摘要
针对不同列车物理属性差异导致的ZPW-2000R轨道电路输出信号数据长度不一致而难以建立统一的神经网络模型问题,提出一种基于数值分析和神经网络的轨道过车信号判断模型。首先,利用数值分析方法将显示轨道是否占用的主机调接入电压数据集处理成长度一致的结构;其次,以处理好的数据集作为输入,以轨道占用或空闲状态作为输出,建立基于神经网络的轨道过车信号判断模型。实际测试结果表明,基于数值分析和神经网络的轨道过车信号判断模型对轨道占用或空闲情况的判断准确率为100%。研究可在保证列车安全可靠运行的基础上实现ZPW-2000R轨道电路输出信号的智能判断,为轨道电路利用主机主接入电压和主机调接入电压信号联合判断轨道占用情况提供了一种智能算法。
Aiming at the problem that the length of output signal data of ZPW-2000R track circuit is inconsistent due to the differences in physical properties of different trains,which makes it difficult to establish a unified neural network model,a model for judging track passing signals based on numerical analysis and neural network is proposed.First,numerical analysis is used to process the host switch-in voltage data set that shows whether the track is occupied or not into a structure with consistent length;then a neural network-based judgement model of track passing signal is established,taking the processed data set as input and track occupancy or idle state as output.The actual test results show that the judgement model of track passing signal based on numerical analysis and neural network can accurately judge the track occupancy or vacancy.The research can realize the intelligent judgment of the output signal of the ZPW-2000R track circuit on the basis of ensuring safety and reliability,and provides an intelligent algorithm for realizing the joint judgment of the track occupancy by using the host main access voltage and the host adjustment access voltage signal in the track circuit.This is of great practical significance for the improvement of the efficiency of the signal recognition of the track traffic and the decrease of the cost.
作者
邓阳
陈洪根
黄春雷
禹建丽
DENG Yang;CHEN Honggen;HUANG Chunlei;YU Jianli(School of Management Engineering,Zhengzhou University of Aeronautics,Zhengzhou,Henan 450046;Heilongjiang Ruixing Technology Co.,Ltd,Harbin,Heilongjiang 150030)
出处
《郑州航空工业管理学院学报》
2022年第1期106-112,共7页
Journal of Zhengzhou University of Aeronautics
基金
国家自然科学基金资助项目(U1904167)
教育部人文社科基金资助项目(20YJCZH235,18YJAZH125)
河南省高校科技创新团队资助项目(21IRTSTHN018)
河南省科技攻关资助项目(212102210338)
郑州航空工业管理学院研究生教育创新计划基金资助项目(2021CX11)。
关键词
数值分析
神经网络
轨道交通
轨道电路
过车信号判断
numerical analysis
neural network
track traffic
track circuit
judgment of track passing signal
