基于RBF-NN近似模型的地铁钢轨探伤车转向架悬挂参数多目标优化  

Multi-objective Optimization of Bogie Suspension Parameters of Metro Rail Flaw Detection Cars Based on RBF-NN Approximate Model

作　　者：肖乾 程玉琦 李超 周生通 高雪山 黄敏 XIAO Qian;CHENG Yuqi;LI Chao;ZHOU Shengtong;GAO Xueshan;HUANG Min(Key Laboratory of Vehicles and Equipment of Ministry of Education,East China Jiaotong University,Nanchang 330013,China;CRRC Zhouzhou Electric Locomotive Co.,Ltd.,Zhuzhou 412000,Hunan,China)

机构地区：[1]华东交通大学载运工具与装备教育部重点实验室,南昌330013 [2]中车株洲电力机车有限公司,湖南株洲412000

出　　处：《噪声与振动控制》2022年第2期59-66,共8页Noise and Vibration Control

摘　　要：以某地铁钢轨探伤车为研究对象,利用多体动力学软件UM建立车辆-轨道耦合动力学模型,分析探伤车在直线、曲线两种工况下的运行平稳性、安全性和曲线通过能力,为后续悬挂参数多目标优化提供原始数据支撑。基于MATLAB编程建立UM-ISight联合仿真平台,在ISight中以转向架一、二系悬挂刚度、阻尼等为目标参数,通过最优拉丁超立方法进行参数组合样本设计,并在此基础上构建符合精度要求的径向基函数神经网络(RBF-NN)近似模型,最后通过多目标优化算法NSGA-Ⅱ对近似模型进行寻优计算,得到最优的转向架悬挂参数组合。结果表明,所建近似模型具有较高的拟合精度,优化后车辆平稳性指标都得到明显改善,最多可降低35.39%,且车辆的脱轨系数、轮轨横向力、轮轴横向力、轮重减载率等曲线通过性能指标也有不同程度好转。With a subway rail flaw detection vehicle as the research object, its vehicle-track coupling dynamics model is established by using the multi-body dynamics software UM. The operation stability, safety and curve passing capacity of the flaw detection vehicle running on straight or curved tracks are analyzed to provide raw data support for the multi-objective optimization of suspension parameters. The UM-ISight joint simulation platform is established based on MATLAB programing. With the bogie primary and secondary suspension stiffness, damping value, etc. as the target parameters, the optimal Latin Hypercube method is applied to design the parameter composition samples. On this basis, a radial base-function neural network(RBF-NN) approximate model, which meets the accuracy requirements, is constructed. Finally, the multi-objective optimization algorithm NSGA-Ⅱis used to optimize the RBF-NN approximate model and obtain the optimal combination of bogie suspension parameters. The results show that the established RBF-NN approximate model has high fitting accuracy. After the optimization, the vehicle stability indicators are significantly improved by up to 35.39 %. The vehicle’s derailment coefficient, wheel-rail lateral force, wheel axle lateral force, wheel load reduction rate and other indicators have also improved.

关 键 词：振动与波 钢轨探伤车 车辆-轨道耦合 悬挂参数 近似模型 多目标优化 

分 类 号：U213.2[交通运输工程—道路与铁道工程]