This Research presents a hybrid load forecasting model that combines successfully Long Short Term Memory (LSTM) neural networks with a Quadratic Regression Model (QRM) for better optimization of short term load prediction of the power distribution systems. The LSTM and the QRM model are combined to take advantage of the sequential learning ability of LSTM and the analytical acumen of QRM, thereby increasing accuracy and reliability in dynamic energy demand environments. Practical relevance and reliability in the real-world electrical load data from distribution networks was ensured for model development, training and validation. The model's performance was compared to the traditional forecasting approaches. Experiments suggest a 2.5% decrease in Mean Absolute Error (MAE), a four-point-zero percent reduction in Root Mean Square Error (RMSE), and an R-squared (R2) of 0.85, which means high forecasting capability. The accuracy of the predicting model is 97% better than the exitsing methods. The recommended approach helps to facilitate timely and informed decision making for utility providers with regard to load management, optimization of resources and planning of infrastructures. It also assists in minimizing energy wastages and increasing the power system's stability through the reduction of forecasting errors. Furthermore, the combination of the LSTM and QRM is consistent with the globally oriented strategy of advancement towards digitization and sustainability at the energy system level. The study further provides development of intelligent forecasting techniques and emphasizes the significance of hybrid models in modern grid-based activities. It is a scalable and responsive solution for utilities that are looking to meet the challenges presented by fluctuating demand and progress toward more resilient and environmental-friendly energy infrastructures.
Gengi et al. (Thu,) studied this question.