192 lines
8.0 KiB
TeX
192 lines
8.0 KiB
TeX
\documentclass[12pt,a4paper,faculty=ea,language=en,doctype=article]{ugent-doc}
|
|
|
|
% Optional: margins and spacing
|
|
%-------------------------------
|
|
% Uncomment and adjust to change the default values set by the template
|
|
% Note: the defaults are suggested values by Ghent University
|
|
%\geometry{bottom=2.5cm,top=2.5cm,left=3cm,right=2cm}
|
|
%\renewcommand{\baselinestretch}{1.15} % line spacing
|
|
|
|
% Font
|
|
%------
|
|
\usepackage[T1]{fontenc}
|
|
\usepackage[utf8]{inputenc} % allows non-ascii input characters
|
|
% Comment or remove the two lines below to use the default Computer Modern font
|
|
\usepackage{libertine}
|
|
\usepackage{libertinust1math}
|
|
% NOTE: because the UGent font Panno is proprietary, it is not possible to use it
|
|
% in Overleaf. But UGent does not suggest to use Panno for documents (or maybe only for
|
|
% the titlepage). For the body, the UGent suggestion is to use a good serif font (for
|
|
% LaTeX this could be libertine or Computer Modern).
|
|
|
|
% Proper word splitting
|
|
%-----------------------
|
|
\usepackage[english]{babel}
|
|
|
|
% Mathematics
|
|
%-------------
|
|
\usepackage{amsmath}
|
|
|
|
% Figures
|
|
%---------
|
|
\usepackage{graphicx} % optional: the package is already loaded by the template
|
|
\graphicspath{{./figures/}}
|
|
|
|
% Bibliography settings
|
|
%-----------------------
|
|
\usepackage[backend=biber, style=apa, sorting=nyt, hyperref=true]{biblatex}
|
|
\addbibresource{./references.bib}
|
|
\usepackage{csquotes} % Suggested when using babel+biblatex
|
|
|
|
% Hyperreferences
|
|
%-----------------
|
|
\usepackage[colorlinks=true, allcolors=ugentblue]{hyperref}
|
|
|
|
% Whitespace between paragraphs and no indentation
|
|
%--------------------------------------------------
|
|
\usepackage[parfill]{parskip}
|
|
|
|
% Input for title page
|
|
%----------------------
|
|
|
|
% The title
|
|
\thetitle{Forecasting and generative modeling of the Belgian electricity market}
|
|
\thesubtitle{November Intermediate Report}
|
|
|
|
%% Note: a stricter UGent style could be achieved with, e.g.:
|
|
\usepackage{ulem} % for colored underline
|
|
\renewcommand{\ULthickness}{2pt} % adjust thickness of underline
|
|
\thetitle{Forecasting and generative modeling of the Belgian electricity market}
|
|
% Note: do not forget to reset the \ULthickness to 1pt after invoking \maketitle
|
|
% (otherwise all underlines in the rest of your document will be too thick):
|
|
%\renewcommand{\ULthickness}{1pt}
|
|
|
|
% The first (top) infobox at bottom of titlepage
|
|
\infoboxa{\bfseries\large Master Thesis}
|
|
|
|
% The second infobox at bottom of titlepage
|
|
\infoboxb{Name:
|
|
\begin{tabular}[t]{l}
|
|
Victor Mylle
|
|
\end{tabular}
|
|
}
|
|
|
|
% The third infobox at bottom of titlepage
|
|
\infoboxc{
|
|
Promotors:
|
|
\begin{tabular}[t]{l}
|
|
prof. dr. ir. Chris Develder \\
|
|
prof. Bert Claessens
|
|
\end{tabular}
|
|
\\\\
|
|
Supervisor:
|
|
\begin{tabular}[t]{l}
|
|
Jonas Van Gompel
|
|
\end{tabular}
|
|
}
|
|
|
|
% The last (bottom) infobox at bottom of titlepage
|
|
\infoboxd{Academic year: 2023--2024} % note dash, not hyphen
|
|
|
|
|
|
\begin{document}
|
|
|
|
% =====================================================================
|
|
% Cover
|
|
% =====================================================================
|
|
|
|
% ------------ TITLE PAGE ---------
|
|
\maketitle
|
|
\renewcommand{\ULthickness}{1pt}
|
|
|
|
% =====================================================================
|
|
% Front matter
|
|
% =====================================================================
|
|
|
|
% ------------ TABLE OF CONTENTS ---------
|
|
% {\hypersetup{hidelinks}\tableofcontents} % hide link color in toc
|
|
% \newpage
|
|
% \begin{titlepage}
|
|
|
|
% \centering % Centers everything on the page
|
|
|
|
% % Logo or Image (Optional)
|
|
% % \includegraphics[width=0.5\textwidth]{path_to_logo.jpg}
|
|
|
|
% \vspace*{2cm} % Add vertical space
|
|
|
|
% {\large Title: Forecasting and generative modeling of the Belgian electricity market\par}
|
|
|
|
% \vspace{2cm}
|
|
% {\Large Victor Mylle\par}
|
|
|
|
% \vspace{1cm}
|
|
% {\large Period of Internship: 3 July 2023 - 31 August 2023\par}
|
|
|
|
% \vspace{1cm}
|
|
% {\large Mentor: dr. ir. Femke De Backere\par}
|
|
% {\large TechWolf supervisor: ir. Jens-Joris Decorte}
|
|
|
|
% \end{titlepage}
|
|
|
|
\newpage
|
|
|
|
\section{Intermediate Results}
|
|
The electricity market is a complicated system with many different factors. During this thesis, we will try to model the day-ahead system imbalance. Using this imbalance, a model can be trained using reinforcement learning to trade on the electricity market to generate profit. The first step is to model the imbalance. The imbalance is the difference between the amount of electricity that is bought and sold on the day-ahead market and the amount of electricity that is actually consumed. Elia (Transmission System Operator) is responsible for keeping the grid stable and takes the steps necessary to do so. They provide electricity when there is a shortage and take electricity when there is a surplus. The amount of energy that is provided or consumed is called the Net Regulation Volume. Based on the Net Regulation Volume and the bid ladder, the electricity price can be calculated. \\\\
|
|
Elia publishes a lot of data on their website. This data can then be used as training data. First, simple baselines can be implemented to forecast the NRV of the next day.
|
|
\\\\
|
|
The data available ranges from 01-01-2015 until the current date. The data is available in minute or quarter intervals. For our use case, the quarter interval will do. The data is split into training data and test data. The data from 2023 is used as the test set.
|
|
|
|
\subsection{Previous day as forecast}
|
|
One baseline can be to use the previous day NRV values as the forecast for the next day. This gives the following results on the test set: \\\\
|
|
MAE: 145.97317296006946 \\
|
|
MSE: 39632.622958020256
|
|
\subsection{All Zeros}
|
|
Using all zeros as forecast gives the following results on the test set: \\\\
|
|
MAE: 106.1727146629051 \\
|
|
MSE: 21977.654647179577 \\
|
|
\\
|
|
The first small conclusion that can be made is that just using all zeros as the forecast gives better results than using the previous day NRV values.
|
|
|
|
|
|
\subsection{Linear Model}
|
|
A simple linear model can also be trained on the data. This doesn't generatively model the NRV but forecasts the next value based on the given NRV values. This model can then be used autoregressively to forecast the NRV of the next day.
|
|
% Table with results with different ranges of training data
|
|
\begin{table}[h]
|
|
\centering
|
|
\begin{tabular}{|l|l|l|}
|
|
\hline
|
|
Training data range & MAE & MSE \\ \hline
|
|
2015-2022 & 78.04712677001953& 10891.9501953125 \\ \hline
|
|
2016-2022 & 77.98072814941406 & 10872.8173828125 \\ \hline
|
|
2017-2022 & 77.94755554199219 & 10859.1943359375 \\ \hline
|
|
2018-2022 & 77.90494537353516 & 10840.017578125 \\ \hline
|
|
2019-2022 & 77.88092041015625 & 10830.2880859375 \\ \hline
|
|
2020-2022 & 77.84571075439453 & 10823.6826171875 \\ \hline
|
|
2021-2022 & 77.86540985107422 & 10831.35546875 \\ \hline
|
|
2022-2022 & 77.95752716064453 & 10871.7685546875 \\ \hline
|
|
|
|
\end{tabular}
|
|
\caption{Results of the linear model with different ranges of training data}
|
|
\label{tab:linear_model}
|
|
\end{table}
|
|
|
|
The experiments performed use a linear model. The input size is 96 (the quarter-hour values of the NRV) and the output is one value that represents the predicted next NRV value. The experiments use Adam as an optimizer with a learning rate of 0.0003. All input values were rescaled using the MinMaxScaler.
|
|
|
|
|
|
\newpage
|
|
\section{Schedule next months}
|
|
An overview of the planning for the next months is given below. The planning is subject to change depending on the results of the experiments.
|
|
|
|
\subsection{Other input features}
|
|
For the moment, only the NRV is used as input. More inputs can be used to model the NRV. For example, Elia provides a load forecast for every quarter hour. This can also be used as input for the model. Weather and other dependencies should be further explored.
|
|
|
|
\subsection{More complex models}
|
|
For now, the models were kept simple. More complex models can however be used to generatively model the NRV. For example, diffusion models can be explored.
|
|
|
|
\subsection{Reinforcement learning}
|
|
Once a model is trained to generatively model the NRV, a reinforcement learning model can be trained to make better decisions on the electricity market. This step however, requires a good generative model of the NRV.
|
|
|
|
|
|
|
|
\end{document} |