An essential prerequisite for proactive and anticipatory real-time regulation of railroad traffic is the accurate forecast of train delays, or deviations from the schedule. To manage the viability of timetable realization, traffic controllers must forecast the trains’ arrival times in (or approaching) their region. Rescheduled travel, traffic control, data on passengers, and effective route design all rely upon the ability to predict train locations in both space and time in real time. The study introduces a data mining system designed to forecast train traffic. The system utilizes a dataset that includes the scheduled times of all trains and the relationships between them, along with weather characteristics. In order to improve the system’s accuracy in predicting future traffic, research was proposed that would use Swarm Intelligence-based feature selection techniques as Harris Hawk Optimization (HHO). To ascertain the best predictive models in terms of predictive capability, machine learning techniques namely Bi-directional Long Short-Term Memory (Bi-LSTM) with entity embedding are utilized. It has been demonstrated that the suggested algorithm for processing and analyzing rail traffic is more efficient than alternative approaches and has shown an accuracy rate of 92.3% and precision of 93%.
Swarm Intelligence Based Feature Selection and Machine Learning Methods for Future Railway Traffic Prediction
2024-06-21
491853 byte
Conference paper
Electronic Resource
English
On feature selection for traffic congestion prediction
Online Contents | 2013
|Swarm Intelligence Inspired Adaptive Traffic Control for Traffic Networks
Springer Verlag | 2018
|