Abstract

Urban Rail Transit (URT) systems play a critical role in modern cities but consume significant amounts of energy. Understanding and predicting energy consumption patterns in these systems is vital for sustainable urban planning, especially during disruptive events. This study presents a Long Short-Term Memory (LSTM) recurrent neural network, the model can accurately predict daily energy consumption and average daily temperature, with root mean squared errors (RMSE) of 50.6 MWh and 6.62°F. Additionally, a decision-making tool was developed to simulate various operational strategies and their impacts on energy consumption and temperature. These findings provide URT operators with a robust framework for making data-driven decisions and improving energy efficiency in URT systems.