This paper proposes an electricity demand and price forecast model of the smart city large datasets using a single comprehensive Long Short-Term Memory (LSTM) based on a sequence-to-sequence network. Real electricity market data from the Australian Energy Market Operator (AEMO) is used to validate the effectiveness of the proposed model. Several simulations with different configurations are executed on actual data to produce reliable results. The validation results indicate that the devised model is a better option to forecast the electricity demand and price with an acceptably smaller error. A comparison of the proposed model is also provided with a few existing models, Support Vector Machine (SVM), Regression Tree (RT), and Neural Nonlinear Autoregressive network with Exogenous variables (NARX). Compared to SVM, RT, and NARX, the performance indices, Root Mean Square Error (RMSE) of the proposed forecasting model has been improved by 11.25%, 20%, and 33.5% respectively considering demand, and by 12.8%, 14.5%, and 47% respectively considering the price; similarly, the Mean Absolute Error (MAE) has been improved by 14%, 22.5%, and 32.5% respectively considering demand, and by 8.4%, 21% and 61% respectively considering price. Additionally, the proposed model can produce reliable forecast results without large historical datasets.

The smart city integrating the smart grid as an integral part of it to guarantee the ever-increasing electricity demand. After the recent outbreak of the COVID-19 pandemic, the socioeconomic severances affecting total levels of electricity demand, price, and usage trends. These unanticipated changes introducing new uncertainties in short-term demand forecasting since its result depends on the recent usage as an input variable. Addressing this challenging situation, this paper proposes an electricity demand and price forecast model based on the LSTM Deep Learning method considering the recent demand trends. Real electricity market data from the Australian Energy Market Operator (AEMO) is used to validate the effectiveness of the proposed model and elaborated with two scenarios to get a wider context of the pandemic impact. Exploratory data analyses results show hourly electricity demand and price reductions throughout the pandemic weeks, especially during peak hours of 8 am-12 noon and 6 pm -10 pm. Electricity demand and price has been dropped by 3% and 42% respectively on average. However, overall usage patterns have not changed significantly compared to the same period last year. The predictive accuracy of the proposed model is quite effective with an acceptably smaller error despite trend change phenomena triggered by the pandemic. The model performance is comprehensively compared with a few conventional forecast methods, Support Vector Machine (SVM) and Regression Tree (RT), and as a result, the performance indices RMSE and MAE have been improved using the proposed LSTM model.

Due to extreme weather conditions and anomalous events such as the COVID-19 pandemic, utilities and grid operators worldwide face unprecedented challenges. These unanticipated changes in trends introduce new uncertainties in conventional short-term electricity demand forecasting (EDF) since its result depends on recent usage as an input variable. In order to quantify the uncertainty of EDF effectively, this paper proposes a comprehensive probabilistic EFD method based on Gaussian process regression (GPR) and kernel density estimation (KDE). GPR is a non-parametric method based on Bayesian theory, which can handle the uncertainties in EDF using limited data. Mobility data is incorporated to manage uncertainty and pattern changes and increase forecasting model scalability. This study first performs a correlation study for feature selection that comprises weather, renewable and non-renewable energy, and mobility data. Then, different kernel functions of GPR are compared, and the optimal function is recommended for real applications. Finally, real data are used to validate the effectiveness of the proposed model and are elaborated with three scenarios. Comparison results with other conventional adopted methods show that the proposed method can achieve high forecasting accuracy with a minimum quantity of data while addressing forecasting uncertainty, thus improving decision-making.