ECAST: A Benchmark Framework for Renewable Energy Forecasting Systems

The increasing capacities of renewable energy sources and the opportunities emerging from the smart grid technology lead to new challenges for energy forecasters. Energy output fluctuates stronger compared to conventional power produc- tion. More time series data is available through the usage of sensor technology. New supply forecasting approaches are developed to better address those characteristics, but meaningful benchmarks of such solutions are rare. Conduct- ing detailed evaluations is time-intensive and unattractive to customers as this is mostly handwork. We define and discuss requirements for ecient and reliable benchmarks of renewable energy supply forecasting tools. To cope with those requirements, we introduce the automated benchmark framework ECAST as our proposed solution. The system's capability is demonstrated on a real-world scenario compar- ing the performance of di↵erent prediction tools against a naive method.