A Multi-Objective Approach for Optimal Energy Management in Smart Home Using the Reinforcement Learning

Area of Science:

• Smart Home Technology
• Artificial Intelligence
• Energy Management

Background:

• Smart homes require efficient energy management through sophisticated real-time algorithms.
• Appliance scheduling is crucial for reducing energy consumption but depends on user behavior.
• Modeling human-appliance interaction is essential for effective scheduling algorithms.

Purpose of the Study:

• To propose a novel scheduling algorithm for smart homes based on human-appliance interaction.
• To utilize reinforcement learning (RL) for optimizing appliance operation schedules.
• To reduce energy consumption and minimize user discomfort in smart homes.

Main Methods:

• Developed a reinforcement learning (RL) based scheduling algorithm.
• Divided the day into states, with agents performing actions for maximum reward.
• Categorized appliances into adoptable, un-adoptable, and manageable groups to address user discomfort.

Main Results:

• The proposed RL algorithm demonstrated superior performance compared to the Least Slack Time (LST) algorithm.
• Achieved significant reductions in energy consumption.
• Successfully minimized the discomfort level experienced by home users.

Conclusions:

• Reinforcement learning provides an effective framework for smart home energy management.
• Modeling human behavior is key to developing user-centric and efficient appliance scheduling systems.
• The proposed algorithm offers a promising solution for optimizing energy usage and user satisfaction in smart homes.