Förbättring av temperaturprognoser i online-miljö för
styrning av fjärrvärmeproduktion
Centre for Mathematical Sciences
Lund Institute of Technology,
In district heating systems the producer needs to be able to predict the
energy consumption. This is part of the production control, in order to meet
the customers demand for heat and hot water in an economically efficient
way. An important factor for the energy consumption is the ambient temperature.
Prediction of the ambient temperature is therefore an important part of
algorithms for prediction of energy consumption. For prediction of the
temperature, weather forecasts from a meteorological institute, in this case
SMHI, are being used. Such forecasts are however not adapted to the location
where the district heating plant and its customers are situated, but they
are made for a larger area. The aim of this thesis is to improve the weather
forecasts by combining forecasts from SMHI with local measurements of the
temperature. The developed method operates on line, that is, the system uses
new forecasts and measurements as soon as they become available.
Three kinds of forecasts are being used. The first kind is an adjustment
of the SMHI forecast. The second kind is based on a statistical model of
the local ambient temperature. The third kind is a combination of the first
two forecast kinds.