Daily allocation of energy consumption forecasting of a power distribution company using optimized least squares support vector machine

Accurate energy consumption forecasting is critical for efficient power distribution management. This study presents a novel approach for optimal allocation forecasting of energy consumption in a power distribution company, utilizing the Least Squares Support Vector Machine (LSSVM) optimized by novel variants of the Barnacle Mating Optimizer (BMO) such as the new Gooseneck Barnacle Optimizer and Selective Opposition-based constrained BMO. The optimized LSSVM hyper-parameters, specifically the regularization parameter ($\gamma$) and the kernel parameter (${\sigma }^2$), were applied to test data to enhance accuracy guided by the Mean Absolute Prediction Error (MAPE), ensuring precise alignment of forecasted values with actual energy consumption data. The results indicate that the novel gooseneck barnacle base-optimized LSSVM provides a robust and reliable solution with accuracy 99.98% for daily energy consumption for allocation forecasting, making it a valuable tool for power distribution companies aiming to optimize their resource allocation and planning processes.