Short-term load forecasting ensures the efficient operation of power systems besides affording continuous power supply for energy consumers. Smart meters that are capable of providing detailed information on buildings energy consumption, open several doors of opportunity to short-term load forecasting at the individual building level. In the current paper, four machine learning methods have been employed to forecast the daily peak and hourly energy consumption of domestic buildings. The utilized models depend merely on buildings historical energy consumption and are evaluated on the profiles that were not previously trained on. It is evident that developing data-driven models lacking external information such as weather and building data are of great importance under the situations that the access to such information is limited or the computational procedures are costly. Moreover, the performance evaluation of the models on separated house profiles determines their generalization ability for unseen consumption profiles. The conducted experiments on the smart meter data of several UK houses demonstrated that if the models are fed with sufficient historical data, they can be generalized to a satisfactory level and produce quite accurate results even if they only use past consumption values as the predictor variables. Furthermore, among the four applied models, the ones based on deep learning and ensemble techniques, display better performance in predicting daily peak load consumption than those of others.

This paper addresses the estimation of household communities' overall energy usage and solar energy production, considering different prediction horizons. Forecasting the electricity demand and energy generation of communities can help enrich the information available to energy grid operators to better plan their short-term supply. Moreover, households will increasingly need to know more about their usage and generation patterns to make wiser decisions on their appliance usage and energy-trading programs. The main issues to address here are the volatility of load consumption induced by the consumption behaviour and variability in solar output influenced by solar cells specifications, several meteorological variables, and contextual factors such as time and calendar information. To address these issues, we propose a predicting approach that first considers the highly influential factors and, second, benefits from an ensemble learning method where one Gradient Boosted Regression Tree algorithm is combined with several Sequence-to-Sequence LSTM networks. We conducted experiments on a public dataset provided by the Ausgrid Australian electricity distributor collected over three years. The proposed model's prediction performance was compared to those by contributing learners and by conventional ensembles. The obtained results have demonstrated the potential of the proposed predictor to improve short-term multi-step forecasting by providing more stable forecasts and more accurate estimations under different day types and meteorological conditions.

Digitalization and decentralization of energy supply have introduced several challenges to emerging power grids known as smart grids. One of the significant challenges, on the demand side, is preserving the stability of the power systems due to locally distributed energy sources such as micro-power generation and storage units among energy prosumers at the household and community levels. In this context, energy prosumers are defined as energy consumers who also generate, store and trade energy. Accurate predictions of energy supply and electric demand of prosuemrs can address the stability issues at local levels. This study aims to develop appropriate forecasting frameworks for such environments to preserve power stability. Building on existing work on energy forecasting at low-aggregated levels, it asks: What factors influence most on consumption and generation patterns of residential customers as energy prosumers. It also investigates how the accuracy of forecasting models at the household and community levels can be improved. Based on a review of the literature on energy forecasting and per- forming empirical study on real datasets, the forecasting frameworks were developed focusing on short-term prediction horizons. These frameworks are built upon predictive analytics including data col- lection, data analysis, data preprocessing, and predictive machine learning algorithms based on statistical learning, artificial neural networks and deep learning. Analysis of experimental results demonstrated that load observa- tions from previous hours (lagged loads) along with air temperature and time variables highly affects the households’ consumption and generation behaviour. The results also indicate that the prediction accuracy of adopted machine learning techniques can be improved by feeding them with highly influential variables and appliance-level data as well as by combining multiple learning algorithms ranging from conventional to deep neural networks. Further research is needed to investigate online approaches that could strengthen the effectiveness of forecasting in time-sensitive energy environments.