基于IGM-BP算法的城轨越区切换研究  

Research on the Handoff for Urban Rail Transit Based on IGM-BP Algorithm

作　　者：魏伟 刘晓娟[1] 张雁鹏[2] 李瑶 WEI Wei;LIU Xiaojuan;ZHANG Yanpeng;LI Yao(School of Electronic and Information Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]兰州交通大学电子与信息工程学院,兰州730070 [2]兰州交通大学自动化与电气工程学院,兰州730070

出　　处：《铁道标准设计》2021年第3期164-170,共7页Railway Standard Design

基　　金：国家自然科学基金项目(61763025);甘肃省自然科学基金项目(18JR3RA106)。

摘　　要：针对城市轨道交通列车在越区切换过程中因受到多径效应、同频干扰等因素的影响,引起接收到的参考信号接收功率(RSRP)值波动较大,导致乒乓切换频繁发生的问题,通过结合TD-LTE标准的A3事件,提出一种改进的GM-BP神经网络(IGM-BP)算法。根据车载数据终端(TAU)接收到的t时刻前4组RSRP值,建立灰色模型GM(1,1)并得到一组预测值,将预测值取均值作为期望值,为改善灰色预测算法对数据的波动性适应不理想的问题,再利用BP神经网络算法修正通过灰色预测算法得到的预测值,从而得到期望的RSRP值。仿真结果表明,与传统切换算法和灰色预测算法相比,利用IGM-BP算法可以使RSRP值波动幅度减小,降低乒乓切换率,提高切换成功率。Aiming at the problem that the urban rail transit train is affected by factors such as multipath effect and co-channel interference during the handoff process,the reference signal received power(RSRP)value fluctuates greatly,resulting in frequent occurrence of ping-pong handoff.An improved grey prediction-BP neural network(IGM-BP)algorithm is proposed by combining the A3 event of the TD-LTE standard.The terminal access unit(TAU)will receive the first four sets of RSRP values at time t to establish GM(1,1)model and get a set of predicted values.The average of the predicted value is taken as the expected value,and finally the predicted values,modified by the BP neural network algorithm,are used to obtain the expected RSRP value.The simulation results show that the IGM-BP algorithm,compared with the traditional handoff algorithm and the grey prediction algorithm,decreases the RSRP value fluctuation range,reduces the ping-pong handoff rate and improves the handoff success rate.

关 键 词：城市轨道交通 越区切换 RSRP TD-LTE A3事件 IGM-BP算法 乒乓切换 

分 类 号：U248.48[交通运输工程—道路与铁道工程]