基于自学习算法的电缆综合诊断故障自动感知系统  

Automatic sensing system of cable comprehensive diagnosis fault based on self-learning algorithm

作　　者：谭云 TAN Yun(Changjiu Intercity Railway Co.,Ltd.,Nanchang 330001,China)

机构地区：[1]昌九城际铁路股份有限公司,江西南昌330001

出　　处：《电子设计工程》2022年第9期121-124,129,共5页Electronic Design Engineering

摘　　要：当铁路电缆发生故障时,为了及时获取故障位置信息,设计了一种基于自学习算法的电缆综合诊断故障自动感知系统。考虑电缆故障模糊性特征,构建具有良好非线性函数逼近特征的模糊基函数网络,获取电缆系统动态行为;对所有新增样本作出迭代处理,完成网络中心与半径的自学习,利用正态型多元模糊分布确定新增样本的理想输出,计算初始样本的矩阵与协方差矩阵,经过递推运算,即可得到新样本中的全部故障可能性分布信息;针对电缆故障诊断缺陷,分别从分布式短路故障采集处理设备与接地故障在线测温设备两方面设计系统硬件模块,并实现软件模块的复位、储存与通信等功能。仿真实验结果表明,该系统故障诊断更加全面且定位精准。In order to obtain the fault location information in time when the railway cable is in fault,an automatic sensing system of cable comprehensive diagnosis based on self-learning algorithm is designed.Considering the fuzziness characteristics of cable fault,a fuzzy basis function network with good nonlinear function approximation characteristics is constructed to obtain the dynamic behavior of cable system;All new samples are iteratively processed to complete the self-learning of network center and radius,the ideal output of new samples is determined by using normal multivariate fuzzy distribution,and the initial sample matrix and covariance matrix are calculated;According to the defects of cable fault diagnosis,the hardware modules of the system are designed from two aspects of distributed short circuit fault acquisition and processing equipment and ground fault online temperature measurement equipment,and the functions of software module reset,storage and communication are realized. Simulation results show that the system fault diagnosis is more comprehensive and accurate.

关 键 词：自学习算法 电缆综合诊断 自动感知 模糊基函数网络 递推运算 

分 类 号：TN325[电子电信—物理电子学]