Zhu, Yuening (Beijing Jiaotong University), ShangGuan, Wei (Beijing Jiaotong University), Chai, Linguo (Beijing Jiaotong University), Chen, Junjie (Beijing Jiaotong University), Zhu, Siyu (Beihang University), Wang, Weijie (Beijing Jiaotong University), Song, Hongyu (Beijing Jiaotong University)

Risk Prediction Method for Urban Rail Transit Emergency Events Based on Explanatory Ensemble Learning Model

Scheduled for presentation during the Invited Session "S16b-Control, Communication and Emerging Technologies in Smart Rail Systems" (TH-EA-T16), Thursday, November 20, 2025, 14:10−14:30, Southport 1

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

To enhance the risk prediction capability for urban rail transit emergencies, this paper utilizes real operational emergency data from Beijing Metro in recent years to investigate in-time safety management for platoon operation, proposing a risk prediction method integrating Random Forests and SHapley Additive exPlanation analysis. Firstly, historical incident data are used to identify risk causation, analyze the diversity and scope of risk causation factors under platoon operation mode, and classify risk severity levels according to their duration. Secondly, a Random Forest ensemble model is established, addressing class imbalance and optimizing the number of decision trees and minimum leaf nodes. Furthermore, it achieves 91.49% accuracy in prediction and classification of risk severity levels, significantly improving the performance compared to traditional ensemble learning methods. Finally, SHapley Additive exPlanation analysis is introduced to quantify feature contribution by calculating their marginal contribution values, generating a feature importance ranking to guide proactive risk prevention and control strategies.