\section{Related Work}
This section will give an overview of related work within the area of price and green energy prediction. The presented concepts will be elaborated as the thesis progress.
\subsection{Price Prediction Systems}
\input{sections/Introduction/PricePredictionSystems.tex}
\subsection{Green Energy Production}
\label{sec:greeEnergyProductionIntroduction}
\input{sections/Introduction/GreenEnergyProduction.tex}
\subsection{Decision Support Systems}
\input{sections/Introduction/StockDSS.tex}
\section{The Method}
\input{sections/Introduction/TheMethod.tex}
\section{The Hypothesis}
\label{sec:theHypothesis}
\input{sections/Introduction/TheHypothesis.tex}
\newpage
\section{The Structure of the Thesis}
\label{sec:structureOfTheThesis}
\input{sections/Introduction/TheStructure.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Main Concepts}
This chapter introduces and explains the technologies and concepts that are the foundation for the thesis.
\label{ch:foundations}
\section{Machine Learning}
\label{sec:machineLearning}
\input{sections/MachineLearning/MachineLearning.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newpage
\section{Artificial Neural Network}
\label{sec:annSection}
\input{sections/NeuralNetworkSection/NeuralNetworkSection.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\newpage
\section{Prediction}
\label{sec:predictionSection}
Being able to predict electricity prices and wind power is the basis for our thesis. Demand has an impact on the electricity price so this is handled here. This section will introduce and give selected examples of these concepts within the different areas.
\subsection{Electricity Demand}
\label{sec:ElectricityDemand}
\input{sections/Prediction/ElectricityDemands.tex}
\subsection{Wind Power Production}
\input{sections/PowerGenerationPrediction/analysisOfPowerGenerationPRediction.tex}
\subsection{Electricity Prices}
\label{sec:electriciyPrices}
\input{sections/SimilarDaysApporach/SDMandANN.tex}
\subsubsection{ARIMA Prediction}
\input{sections/ARIMA/ArimaPrediction.tex}
\subsubsection{Support Vector Machine Prediction}
\label{sec:svmPrediction}
\input{sections/SVMforecasting/SVMforecasting.tex}
\input{sections/PredictionSummary/PredictionSummary.tex}
\newpage
\section{Decision support}
This section introduces the concept of a Decision Support System and how it relates to Artificial Neural Network. Secondly, it presents how uncertain information (that is highly represented in the electricity market) can be handled through transparency and analysis.
\label{sec:dssAndUncertain}
\subsection{Decision Support Systems}
\label{sec:dssSection}
\input{sections/DecisionSupport/DecisionSupport.tex}
\subsection{Presentation of Uncertain Information}
\label{sec:uncertainInformation}
\input{sections/UncertainInformationPresentation/UncertaintInfo.tex}
\subsection{Summary}
\input{sections/SummaryDecisionSupport/decisionSupportSummary.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Dataset Analysis}
\label{ch:theANNs}
The purpose of this chapter will be to analyse what influences the price and wind production. The influences are based on concepts and knowledge from Section \ref{sec:predictionSection} and need further validation by analysing them in regard to the Danish electricity market. Furthermore the objective uncertainties presented in Section \ref{sec:uncertainInformation} will be elaborated on when analysing what influences the electricity price/wind power. The analysis will contribute to locating the potential input parameters to be used in the Artificial Neural Network.
\section{Data Collection}
\label{sec:dataCollection}
\input{sections/DataCollection/datacollection.tex}
\newpage
\section{Wind Power Analysis}
\label{sec:windPowerAnalysis}
\input{sections/GreenEnergyTrainingset/greenEnergyTrainingSet.tex}
\newpage
\section{Electricity Price Analysis}
\label{sec:ElectricityPriceAnalysis}
\input{sections/DataCollection/electricityPricePrediction.tex}
\section{Discussion}
\input{sections/NetworkDiscussion/dataSetDiscussion.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Forecasting Model}
\label{ch:forecastingModel}
The wind power production and electricity price time-series have similarities in relation to volatility, seasonality and are both related to the electricity market seen in the co-relation to consumption. We will use the same network structure and learning algorithm in both cases. This Chapter will describe the used Artificial Neural Network, the learning algorithm and the different strategies that will be experimented with in our model.
\input{sections/NeuralNetworkModelling/Overview.tex}
\input{sections/NeuralNetworkModelling/NeuralNetworkAnalysis.tex}
\input{sections/NeuralNetworkModelling/DataSetManipulation.tex}
\input{sections/NeuralNetworkModelling/PredictionStrategies.tex}
\input{sections/NeuralNetworkModelling/Conclusion.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Experimental Results}
\label{ch:experimentalResults}
This section describes the experimental results of the prediction simulations for both wind power and electricity prices. These experiments are done to verify or reject the assumptions from the dataset analysis. Before the actual experiments we will introduce the procedure for testing.
\section{Test Procedure}
\label{sec:testProcedure}
\input{sections/Experiments/TestProcedure.tex}
\section{Statistical Evaluation Methods}
\label{sec:statisticalEvaluation}
\input{sections/Statistics/Statistics.tex}
\newpage
\section{Wind Power Experiments}
\label{sec:windProductionExperiments}
\input{sections/Experiments/WindProductionExperiments.tex}
\newpage
\section{Electricity Price Experiments}
\label{sec:priceExperiments}
\input{sections/DataCollection/pricePredictionExperiments.tex}
\newpage
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Experimental Result Discussion}
\label{ch:experimentalResultDiscussions}
\input{sections/NetworkDiscussion/annDiscussion.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Conclusion}
\label{ch:conclusion}
\input{sections/Conclusion/Conclusion.tex}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\addcontentsline{toc}{chapter}{Bibliography}
\bibliographystyle{plain}
\bibliography{refs}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\chapter{Appendix}
\label{ch:appendix}
\section{Abbreviation List}
\label{sec:abbreviationList}
\input{sections/Appendix/AbbreviationList.tex}
\newpage
\section{Weather Data}
\subsection{Stations}
\label{sec:weatherStations}
\input{sections/Appendix/WeatherStations.tex}
\subsection{Format}
\label{sec:weatherDataFormat}
\input{sections/Appendix/WeatherDataFormat.tex}
\section{Price experimental results}
\subsection{Price}
\label{sec:priceResultAppendix}
\input{sections/ExperimentalAppendix/PriceAppendix.tex}
\section{Wind power experimental results}
\label{sec:windResultsAppendix}
\input{sections/ExperimentalAppendix/WindResultAppendix.tex}
\end{document}