ZHANG Guorui1,2, SI Yongbo1,2, CHEN Guangwu1,2, WEI Zongshou1,2
（1. Automatic Control Research Institute, Lanzhou Jiaotong University, Lanzhou 730070, China；2. Gansu Provincial Key Laboratory of Traffic Information Engineering and Control, Lanzhou 730070, China）
Abstract： Turnout is one of the important signal infrastructure equipment, which will directly affect the safety and efficiency of driving. Base on analysis of the power curve of the turnout, we extract and select the time domain and Haar wavelet transform characteristics of the curve firstly. Then the correlation between the degradation state and the fault state is established by using the clustering algorithm and the Pearson correlation coefficient. Finally, the convolutional neural network (CNN) and the gated recurrent unit (GRU) are used to establish the state prediction model of the turnout to realize the failure prediction. The CNN can directly extract features from the original data of the turnout and reduce the dimension, which simplifies the prediction process. Due to its unique gate structure and time series processing features, GRU has certain advantages over the traditional forecasting methods in terms of prediction accuracy and time. The experimental results show that the accuracy of prediction can reach 94.2% when the feature matrix adopts 40-dimensional input and iterates 50 times.
Key words： turnout; clustering; convolutinal neural network (CNN); gated recurrent unit (GRU); fault prediction
References
［1］Kai Z. The railway switch fault diagnosis algorithm based on BP neural network. In: Proceedings of IEEE International Conference on Control Science & Systems Engineering. 2015.
［2］Zhong Z W, Tang T, Wang F. Research on fault feature extraction and diagnosis of railway switches based on PLSA and SVM. Journal of the China Railway Society, 2018, 40(7): 80-87.
［3］Xu Q Y, Liu Z T, Zhao H B. Method of switch fault diagnosis based on hidden Markov model. Journal of the China Railway Society, 2018, 40(8)： 98-106.
［4］Zhong Z W. Research on methods of health condition assessment and prediction of railway switch. Beijing: Beijing Jiaotong University, 2019.
［5］Serdar D, Sakdirat K, An M, et al. Bayesian network-based probability analysis of train derailments caused by various extreme weather patterns on railway turnouts. Safety Science, 2018, 110: 20-30.
［6］Zhao L H, Lu Q. Method of switch fault diagnosis based on grey correlation Analysis. Journal of the China Railway Society, 2014, 36(2)： 69-74.
［7］Zhang D W. Research on railway signal expert system. Beijing: China Academy of Railway Sciences, 2014.
［8］Eker O F, Camci F, Guclu A, et al. A simple state-based prognostic model for railway switch systems. IEEE Transactions on Industrial Electronics, 2012, 58(5): 1718-1726.
［9］Melnyk P, You Z Q, Li K Q. A high-performance CNN method for offline handwritten Chinese character recognition and visualization. Soft Computing, 2019: 1-9. arXiv:1812.11489V
［10］Zhao B, Wang Z P, Ji W J. CNN-GRU short-term power load forecasting method based on attention mechanism. Power System Technology, 2019, 43(12)： 4370-4376.
［11］Li Z Q, Du J Q, Nie B. Summary of feature select-ion methods. Computer Engineering and Applications, 2019, 55(24)： 10-19.
［12］Wang Q Y, Cao S R, Chen B Y. Research on fault diagnosis of high-voltage circuit breakers based on spectral kurtosis analysis and particle swarm K-Means algorithm. High Voltage Electrical Apparatus, 2019, 55(5)： 23-28.
基于门控循环单元模型的道岔故障预测方法
张国瑞1,2, 司涌波1,2, 陈光武1,2, 魏宗寿1,2
（1. 兰州交通大学 自动控制研究所， 甘肃 兰州 730070；2.甘肃省高原交通信息及控制重点实验室， 甘肃 兰州 730070）
摘要:道岔是铁路上重要的信号基础设备之一， 其故障将直接影响行车安全和效率。 本文从分析道岔的功率曲线入手， 首先提取其时域及哈尔（Haar）小波变换特征并进行特征选择， 然后通过聚类算法和皮尔逊（Pearson）相关系数建立退化状态与故障状态之间的关联， 最后利用卷积神经网络（Convolutional neural network, CNN）和门控循环单元（Gated recurrent unit, GRU）建立道岔的状态预测模型， 实现道岔的故障预测。 CNN可以直接提取原始功率数据的特征同时降低维数， 简化了预测过程。 GRU独特的门结构和处理时间序列的特点在预测精度和时间上相比传统的预测方法具有一定优势。 实验结果表明， 当特征矩阵采用40维输入， 迭代50次时， 道岔状态预测准确率达94.2%。
关键词:道岔； 聚类； 卷积神经网络； 门控循环单元； 故障预测
引用格式：ZHANG Guorui, SI Yongbo, CHEN Guangwu, et al. Turnout fault prediction method based on gated recurrent units model. Journal of Measurement Science and Instrumentation, 2021, 12（3）： 304-313. DOI： 10．3969／j．issn．1674-8042．2021．03．008
[full text view]
