Dynamic load scheduling and power allocation for energy efficiency and cost reduction in smart grids: An RL-SAL-BWO approach

The demand-side management (DSM) research field has expanded due to rising energy consumption. In the traditional electrical grid, unknown energy usage results in high costs. This paper introduces a reinforcement learning-based self-adaptive learning-black widow optimization (RL-SAL-BWO) approach for dynamic load scheduling and power allocation, aimed at improving energy efficiency and reducing costs and energy consumption. The proposed strategy utilizes pricing signals and real-time load profiles to estimate the changing energy consumption within residential buildings. To optimize energy allocation across different appliances, this algorithm considers both energy efficiency and load characteristics. The RL agent, comprising action space, reward function, and Q-value function, is utilized for decision-making on power allocation and load scheduling. The SAL algorithm automatically adjusts the exploration rate and learning rate which leads to enhanced efficiency. By exploring the solution space, the BWO improves the learning process. Through the integration of RL, SAL, and BWO techniques, energy efficiency is increased, energy consumption is reduced, and electricity costs are lowered. The smart grid is utilized for estimating changes in energy consumption. The purpose of this is to estimate changes in energy consumption, aiding in informed decisions about energy management and infrastructure planning. The proposed approach is implemented using MATLAB R2021b software, followed by the evaluation and calculation of performance metrics. The findings demonstrate that the proposed strategy significantly enhances energy efficiency by 18.5%, reduces energy consumption by 31.91%, and decreases electricity costs by 40.66%. Furthermore, the computation time reduction of the proposed approach is 13.7 s.