基于神经网络的高速铁路安全态势预测方法研究

为提高高速铁路运营安全水平，提出用事故次数、事故联动系数、月均事故率作为高速铁路运营安全态势评价指标，对高速铁路的运营安全态势进行时空分布特性聚类分析，将国内高速铁路的线路运营现状分为危险、较危险、一般、安全四个等级。以2013—2015年存在运营安全风险的高速铁路线路运营事故数据为基础，利用BP （Back Propagation简称BP）神经网络具有较强非线性拟合能力的特点，建立高速铁路运营安全态势预测模型。预测结果表明，除京沪线、达成线、福夏线、胶济线外，其他高速铁路线路安全运营水平均有上升趋势。

On the safety operation state prediction of high-speed railway based on neural network

For the safety operation level of high‐speed railways ,the number of accidents ,accident linkage coefficient and average month accident rate are proposed as the evaluation indexes in this paper . With space‐time distribution analysis by clustering theory ,the operation state of high‐speed railway is classified into four levels :dangerous ,comparison of risk ,generation ,and safety .Then ,the high‐speed railway lines which have risk on safety operation are predicted from 2013 to 2015 by Back Propagation Neural Network . The result shows that except the lines of Jin‐Hu ,Da‐Chen ,Fu‐Xia ,Jiao‐Ji ,all other high‐speed railway lines’ safety level have a certain improvement .