Short-term energy forecasting for public educational institution

Forecasting energy consumption has long been a question of great interest in a wide range of fields. The last years, energy usage of public institution is attracting considerable attention as result of energy crises we faced in North Macedonia. The consumption of public building is highly correlated with the working habits of employees or students. In this research work, we tend to forecast the energy consumption of public education institution Faculty of Electrical Engineering and Information Technologies in Skopje. For this purpose, CNN-LSTM model with attention is proposed that performs short-time forecasting. Model is trained and tested on multivariate time series that present energy consumption of the building of the Faculty of Electrical Engineering and Information Technologies. The model is evaluated using three metrics MAE, MSE and MAPE. The obtained results show that the model achieves great performance in multi-step forecasting on hourly basis.