铁道科学与工程学报2025，Vol.22Issue(1)：136-145,10.DOI:10.19713/j.cnki.43-1423/u.T20240453
稳定车作业下道床横向阻力在线检测模型研究
Online detection model of lateral resistance of sleeper under stabilizer operating
摘要
Abstract
Stabilizer is a kind of large-scale rail transit operation and maintenance equipment,which can improve the lateral resistance of sleeper by stabilizing operation,but can not detect it online.In order to explore the online detection method of the lateral resistance of sleeper under stabilizer operating,field tests were carried out on the ballast track line to obtain the online signals of stabilizing operation parameters and the lateral acceleration of stabilizing device.Before and after the operation,the lateral resistances of sleepers were detected offline.A space window with the same length as the longitudinal spacing of the sleepers was used to intercept the online signal,so that the online test results match the offline test results of the lateral resistance of the sleepers after stabilizing.The correlation between the test data of stabilizing operation and the lateral resistance of sleeper after stabilizing was evaluated by grey relational analysis method.The characteristic parameters of lateral resistance of sleeper were extracted.According to the test data set,the structure of RBF neural network model was determined.The parameters of RBF neural network model were optimized by PSO algorithm,and then the calculation errors of PSO-RBF model and RBF model for lateral resistance of sleeper were compared.The results are drawn as follows.The grey correlation between excitation frequency,running velocity,left downward pressure,right downward pressure,lateral acceleration of stabilizing device and lateral resistance of sleeper after stabilizing is 0.68,0.64,0.70,0.70 and 0.71,respectively.The online characteristic parameters can reflect the off-line lateral resistance characteristics of sleeper.In the test set verification,compared with RBF model,the maximum absolute error of PSO-RBF model is reduced by 54.12%.The average absolute error is reduced by 47.30%,the root mean square error is reduced by 44.21%,and the R squared is increased from 0.90 to 0.97.The introduction of PSO algorithm improves the calculation accuracy of the lateral resistance model of sleeper.The research results can provide a theoretical basis for online detection of lateral resistance of sleeper under stabilizer operating,and promote the intelligent development of rail transit operation and maintenance technology.关键词
稳定车/道床横向阻力/RBF神经网络/粒子群优化/灰色关联度Key words
stabilizer/lateral resistance of sleeper/RBF neural network/particle swarm optimization/gray relational analysis分类
交通工程引用本文复制引用
陈春俊,江浩,林梦..稳定车作业下道床横向阻力在线检测模型研究[J].铁道科学与工程学报,2025,22(1):136-145,10.基金项目
国家自然科学基金资助项目(52372402,U2034210) （52372402,U2034210）
