GPS信号失锁下的列车组合定位方法  

Train integration positioning method in GPS signal lost lock

作　　者：王妍 陈永刚[1] WANG Yan;CHEN Yonggang(School of Automatization and Electric Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]兰州交通大学自动化与电气工程学院,兰州730070

出　　处：《导航定位学报》2020年第4期50-57,共8页Journal of Navigation and Positioning

基　　金：国家自然科学基金地区科学基金项目(61763023)。

摘　　要：针对列车在卫星信号失锁情况下定位精度低的问题,提出1种基于灰色模型(GM(1,1))和长短时记忆网络(LSTM)的全球定位系统(GPS)与捷联惯性导航系统(SINS)组合定位方法:使用GM(1,1)模型对传感器原始数据进行粗预测,以降低数据存在的误差;然后利用LSTM网络进行网络训练,并利用训练好的网络预测GPS信号失锁情况下的列车位置序列;最后通过模拟列车运行,分析不同参数对网络训练的影响,并且从模型输出对单个特征的预测效果的角度出发,与GM(1,1)-BP神经网络预测结果进行对比。结果表明,在GPS信号失锁情况下,GM(1,1)-LSTM网络预测的定位结果误差较小,能够满足列车定位的要求。Aiming at the problem of low positioning accuracy under the condition of GPS signal lost lock,the paper proposed a GPS/SINS integration positioning method based on GM(1,1)and LSTM:GM(1,1)model was used to make a rough prediction of the sensor original data to reduce the errors;LSTM network was used for network training and the trained network was implemented to predict the train position sequences under the condition of GPS signal lost lock;then the train operation was simulated to analyze the influence of different parameters on the network training,and the outputs between SINS single positioning and GM(1,1)-BP neural network prediction in the signal lost lock were compared from the aspect of prediction effect of model output on individual features finally.Results showed that the proposed method would have less errors of location,which could meet the requirement of train positioning.

关 键 词：列车定位 运营安全 卫星信号失锁 灰色模型(GM) 长短时记忆网络(LSTM) GM(1 1)-LSTM网络 

分 类 号：P228[天文地球—大地测量学与测量工程]