摘要
为了实现对低真空管道中运行列车的最大阻力预测研究,本文采用数值仿真和神经网络结合的方法。选取不同阻塞比、运行速度和管道压力,利用流体仿真软件计算100种运行工况下列车的最大阻力;以96组仿真数据作为网络模型训练样本,选取RBF和BP两种三层神经网络,经多次调试确定最佳隐层神经元数目,利用训练函数训练两种预测模型;利用随机选取的4组验证样本验证两种网络模型。研究表明:RBF和BP神经网络模型能较好的预测列车在真空管道中运行的最大阻力,其中RBF神经网络预测值的最大误差不高于5%,相比BP神经网络,RBF预测精度更优。
The flow field,around the high-speed train in vacuum tube,was mathematically formulated with 3 D standard k-εturbulence model and numerically simulated with Fluent software.To begin with,the influence of the possible situations,including the 100-cases specified in a combination of variables of blocking-ratio,train-speed and pressure,on the aerodynamic resistance was investigated;second,96-sets of simulated results were analyzed as the training samples of the three-layer RBF and BP neural network models,for determination and optimization of the hidden-layer-neurons number and training functions;and finally,the two neural network models were verified with the 4 randomly selected samples,respectively.The analysis results show that when it comes to prediction of the largest resistance,RBF neural network model,with the largest prediction error of≤5%,outperforms that of BP because of better accuracy.
作者
冯瑞龙
王志飞
冯海全
李樊
杜呈欣
Feng Ruilong;Wang Zhifei;Feng Haiquan;Li Fan;Du Chengxin(School of Mechanical Engineering,Inner Mongolia University Of Technology,Hohhot 010051,China;Instiue of Computing Technology,China Acadery Of Railway Sciences,Beijing 100081,China)
出处
《真空科学与技术学报》
EI
CAS
CSCD
北大核心
2020年第9期827-832,共6页
Chinese Journal of Vacuum Science and Technology
基金
中国铁路总公司科技研究开发计划课题(K2018T004)
中国工程院重大咨询项目(2018-ZD-16)。
关键词
数值仿真
神经网络
预测模型
最大阻力
Numerical simulation
Neural network
Prediction models
Maximal resistance