No Thumbnail Available
Data-driven modeling to predict the consumption of energy in one or multiple sites
Files
Dacier_36231400_2019.pdf
Open access - Adobe PDF
- 16.01 MB
Details
- Supervisors
- Faculty
- Degree label
- Abstract
- This thesis deals with the extension of a modeling method in the field of time-series estimation, namely the Hannan-Rissanen(HR) method, and its application to a real life case. The method is used to construct models which are able to describe processes with non-seasonal, single seasonal or double seasonal behaviour. The purpose of modeling is to predict the output of these processes, which in this thesis represent energy consumption in one or multiple sites. The HR method has been proposed before for the estimation of non-seasonal and single seasonal models. In this work, a modification of the method is proposed for the estimation of single seasonal models and the method is extended to cover the modeling case of double seasonal processes. Both at a simulation level as well as at a real process modeling case, results show that time-series estimates can be built in an efficient and systematic way using this method, in order to predict the output of processes with double seasonal characteristics. Last but not least, a way to automatize the modeling procedure is proposed in terms of this thesis.