Electricity Consumption Time Series Profiling: A Data Mining Application in Energy Industry

摘要

The ongoing deployment of Automated Meter Reading systems (AMR) in the European electricity industry has created new challenges for electricity utilities in terms of how to fully utilise the wealth of timely measured AMR data, not only to enhance day-to-day operations, but also to facilitate demand response programs. In this study we investigate a visual data mining approach for decision-making support with respect to pricing differentiation or designing demand response tariffs. We cluster the customers in our sample according to the customers' actual consumption behaviour in 2009, and profile their electricity consumption with a focus on the comparison of two sets of seasonal and time based variables. The results suggest that such an analytical approach can visualise deviations and granular information in consumption patterns, allowing the electricity companies to gain better knowledge about the customers' electricity usage. The investigated electricity consumption time series profiling approach will add empirical understanding of the problem domain to the related research community and to the future practice of the energy industry.