Smart Meters and Customer Consumption Behavior: An Exploratory Analysis Approach

As the economy and technology continue to advance, the need of energy for humans' activities is growing, placing significant pressure on power distribution to reach this demand instantly. Household energy behaviors can be tracked by using Smart Meters (SM), whose data undoubtedly contains valuable insights into household electricity consumption. However, it is challenging to effectively perceive customers' behavior from the massive SM data. Moreover, this information needs to be captured by a data model; the workflow to understand customer behavior needs to be clearly defined. Our research main goal is threefold: we aim to exploit SMs data to train unsupervised Machine Learning (ML) models to forecast the energy load for a specific customer; we want to cluster customers into appropriate equivalence classes characterized by a distinct consumption pattern; and, last but not least, we pursue the profiling of customers according to their habits, with the goal of discriminating the appliances actually in use and/or the charging of electric vehicles. Since this is currently work-in-progress, in this manuscript we briefly describe our research and report the current preliminary achievements.