| Title |
Development of an AI-Based Optimal Decision-Making Algorithm for Train Operation Safety with Multi-Accident Defense |
| Authors |
송은주(Eun-Ju Song) ; 김상암(Sang-Ahm Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2025.74.12.2495 |
| Keywords |
Railway Safety; Algorithm; AI; Deep Learning; Accident prevention |
| Abstract |
Due to frequent train derailments and vehicle accidents, the national railway safety level in Korea has been downgraded from Grade 1 to Grade 2. Since the domestic railway safety prediction and evaluation system remains insufficient, there is a pressing need to shift and advance the paradigm of safety management technologies. Therefore, this study aims to develop an algorithm for constructing a prototype platform that enables optimal decision-making and alternative selection for train operation safety based on digital simulation and AI technologies. Using railway accident and failure data (2002?2021, 13,255 cases), risk patterns were analyzed, and risk factors for major accident types were identified and labeled. The proposed system consists of an Accident Type Classification Model, a Risk Factor Analysis Model, and a Countermeasure Recommendation Algorithm, all of which were validated for performance. |