A human error quantitative analysis of the railway ATO system based upon improved CREAM

Abstract

Automatic Train Operation (ATO) system, whose function is to drive the train automatically, can improve the efficiency of operation. However, the driver’s errors often leads to the failure of the ATO system in actual operation. So it is significant to evaluate driver’s human reliability for improving system effectiveness. Considering the scenario-based characteristics of railway operation, this paper proposes a three-layer Human Reliability Analysis (HRA) model based on Bayesian network, Event Tree Analysis(ETA) to optimize Cognitive Reliability and Error Analysis Method (CREAM). Special Bayesian networks are used to transform expert experience into driver’s Human Error Probability (HEP) value in the ATO system of the railway, rather than the probability interval. The model is used to calculate the HEP of the scene of manual train door and platform door linkage. The overall idea of the model is consistent with the CREAM method, and the application results are also conform to the conclusion of the basic CREAM . Therefore, the improved HRA model can provide reliable HEP of the scenario for the driver in ATO system of the railway.