China Mechanical Engineering ›› 2022, Vol. 33 ›› Issue (01): 62-69.DOI: 10.3969/j.issn.1004-132X.2022.01.007

Previous Articles     Next Articles

Road Roughness Recognition Based on Vehicle Body Dynamic Response of Tracked Vehicles

LING Qihui1;DAI Juchuan1;CHEN Shengzhao1;SUN Feiying2;WANG Guosheng3;LIAO Lili1   

  1. 1.School of Mechanical Engineering,Hunan University of Science Technology,Xiangtan,Hunan,411201
    2.Jianglu Machinery Electronics Group Co.,Ltd.,Xiangtan,Hunan,411001
    3.China North Vehicle Research Institute,Beijing,100072
  • Online:2022-01-10 Published:2022-01-19

基于履带车辆车体动态响应的行驶路面不平度识别

凌启辉1;戴巨川1;陈盛钊1;孙飞鹰2;汪国胜3;廖力力1   

  1. 1.湖南科技大学机电工程学院,湘潭 411201
    2.江麓机电集团有限公司,湘潭 411100
    3.中国北方车辆研究所,北京 100072
  • 作者简介:凌启辉,男,1986年生,副教授。研究方向为复杂机电系统动力学行为监测与控制。发表论文20余篇。E-mail:lqh_hunan@163.com。
  • 基金资助:
    湖南省教育厅优秀青年项目(157948);
    湖南省科技创新计划(2021RC4038);
    机械设备健康维护湖南省重点实验室开放基金(202002)

Abstract: A model for road surface roughness recognition was established based on dynamic response of tracked vehicle body. NARX neural network structure was adopted in the model, and dynamic response signals of tracked vehicle body were taken as inputs and road surface roughness values were taken as outputs. Correlation coefficient, root mean square error, and absolute error cumulative probability density were proposed as indexes of recognition effectiveness evaluation, and the fusion method of the three indexes was proposed. Based on orthogonal experimental design, the balance between the number of input and the recognition effectiveness of road roughness recognition model was analyzed and realized, which simplified the layout of the sensor test system. The recognition effectiveness of road roughness under different road surfaces, different sampling frequencyies, and different speeds was analyzed. The results show that the proposed model may satisfy the practical engineering needs.

Key words: tracked vehicle, road roughness recognition, dynamic response, nonlinear auto-regressive with exogeneous inputs(NARX) neural network

摘要: 建立了基于履带车辆车体动态响应的行驶路面不平度识别的模型。该模型采用带外源输入的非线性自回归神经网络结构,以履带车辆车体动态响应为输入、路面不平度为输出。将相关性系数、均方根误差和绝对误差累计概率密度作为识别效果的评价指标,并给出了上述三个指标的融合方法。基于正交试验设计的思路分析并实现了路面不平度识别模型输入数量和识别效果的平衡,简化了测试系统传感器的布置。分析了不同的路面、采样频率和车速下的路面不平度识别效果。结果表明,提出的不平度识别方法满足工程实际需求。

关键词: 履带车辆, 路面不平度识别, 动态响应, 带外源输入的非线性自回归神经网络

CLC Number: