A Recommender System for Increasing Energy Efficiency of Solar-Powered Smart Homes

Area of Science:

• Renewable Energy Systems
• Artificial Intelligence
• Smart Home Technology

Background:

• Photovoltaic (PV) energy optimization is crucial for environmental benefit.
• Battery storage for PV systems has a significant ecological footprint.
• Smart homes require efficient energy management solutions for battery-free solar panels.

Purpose of the Study:

• To develop and evaluate an AI-powered recommender system for optimizing energy production in smart homes with battery-free solar panels.
• To predict daily energy production and consumption using multiple data sources.
• To generate and rank relevant energy usage recommendations for residents.

Main Methods:

• An AI recommender system was developed, utilizing home automation sensor logs, solar inverter data, and weather data.
• Two system variants were trained using 76 days of data: one considering energy consumption, the other not.
• Recommendations were evaluated for relevance, accuracy, and ranking logic by 11 human participants over 14 days.

Main Results:

• Predicting resident energy consumption solely from sensor logs proved challenging.
• An average of 74% of recommendations were deemed relevant by participants.
• The accuracy of specific energy values (kW) in recommendations was 66% for variant 1 and 77% for variant 2.
• Recommendation ranking was considered logical in over 88% of cases.

Conclusions:

• The AI recommender system shows potential for optimizing energy production in solar-powered smart homes.
• Residents may be receptive to using such systems for improved energy management.
• Further research is needed to enhance the accuracy of the specific energy values within the system's recommendations.