Authors:
Stefan Übermasser
1
;
Simon Kloibhofer
1
;
Philipp Weihs
2
and
Matthias Stifter
1
Affiliations:
1
AIT Austrian Institute of Technology, Austria
;
2
University of Natural Resources and Life Sciences, Austria
Keyword(s):
Distributed Data, Machine Learning, Photovoltaic Systems, Recurrent Neural Network, Power Forecast, Short-Term Forecast, Renewable Energy, Distributed Energy Resources.
Related
Ontology
Subjects/Areas/Topics:
Energy and Economy
;
Energy Management Systems (EMS)
;
Energy Monitoring
;
Energy Profiling and Measurement
;
Energy-Aware Systems and Technologies
;
Greener Systems Planning and Design
;
Load Balancing in Smart Grids
;
Optimization Techniques for Efficient Energy Consumption
;
Renewable Energy Resources
;
Smart Grids
Abstract:
The mass-introduction of small scale power generation units like photovoltaic systems at household levels increase the risk for system unbalances, due to their stochastic generation profile. Additionally, upcoming technologies such as electric vehicles, battery storage systems and energy management systems lead to a change from consumer households to prosumers with a significant different residual load profile. For optimizing the profile of future prosumers, especially the forecast for PV generation is crucial. Whilst traditional weather forecasts are based on a few hundred metering locations in the case of Austria, more than 55000 PV systems are currently connected to the Austrian Power grid. Due to the low areal coverage of common metering locations, weather forecasts do not take local phenomena like shadows from clouds into account. An approach using generation data from neighbouring PV systems together with machine learning methods provides a promising alternative for individual
location based intra-hour forecasts. This paper describes the requirements and methods of such a concept and concludes with a first proof of concept.
(More)