Abstract: This paper proposed a fault identification method of platform door based on Bayesian neural network model to address the problems faced by the platform door system of urban rail transit, such as low maintenance efficiency, missing operation data, and heavy reliance on the technical level of operation and maintenance personnel. The paper collected and preprocesses multidimensional data, and used Adam optimizer to train and optimize the model. Based on real-time monitoring of platform door operation status, it implemented key component fault identification and intelligent operation and maintenance. Through detailed testing and verification, the results show that the model has superior performance in platform door fault mode recognition, can reduce the average maintenance time of the platform door system, improve its availability and reliability, and provide strong support for the safe operation of urban rail transit.