WU Xiaochun， LIU Jiexin
（School of Automation & Electrical Engineering， Lanzhou Jiaotong University， Lanzhou 730070， China）
Abstract： To solve the problems of poor real-time performance and uncertain fault location in track circuit fault detection, a fault detection method for track circuit was proposed based on improved intrinsic time-scale decomposition (IITD) and sample entropy input as eigenvalues for improved deep belief network (DBN) recognition. Firstly, the fault signals are decomposed by IITD method, and the PR component containing the main fault characteristic information is screened. Secondly, the sample entropy is calculated as the characteristic value of the signal. Finally, the characteristic value is input into the improved DBN network, which can effectively detect the faults. According to 180 sets of historical fault data of a electricity service section, simulation results show that the accuracy of the algorithm is 97.22%, which is 7.12%, 4.98%, 6.34% and 3.82% higher than that of BP neural network, PMFCC-DTW, Fuzzy neural network and non-optimized DBN. It provides a new solution for fault detection of track circuit.
Key words： improved intrinsic time-scale decomposition (IITD)； sample entropy； deep belief network (DBN)； fault detection
References
［1］SUN S P. Research on fault diagnosis for railway jointless track circuit. Beijing: Beijing Jiaotong University, 2014： 22-23.
［2］YANG J, WANG X M, ZHANG W F, et al. Transient fault diagnosis of track circuit based on PWFCC-DTW. Journal of the China Railway Society, 2021, 43(5)： 104-111.
［3］HU L Q, HE C F, HE Q C, et al. Track circuit fault prediction method based on grey theory and expert system. Journal of Visual Communication and Image Representation, 2019, 58： 22-23.
［4］CHEN G W, GAO Y L, JIAO X M. Track circuit fault diagnosis based on adaptive mutation SAPSO-LSSVM. Journal of Beijing Jiaotong University, 2021, 45(2)： 1-7.
［5］WANG Z C. Research on Transmission Model and Fault Diagnosis Method of Track Circuit. Chengdu: Southwest Jiaotong University, 2019： 44-46.
［6］ZHENG Z, DAI S, XIE X. Research on fault detection for ZPW-2000A jointless track circuit based on deep belief network optimized by improved particle swarm optimization algorithm. IEEE Access, 2020, 8： 175981-175997.
［7］XIANG L, GUO P F, GAO N, et al. Rolling bearing fault diagnosis based on IITD and FCM clustering, Journal of Aerospace Power, 2018, 33(10)： 2553-2560.
［8］CHEN J, CAI K Q, TAO S Y, et al. Fault diagnosis method of rolling bearing based on IITD fuzzy entropy and random forest. Acta Metrologica Sinica, 2021, 42(6)： 774-779.
［9］HALIT A, MUHAMMAD S. A multivariate streamflow forecasting model by integrating improved complete ensemble empirical mode decomposition with additive noise, sample entropy, Gini index and sequence-to-sequence approaches. Journal of Hydrology, 2021, 603： 63-65.
［10］QU H Q, WEI B B, ZHANG Z, et al. Feature extraction method based on FDM energy entropy and its application on optical fiber vibration recognition. Laser & Optoelectronics Progress, 2021, 58(7)： 168-175.
［11］WEI L, ZHANG Y J. Research on fault diagnosis of rotating machinery based on improved deep belief network. Journal of North China Electric Power University (Natural science edition), 2020, 47(6)： 99-106.
［12］CHEN X, JIANG L, SONG Z H. Nonlinear process fault detection based on gaussian restricted boltzmann machine. Journal of Technology, 2015, 15(2)： 139-143.
［13］HINTON G E. A practical guide to training restricted Boltzmann machines, Momentum, 2010, 9(1)： 926.
［14］REN C H, YU T Z, DING D, et al. Fault diagnosis method of rolling bearing based on VMD-DBN. Journal of Northeastern University(Natural science), 2019, 42(8)： 1105-1110.
［15］RAMANATHAN L, PARTHASRATHY G, VIJAYAKUMAR K, et al. Cluster-based distributed archi -tecture for prediction of student’s performance in higher education. Cluster Computing, 2019, 22(1)： 1329 -1344.
［16］GAN M. Study of sparse representation of signal in fault feature extraction and intelligent diagnosis of rolling bearing. Heifei: University of Science and Technology of China, 2017： 333-345.
［17］CAO X S, Research on Fault diagnosis technology of industrial robot based on deep learning. Handan: Hebei University of Engineering, 2021： 35-39.
［18］LI Y B, HUANG D H, MA J B, et al. A gear fault diagnosis method based on deep belief network and information fusion. Journal of Vibration and Shock, 2021, 40(8)： 62-69.
基于IITD样本熵与改进深度置信网络的轨道电路故障检测
武晓春， 刘杰鑫
（兰州交通大学 自动化与电气工程学院， 甘肃 兰州 730070）
摘要：针对轨道电路故障检测实时性差且故障位置不确定等问题， 提出了一种基于改进固有时间尺度分解（Improved intrinsic time-scale decomposition, IITD）和样本熵值作为特征值输入以改进深度置信网络（Deep belief network, DBN）识别的轨道电路故障检测方法。 首先， 采用IITD方法对故障信号进行分解， 筛选包含主要故障特征信息的PR分量。 其次， 计算其样本熵值作为信号的特征值。 最后， 将特征值输入至改进的DBN网络中， 进行故障检测。 以180组轨道电路历史故障数据为输入， 本算法准确率达97.22%， 较BP神经网络、 PMFCC-DTW、 模糊神经网络以及未经优化的DBN， 其检测准确率分别提高7.12%, 4.98%, 6.34%和3.82%， 可以作为轨道电路的故障检测的有效方法。
关键词：改进固有时间尺度分解； 样本熵； 深度置信网络； 故障检测
引用格式：WU Xiaochun， LIU Jiexin． Track circuit fault detection based on IITD sample entropy and improved deep belief network． Journal of Measurement Science and Instrumentation， 2023， 14（1）： 9-16. DOI： 10．3969／j．issn．1674-8042．2023．01．002
[full text view]
