Home Activity Pattern Estimation Using Aggregated Electricity Consumption Data

In this paper, we propose a low-cost, noninvasive home activity recognition method using low-resolution power consumption data. Notably, we tackle the following two challenges. Firstly, we use only the time series of power consumption data aggregated per house and measured approximately every 20 s, which is usually used for demand monitoring by smart meters. We design a set of activities that can be recognized using such low-resolution data and find an appropriate feature set to train and test the balanced random forest classifier. Secondly, we consider the divergence of activity patterns seen in different households. Since supervised learning dedicated to each household is not a realistic solution, we arrange different classifiers trained with different household data in supervised learning and present a method of automatically choosing the best-fit classifier for an unseen household in the online phase. We evaluated our method with an aggregated power consumption dataset collected from eight real homes over 191 days. We confirmed that our method achieved a recognition accuracy of 70% for activities using such a low-resolution aggregated power consumption dataset, and that our proposed fitness scores were effective for choosing the best classifiers.