A Review of Short-Term Electrical Load Forecasting Using Ensemble Stacking Generalization with Artificial Neural Network

Abstract

Electric load forecasting has gained much attention in electricity
production due to its important role in electric power system
management. Short-term load forecasting (STLF) uses the
perception of ensemble learning approaches as a general scheme for
educating the prognostic skill of a machine learning model (MLM).
STLF is subjected to numerous errors /problems like high bias and
variance. This prompts the need for the employment of ensemble
stacking generalization with artificial neural networks (ANN) to
ensure an improved performance with accurate results. This
approach combined four models namely random forest (RF),
generalized boosted regression model (GBRM), Evolutional
Algorithm (EvA), and artificial neural network (ANN). The inner
mechanism of the stacked EvA-RF-GBRM-ANN model involves
creating meta-data from EvA, RF, and GBRM models to calculate
the final estimates using ANN. This work proposes a stacked neural
network for short-term load forecasting through a view of dropping
predicting faults besides their discrepancy associated with solebased models and stacked neural networks (SNN