Advanced

Modelling and Forecasting Electricity Load in Secondary Substations

Sjöborg, Emma (2017) FMS820 20171
Mathematical Statistics
Abstract
In the energy sector a transition towards smart grid is now taking place as a step towards a sustainable energy distribution. In addition to many other solutions, this transition will depend upon extended measurements and data management to increase the knowledge about load flows (i.e electricity use) in the network. This thesis will concentrate on data management and statistical analysis of measurements based on a smart grid project in Hyllie Malm¨o’s largest development area with extensive environmental goals. The purpose of this thesis is to develop models for describing and forecasting the load in secondary substations as accurate as possible. In order to fulfill this purpose, measurements have been collected from secondary... (More)
In the energy sector a transition towards smart grid is now taking place as a step towards a sustainable energy distribution. In addition to many other solutions, this transition will depend upon extended measurements and data management to increase the knowledge about load flows (i.e electricity use) in the network. This thesis will concentrate on data management and statistical analysis of measurements based on a smart grid project in Hyllie Malm¨o’s largest development area with extensive environmental goals. The purpose of this thesis is to develop models for describing and forecasting the load in secondary substations as accurate as possible. In order to fulfill this purpose, measurements have been collected from secondary substationsinHyllie. Forcomparison,dataisalsocollectedfromFigeholm,wheremeasuringhas been going on for a longer time. Together with weather data from SMHI, the load data has beenusedtocreateastatisticalmodel-ageneralizedadditivemodel(GAM).GAMisatype ofregressionmodeldescribingtheload(activepower)inthesecondarysubstationbasedon a number of explanatory variables. These parameters are mainly weather variables, such as temperature and wind speed, and calendar variables; as time of day, day of week and time of year. The models also take into account the load and temperature of the days before, by including the lagged values as explanatory parameters. The data from Figeholm has been used for better detection of the annual pattern, since this data covers a whole year. All of the models show a significant relation between the load and the time of the day as well as day of week. For the stations in Figeholm a distinct annual pattern is also visible. This is not as pronounced for Hyllie, due to the shorter measure period. Furthermore, the laggedvaluesalsoseemtohaveinfluenceontheload. Consideringtheweatherdependence, all stations show a significant relation between the load and the temperature. For the Hyllie stations,therealsoexistarelationbetweentheloadandthewindspeedandglobalradiation. For all station in Hyllie, the same model, using the same explanatory variables has been used. This shows good model flexibility, as the load profiles of the Hyllie stations differs a lotbetweenthestations. Themodelshavealsobeentestedforpredictionoftheloadoneday ahead with relatively good results. Lastly, this thesis will discuss the problems with load modelling and prediction, and how it can be improved with more information and longer measure periods. (Less)
Please use this url to cite or link to this publication:
author
Sjöborg, Emma
supervisor
organization
course
FMS820 20171
year
type
H2 - Master's Degree (Two Years)
subject
keywords
LoadModelling, LoadForecasting, GeneralizedAdditiveModels(GAM)Semiparametric Regression, Smart Grid, Load Profiles, Secondary Substation.
language
English
id
8902631
date added to LUP
2017-02-07 09:30:33
date last changed
2017-02-07 09:50:02
@misc{8902631,
  abstract     = {In the energy sector a transition towards smart grid is now taking place as a step towards a sustainable energy distribution. In addition to many other solutions, this transition will depend upon extended measurements and data management to increase the knowledge about load flows (i.e electricity use) in the network. This thesis will concentrate on data management and statistical analysis of measurements based on a smart grid project in Hyllie Malm¨o’s largest development area with extensive environmental goals. The purpose of this thesis is to develop models for describing and forecasting the load in secondary substations as accurate as possible. In order to fulfill this purpose, measurements have been collected from secondary substationsinHyllie. Forcomparison,dataisalsocollectedfromFigeholm,wheremeasuringhas been going on for a longer time. Together with weather data from SMHI, the load data has beenusedtocreateastatisticalmodel-ageneralizedadditivemodel(GAM).GAMisatype ofregressionmodeldescribingtheload(activepower)inthesecondarysubstationbasedon a number of explanatory variables. These parameters are mainly weather variables, such as temperature and wind speed, and calendar variables; as time of day, day of week and time of year. The models also take into account the load and temperature of the days before, by including the lagged values as explanatory parameters. The data from Figeholm has been used for better detection of the annual pattern, since this data covers a whole year. All of the models show a significant relation between the load and the time of the day as well as day of week. For the stations in Figeholm a distinct annual pattern is also visible. This is not as pronounced for Hyllie, due to the shorter measure period. Furthermore, the laggedvaluesalsoseemtohaveinfluenceontheload. Consideringtheweatherdependence, all stations show a significant relation between the load and the temperature. For the Hyllie stations,therealsoexistarelationbetweentheloadandthewindspeedandglobalradiation. For all station in Hyllie, the same model, using the same explanatory variables has been used. This shows good model flexibility, as the load profiles of the Hyllie stations differs a lotbetweenthestations. Themodelshavealsobeentestedforpredictionoftheloadoneday ahead with relatively good results. Lastly, this thesis will discuss the problems with load modelling and prediction, and how it can be improved with more information and longer measure periods.},
  author       = {Sjöborg, Emma},
  keyword      = {LoadModelling,LoadForecasting,GeneralizedAdditiveModels(GAM)Semiparametric Regression,Smart Grid,Load Profiles,Secondary Substation.},
  language     = {eng},
  note         = {Student Paper},
  title        = {Modelling and Forecasting Electricity Load in Secondary Substations},
  year         = {2017},
}