A Performance-Driven Economic Analysis of a LSTM Neural Network Used for Predicting Building Energy Consumption

Abstract The energy sector occupies a central role in European policy regarding the sustainability and climate change ambitions. This economic gained even more importance context of growing demand threat to security stability Eastern Europe. administrative changes from past years shift towards renewable sources created momentum for general transformation sector. research this field is much needed order find best solutions predict how efficiently satisfy it. purpose paper assess impact using artificial neural networks on dataset that captures information one building about consumption. network follows Long – Short Term Memory architecture it characterized by Root Mean Squared Propagation function optimization. analysis consists comparison performance three activation functions: Rectified Linear Unit, Sigmoid Tanh. results complement existing focusing prediction consumption at level. case study indicates Unit Tanh functions are appropriate than be used an LSTM applied data. former performs better terms accuracy, measured Absolute Value, has similar computational costs Tanh, with slightly larger value training time.