A Data-Driven Calibration Method of Short-Term Passenger Flow Prediction Models for Urban Rail Transit Systems

At present, a variety of models exist for short-term passenger flow prediction, each of which requires parameter calibration in practical applications. However, the parameters for most currently used models are set using empirical values, leading to occasional discrepancies between actual outcomes and model predictions. This paper proposes a data-driven parameter calibration method for short-term passenger flow prediction models of urban rail transit systems with a genetic algorithm. The method’s accuracy and effectiveness are initially verified through a practical case study of the Nanchang Metro. The results indicate that the method can enhance the precision of parameter values and reduce the deviations between actual outcomes and model predictions in urban rail transit systems.