【作者】 李建伟； 程晓卿； 秦勇； 张媛； 邢宗义；

【Author】 LI Jianwei~1,CHENG Xiaoqing~2,QIN Yong~2,ZHANG Yuan~2,XING Zongyi~3 (1.School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China; 2.State Key Laboratory of Rail Traffic Control and Safety,Beijing Jiaotong University,Beijing 100044,China; 3.School of Automation,Nanjing University of Science and Technology,Nanjing 210094,China)

【机构】 南京理工大学机械工程学院； 北京交通大学 轨道交通控制与安全国家重点实验室； 南京理工大学自动化学院；

【摘要】 针对传统可靠性预测方法对非线性故障数据预测效果较差的特点,采用BP神经网络对城市轨道交通车辆的可靠性进行预测。首先,介绍可靠性常用指标和可靠性预测模型:然后,建立BP神经网络三层模型,输入层到中间层采用S型正切函数,中间层到输出层采用线性函数,并采用基于梯度下降法与高斯牛顿法结合的反传算法作为学习函数:最后,利用广州地铁故障数据进行仿真分析。研究结果表明:预测效果较佳,相关性为0.900 69。更多还原

【Abstract】 As the traditional reliability prediction methods for nonlinear fault data are poor,BP neural network was applied to predict the reliability of urban rail transit vehicles.Firstly,the common reliability index and the reliability prediction models were introduced.Then a three layers model of BP neural network was built.In the model,the S-shaped tangent function was used in the input layer to the middle layer,while linear function was used in the middle layer to the output layer.Also in the model,Levenberg-Marquadt back-propagation algorithm was used as the learning function. Finally,the Guangzhou subway fault data was used for simulation analysis.The results show that the prediction is accurate and the correlation coefficient can reach 0.900 69.更多还原