【作者】 张磊；

【导师】 凤雷；

【作者基本信息】 哈尔滨工业大学 ， 仪器仪表工程（专业学位）， 2021， 硕士

【摘要】 城市轨道交通作为一种新兴、大运量、高效率的客运交通方式,在缓解常规公共交通压力、保障市民安全出行和加速城市发展中扮演着重要角色。随着城市人口增多、轨道交通负荷增大和列车运行速度提高,钢轨磨损越来越快,钢轨磨损程度直接关乎轨道交通的运营安全,若维护不当极容易发生重大安全事故。钢轨磨耗预测技术,可以准确掌握钢轨变化趋势,为指导钢轨打磨或更换作业提供科学依据,保证轨道交通安全运行。钢轨的磨损种类和磨损速度受多种因素影响,其变化规律属于复杂非线性问题,传统方法不能准确地进行预测。针对哈尔滨地铁小半径曲线段钢轨的磨耗测量工作需求,设计了一种基于机械电子式测量原理的便携式侧面磨耗测量小车,解决了哈尔滨地铁原有点接触式测量仪器的测量效率低、测量精度受人为作业水平干扰较大等弊端,提升了钢轨磨耗测量工作的效率和精度,同时减少了钢轨磨耗测量作业的人力成本投入,为哈尔滨地铁钢轨养护工作提供了一种新方法。本文以哈尔滨地铁小半径曲线处侧磨为研究对象,通过测量实验段轨道12个周期的侧磨值发生情况,对影响钢轨磨损速度的主要变量进行信息收集。随后,建立了基于BP神经网络哈尔滨地铁小半径曲线侧磨仿真模型,并以收集到的数据样本进行训练,提升模型预测精度。最后,对所建立的基于BP神经网络哈尔滨地铁小半径曲线侧磨仿真模型进行实际验证,预测样本均方误差为0.233,满足哈尔滨地铁小半径曲线段钢轨的磨耗预测需求。更多还原

【Abstract】 As a kind of emerging,large volume and high efficiency modes of passenger transportation,the rail traffic plays an important role in relieving regular public transportation pressure,ensuring the safety of citizens travel and accelerating urban development.As the increasing of the city’s population,the load and the improving of train speed,the abrsion of the rail get faster.The degree of abrsion is directly related to the operation safety of the rail traffic.If the maintenance is not appropriate,accidents are prone to occur.The prediction technology of rail traffic can master accurately the variation trend of rail,which can provide scientific basis in guiding rail grinding and changing work,and guarantee the safety of rail traffic.The kinds and speed of the abrsion of the rail are affected by many factors.The changing rule belongs to non-linear problem and the traditional method can’t predict accurately.According to the abrsion measurement requirements of rail in the small radius curve sections of Harbin metro,this thesis designs a portable side abrsion measurement trolley based on the principle of mechatronic measurement,which makes the measurement more efficient and more accurate.This thesis takes the small radius curve side wear of Harbin subway rail as the research object,though the measurement of occurrence of side abrasion values in 12 cycles of the experimental section rail,conduct information collections in the main variation that affects the rail abrasion speed.Subsequently,it builds the side abrsion simulation model of the small radius curve of Harbin metro based on BP neural network.It trains the data samples and improves the model’s prediction accuracy.Finally,Mean square error of prediction sample reach to 0.233,which meets the rail wear prediction demand of the small radius curve of Harbin metro.更多还原