The effectiveness of traditional risk assessment methods greatly relies on the quantity and quality of expert knowledge, featuring strong human factor dependence. The failure modes of railway operation scenes are complex and volatile, causing a significantly different distribution of expert scoring data for decision-making. Focusing on railway signaling safety, this paper puts forward an ensemble learning algorithm (ELA), constructing a comprehensive risk assessment approach that can better use limited expert knowledge to achieve a high fuzziness level. This approach improves the fuzzy risk assessment based on the failure mode, effects, and criticality analysis framework. A risk evaluation algorithm is proposed to construct an ensemble model with different voting rules combined with the k-nearest neighbor, support vector machine, random forest, and other elementary classifiers. Further, the ELA is applied to a case study on a typical signaling apparatus. The results show that using this ensemble model for the risk evaluation can achieve 96.0% testing accuracy. The novel approach simplifies the on-site investigation while weakening the subjectivity of expert decision-making, presenting more advantages in high-speed railway signaling susceptible to intricate factors.
Application of Ensemble Learning and Expert Decision in Fuzzy Risk Assessment of Railway Signaling Safety
2022-10-08
2821688 byte
Conference paper
Electronic Resource
English
Design, development, application, safety assessment and simulation of the railway signaling system
British Library Conference Proceedings | 2010
|Studies on safety assessment of signaling system for high-speed railway
IuD Bahn | 2009
|Fuzzy-reasoning-based approach to qualitative railway risk assessment
British Library Online Contents | 2006
|Fuzzy-Reasoning-Based Approach to Qualitative Railway Risk Assessment
SAGE Publications | 2006
|