SPEMS: A sustainable parasitic energy management system for smart homes: Fid-Bau Portal

SPEMS: A sustainable parasitic energy management system for smart homes

Ismail, Sadia / Mujtaba, Hasan / Beg, Mirza Omer

Graphical abstract Display Omitted

Highlights • A new competitive yet co-operative multi-ojbective MAS based SPEMS proposed for HEMS. • It minimizes energy cost, use, PAR and user's discomfort using Host-Parasite Model. • User comfort evaluated quantitatively with unobtrusive adaption to user's choice. • Using aspects like user's occupancy and weather elements to devise scheduling plans. • A Hall of Fame retains all time best schedules with experiments on real time data.

Abstract The amount of electricity consumed grows exponentially with each passing year. Energy management (EM) has the potential to help satisfy electricity demands more efficiently. It implements strategies to reduce energy consumption during peak time or shift its usage to off-peak hours. This research introduces SPEMS (Sustainable Parasitic EM System), a competitive yet cooperative Multi-agent System (MAS) based framework that optimizes energy cost, consumption, Peak-to-Average Ratio (PAR), and user’s discomfort unobtrusively. The framework devises an optimized load scheduling plan for smart appliances using the Host-Parasite Model where the agents with conflicting objectives need to have a symbiotic relationship. Our work follows a quantitative approach towards evaluating user comfort. It takes into account environmental aspects like user presence and weather elements including user temperature, apparent temperature, and visibility. SPEMS adapts itself to user’s behavior patterns without requiring the user to enter their everyday feedback. The concept of sampled Hall of Fame is incorporated to preserve and use the best schedules from previous days ensuring progress towards finding optimal schedules. Experiments have been performed on real-time data collected from a real home for about one and a half years. Results show that our model is computationally inexpensive, sustainable, and minimizes cost, consumption, and PAR with reduced discomfort of the user.