Evaluation of neural networks for residential load forecasting and the impact of systematic feature identification

Abstract Energy systems face challenges due to climate change, distributed energy resources, and political agenda, especially distribution system operators (DSOs) responsible for ensuring grid stability. Accurate predictions of the electricity load can help DSOs better plan maintain their grids. The study aims test a systematic data identification selection process forecast Danish residential areas. five-ecosystem CSTEP framework maps relevant independent variables on cultural, societal, technological, economic, dimensions. Based literature, recurrent neural network (RNN), long-short-term memory (LSTM), gated unit (GRU), feed-forward (FFN) are evaluated compared. models trained tested using different inputs forecasting horizons assess impact approach practical flexibility models. findings show that achieve equal performances around 0.96 adjusted R 2 score 4–5% absolute percentage error 1-h predictions. Forecasting 24 h gave an 0.91 increased slightly 6–7% error. depended type network, with FFN showing highest increase in when removing supporting variables. GRU LSTM did not rely identified variables, minimal changes performance or without them. researchers understand target variable. results indicate focus curating affects more than choosing specific architecture.