An recommender system for energy management in smart homes based on user behavior data

Nowdays, smart homes are equipped with a variety of sensors which are able to monitor human activities. In these houses, the automation system can remotely control activities such as heat, light, energy, security, and so on, which use data collected by sensors to analyze behaviors. One of the challenges in smart homes is reduction of energy consumption. In this research, the application range of energy management, electricity, has been specifically addressed considering smart homes. In addition, the main purpose of this research is energy saving based on user behavior data by identifying repetitive consumer behavior patterns. To do this, CRISP-DM methodology have been employed, in which after collecting data and performing preprocess using FP-GROWTH algorithm, repeated patterns of user behavior are obtained. Based on the existing criteria and the FP-GROWTH algorithm results, a community-based behavioral cluster is created. It should be noted that in these patterns, the names of active sensors also has been used in behavioral patterns which is specified according to the calculated average energy consumption of each behavioral pattern, and further analysis of each behavioral pattern. Based on these data, the recommender system suggests how to store energy in the smart home. Finally, we examined the energy consumption of the home in the pre- and post-proposals which show a 32.38% decrease in electricity consumption.

کلیدواژه ها:

energy management ، smart home ، repetitive pattern recognition ، CRISP-DM ، exploring community rules ، FP-GROWTH.