Short term load forecasting using optimized deep learning based weighted DenseBiGRU for smart grids

T.M. Angelin Monisha Sharean, R.S. Shaji,
Short term load forecasting using optimized deep learning based weighted DenseBiGRU for smart grids,
Sustainable Computing: Informatics and Systems,
Volume 48,
2025,
101240,
ISSN 2210-5379,
https://doi.org/10.1016/j.suscom.2025.101240.
(https://www.sciencedirect.com/science/article/pii/S2210537925001611)
Abstract: Distribution system operators can successfully manage energy through the use of advanced demand-response programs in the smart grid (SG) due to short-term load forecasting. The short-term load forecasting approach is essential for effective energy management when taking into account the electric fields in the energy trade. Short-term load forecasting can be applied to many aspects of daily operations in infrastructure maintenance, energy purchase, contract analysis, energy generation planning, including load shedding, and electric utilities. There are a number of techniques for predicting short-term load. Still, all suffer from a lack of model parameter adaptability, making it impossible to meet the demand for precise and efficient smart grid load forecasting. In order to improve the model's predictive accuracy, an optimized deep learning (DL) model is employed in this study. The proposed Improved Weighted Mean of Vector based Dense Bidirectional Gated Recurrent Unit (I-INFO_DenseBiGRU) is utilized for the short term load forecasting with the weather data. The proposed I-INFO_denseBiGRU performance is calculated based on numerous events like MAPE, MSE, MAE, NRMSE, and R2, and achieves superior performance compared to state-of-the-art methods.
Keywords: Short term load forecasting; Weighted mean of vector; Hybrid deep learning; Smart grids; Wrapper feature selection; Gated recurrent unit; Weather data