| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 734.85 KB | Adobe PDF |
Advisor(s)
Abstract(s)
The increasing penetration of renewable generation
brings a significant escalation of intermittency to the power and
energy system. This variability requires a new degree of flexibility
from the whole system. The active participation of small and
medium players becomes essential in this context. This is only
possible by using adequate forecasting techniques applied both to
the consumption and to generation. However, the large number
of incontrollable factors, such as the presence of consumers in the
building, the luminosity, or external temperature, makes the
forecasting of energy consumption an arduous task. This paper
addresses the electrical energy consumption forecasting problem,
by studying the correlation between the solar radiation and the
electrical consumption of lights. This study is performed by
means of three forecasting methods, namely a multi-layer
perceptron artificial neural network, a support vector regression
method, and a linear regression method. The performed studies
are analyzed using data gathered from a real installation –
campus of the Polytechnic of Porto, in real time.
Description
Keywords
Artificial Neural Network Electricity Consumption Solar Radiation Support Vector Regression
Citation
Publisher
Institute of Electrical and Electronics Engineers