Association rule mining based approach to consider users' preferences in the energy management of smart homes

In this paper, a novel data-driven framework for smart home energy management is proposed that integrates electricity demand forecasting and user behavior recognition. The framework aims to maximize smart home profit and consider user comfort. To confront the global energy crisis, we need efficient energy management solutions. Smart homes that benefit from smart appliances and energy management systems have great potential for energy savings. Accurately predicting electricity demand and scheduling appliances while considering user preferences are crucial challenges. In this regard, a data-driven approach based on Long-Short-Term-Memory (LSTM) and association rule mining is utilized to predict "day-ahead" electricity consumption in the smart home and understand user habits and routines. The association rule mining extracts rules that contain the day and order of using flexible appliances. An optimization model is used to maximize smart home profit and schedule appliances based on the users' energy consumption patterns and routines. This framework allows the smart home to optimally schedule flexible appliances based on user preferences and predicted electrical energy demand. A simulation analysis shows the applicability of the proposed approach in scheduling the appliances based on the imposed patterns and routines.