Optimized Home Appliance Scheduling Utilizing Reinforcement Learning for Enhanced Load Profile

With the recent multiple increases in the fuel prices, and as most of electricity generation are fossil-fuelled, utilities were obliged to raise the electricity tariffs accordingly. Subsequently, individuals are compelled to reduce their energy usage to cut the daily energy bills. One approach to address this challenge is to enhance the overall efficiency of electrical systems by shifting peak demand periods, rather than relying on expanding the transmission network to correspond to peak demand indirectly. With the smart grid and deploying bi-directional smart meters, it is probable to perform load control using innovative Home Energy Management (HEM) systems with demand responseenabled appliances. A HEM system utilizes automated control and monitoring capabilities for household appliances, enabling it to make decisions regarding load shedding and shifting according to a predefined set of preferences. This in turn helps alleviate power system stress conditions and shift the peak demand. This study aims to develop two intelligent HEM systems with responseenabled appliances considering Reinforcement Learning (RL) as the main controlling algorithm and Time-of-Use (ToU) electricity price. The objective of the first controller is to minimize the power consumption while the second controller aims to enhance the load profile. The proposed RL-based intelligent HEM systems achieved 8.3% reduction and 1.25% for the minimization and load enhancement cases, respectively, achieving flexible appliance usage schedules that will minimize consumer energy bill without user lifestyle disruptions.