Nowadays, there is an undoubted change of trend toward a decentralized and decarbonized electric grid, where the electric generation based on local resources will take on special relevance. In this context, the encouragement of collective self-consumption (CSC) becomes one of the key issues. One of the aspects that will contribute to this aim is the development of power consumption-forecasting tools. This article proposes the comparison of three models to perform a day ahead consumption forecasting of ESTIA 2 building: nonlinear autoregressive neural network with exogenous inputs (NARX), long-short-term memory cell (LSTM) and support vector regression (SVR). First, the model structure has been designed by selecting the suitable-input combination and the optimal time window (TW) for the three models. Then, parameters of each model have been adjusted to achieve the most accurate prediction. After forecasting separately winter and summer seasons, experiments reveal that the proposed NARX neural network is the one that predicts with the highest accuracy in both winter and summer months, obtaining a mean absolute percentage error (MAPE) of 14,1% and 12%, respectively. Likewise, regardless of the model, better results have been obtained in summer predictions, which is closely related to the dependence of the building’s consumption on the heating, ventilation, and air conditioning (HVAC) system.

The existing trend towards increased penetration of renewable energies in the traditional grid, and the intermittent nature of the weather conditions on which these energy sources depend, make the development of tools for the forecasting of renewable energy production more necessary than ever. Likewise, the prediction of the energy generated in these renewable production plants is key to the implementation of efficient Energy Management Systems (EMS) in buildings. These will aim both to increase the energy efficiency of the building itself, as well as to encourage self-consumption or, where appropriate, collective self-consumption (CSC). This paper presents a comparison between four different models, the former one being an analytical model and the remaining three machine learning (ML) based models. All of them will forecast the photovoltaic (PV) production curve for the next day. In order to validate these models, a case study of a PV system installed on the roof of a university building located in Bidart (France) is proposed. The model that most accurately forecasts the PV production during the period of July 2021 is the support vector regression (SVR), which has a mean R² of 0.934 for July, being 0.97 on sunny days and 0.85 on cloudy ones. This is an improvement of 5.14%, 4.07%, and 4.18% over the nonlinear autoregressive with exogenous inputs (NARX), feedforward neural network (FFNN), and analytical model, respectively.

The building sector has an important role in the decarbonization of energy. Buildings energy management systems can act on flexible loads to contribute to the integration of renewable energies. This article presents the design, implementation and evaluation of a rule-based energy management systems (RB-EMS) in the frame of a collective self-consumption (CSC) system based on photovoltaic (PV) energy in the Technology Park of Izarbel, in Bidart, France. The RB-EMS acts on the heat ventilation and air conditioning (HVAC) system of one of the buildings involved in the CSC in order to maximize the PV energy self-consumption rate (SCR). Internet of things (IoT) has been developed and implemented in order to retrieve consumption, production and temperature data, and to be able to act on the HVAC. A new 5-step methodology has been developed to design and adjust the RB-EMS. This methodology is mainly based on tests carried out to find the relationship between ON/OFF states of the internal units of the HVAC and the heat pumps’ consumption. Finally, the RB-EMS is tested in heating operating mode. The results show that the RB-EMS allows obtaining a SCR of 89%, and that the users comfort limits are respected.

In self-consumption (SC) configurations, energy management systems (EMSs) are increasingly being implemented to maximise the self-consumption ratio (SCR). Recent studies have demonstrated that prediction-based EMSs significantly enhance decision-making capabilities compared to non-predictive EMSs. This paper presents the design, implementation, and testing on a real system of two machine learning (ML)-type predictive models capable of forecasting the electricity consumption of an individual building using a small dataset. A nonlinear autoregressive with exogenous input (NARX) neural network model and a support vector regression (SVR) model were designed and compared. These models predict day-ahead hourly electricity consumption using forecasted meteorological data from Meteo Galicia (MG) and building occupancy data, both automatically obtained and pre-processed. In order to compensate for the lack of recurrence of the SVR model, the effect of introducing an additional input, a time vector, was analysed. It is proved that both ML models trained with a small dataset are able to predict the next day’s average hourly power with a mean MAPE below 13.96% and a determination coefficient (R²) greater than 0.78. The model that most accurately predicts the hourly average power of a week is the SVR, which achieves a mean MAPE and R² of 10.73% and 0.85, respectively.