Prediction of energy consumption in Mexico by integrating environmental, economic, and energy data using artificial neural network models

Preventing energy consumption in emergencies such as the last COVID-19 pandemic can ensure the continued operation of hospitals and food supply centres. In addition, considering the relationship of energy consumption with various factors generates points of attention. This work is focused on presenting the prediction of energy consumption in Mexico using data related to environmental, economic and energy aspects recorded from 1965 to 2021. The input variables were: year, carbon dioxide emissions, Gross Domestic Product per capita, number of power plants, increase in temperature in the world and oil production. The models of Artificial Neural Networks (ANN’s) based on a single layer and hidden multilayer obtained a good correlation between the real values and the simulated ones with a coefficient of determination (R2) of 0.9999 and Mean Absolute Percentage Error (MAPE) of 0.37%. For prediction, the Recurrent Neural Network-Long Short-Term Memory (RNN-LSTM) model generated a correlation with an R2 of 0.8910 between the real and forecast data. The data demonstrated that ANN-based models are a tool capable of predicting energy consumption to support decision-making on the distribution and consumption of energy resources in the face of future emergencies.