Abstract: Targeting the turnout loss of indication fault disposal operation of urban rail transit station attendant, the THERP (technology for human error rate prediction) and the HCR (human cognitive reliability) models are constructed. The operation process of turnout loss of indication is refined, and a quantitative analysis of human error under rail transit standardized operation is carried out. At the same time, based on Bayesian estimation model, the two shape parameters of prior distribution function are estimated by means of operation simulation test of station attendant, which verifies the rationality of the THERP+HCR model.