Prediction of Train Arrival Delay Using Hybrid ELM-PSO Approach

In this study, a hybrid method combining extreme learning machine (ELM) and particle swarm optimization (PSO) is proposed to forecast train arrival delays that can be used for later delay management timetable optimization. First, nine characteristics (e.g., buffer time, the number, station code) associated with are chosen analyzed using extra trees classifier. Next, an ELM one hidden layer developed predict by considering these mentioned before as input features. Furthermore, PSO algorithm optimize hyperparameter of compared Bayesian genetic solving arduousness problem manual regulating. Finally, case studied confirm advantage model. Contrasted four baseline models (k-nearest neighbor, categorical boosting, Lasso, gradient boosting decision tree) across different metrics, model demonstrated proficient achieve highest prediction accuracy. addition, through detailed analysis error, it found our possesses good robustness correctness.