摘要
为降低连续测力轮对在横向力和垂向力解耦过程中产生的系统误差,基于神经网络的原理,提出一种基于GA-LSTM的轮轨力连续测量方法。研究表明:相比于GB/T 5599—2019中的轮轨力测量方法,本轮轨力连续测量方法具有更高的精度和效率。将本轮轨力连续测量方法运用于国内某运营地铁线路的轮轨力信号实测,测试数据验证了其有效性。
In order to reduce the systematic error in the decoupling process of lateral and vertical forces of continuous instrumented wheelsets,a continuous measurement method of wheel-rail forces based on GA-LSTM is proposed according to the principle of neural network.The study shows that compared with the method in the national standard GB/T 5599-2019,the proposed method has higher accuracy and efficiency,and the measured data of wheel-rail forces gained by the method in a domestic operating subway line verify its effectiveness.
作者
汪卫
陈建政
吴越
WANG Wei;CHEN Jianzheng;WU Yue(State Key Laboratory of Traction Power,Southwest Jiaotong University,Chengdu 610031,China)
出处
《机械制造与自动化》
2024年第4期111-118,共8页
Machine Building & Automation
基金
国家自然科学基金项目(u1734201)
关键词
测力轮对
神经网络
GA-LSTM
轮轨力测量
instrumented wheelset
neural network
GA-LSTM
measurement of wheel rail force
