Energy management in smart grids using internet of things and price-based demand response with a hybrid EVO-PDACNN approach

Abstract

Network control systems for energy distribution play an essential role when renewable energy sources (RES) expand and the smart grid (SG) infrastructure increases. A new approach to energy management (EM) in SGs combines energy valley optimizer (EVO) with pyramidal dilation attention convolutional neural network (PDACNN) to achieve its objectives. Through EVO-PDACNN, the system performs accurate energy consumption forecasting with PDACNN, while the EVO algorithm supports systematic scheduling capabilities. Due to its use, this method reduces the peak-to-average ratio (PAR) by 22% also the cost of electricity (COE) by 12%. This method performs better than the wind-driven bacterial forging algorithm (WBFA), genetic algorithm (GA), particle swarm optimization (PSO), modified elephant herd optimization algorithm (MEHOA), and ant colony optimization (ACO) because it has a new prediction ability and quick response. EVO-PDACNN establishes better performance through lower root mean square error (RMSE), together with mean squared error (MSE) and mean absolute error (MAE), which indicates enhanced cost efficiency and resource management capabilities for SGs. The method strengthens both energy forecasting and operational scheduling operations while effectively dealing with changes in supply and demand, which helps build resilient power systems.