中文摘要：

为科学地评估地铁自动扶梯通道安全现状，根据地铁自动扶梯的独特属性，从“设备、乘客、使用环境、安全管理”4个方面建立安全综合评价指标体系。针对一般安全评价方法单一性及易受人为主观影响的问题，提出采取模糊综合评价和概率神经网络（PNN）进行安全状态的初级评价，得到隶属度函数和基本概率分配函数，并通过转换函数转化为安全状态等级的信度函数。由Dempster—Sharer（D—S）证据理论方法进行信度融合，最终获得不同方法综合后的评价结果。研究结果表明，该方法能有效地融合不同途径获得的评价信息，并提高了评价精度，可以作为地铁安全管理者宏观决策的工具。

英文摘要：

In order to scientifically evaluate the safety status of the metro escalator passageway and integrate the u-nique properties of the subway escalator, the paper proposed to establish a comprehensive evaluation index systemfrom 4 aspectsof ＂equipment, passengers, the use of the environment, and safety management＂. Fo, r thegenralsafety evaluation method and the problem of the subjective influence, fuzzy comprehensive evaluation and probabi-listic neural network （PNN） were proposed to evaluate the security status of the primary evaluation. The member-ship function and the basic probability distribution function were obtained, and the reliability function of the secur-ity state level was obtained by transforming. The reliability fusion was then carried out by the Dempster - Shafer （D-S） evidence theory and the different method synthesis was obtained after the appraisal result. The results showthat the method can effectively integrate the evaluation information obtained by different methods, and improve theaccuracy of the evaluation. The method can be used as a macro decision tool for the safety management of the metro.