No abstract

No abstract

No abstract

The increased penetration of smart meters generates huge amounts of fine-grained data, which may empower a new generation of energy related applications and services. Significant research efforts focus on the usage of such data to mainly improve the business processes of the electrical grid operators and provide some value added services to the endusers. Forecasting has a prominent position as it is a crucial planning step, and is mostly used to predict the grid load through highly-aggregated data. However, with the dramatic increase on fine-grained data, new challenges arise as forecasting can now also be done on much shorter and detailed time-series data, which might provide new insights for future applications and services. For the smart grid era, being able to segment customers on highly predictable groups or identify highly volatile ones, is a key business advantage as more targeted offers can be made. This work focuses on the analysis and impact assessment of in the context of smart metering data aggregation. A system to measure the impact of aggregation is designed and its performance is assessed. We experiment with measuring of the forecast accuracy on various levels of individual load aggregation, and investigate the identification of highly predictable groups.

No abstract