学术论文

      基于自动提取分段点的车轮外形轮廓拟合方法

      A wheel profile fitting method based on automatically extracting subsection points

      摘要:
      为准确拟合轮对外形轮廓线,提出了一种基于自动提取分段点的轮对外形轮廓拟合方法.首先根据设定的误差和阈值对激光位移传感器探测得到的轮对外形轮廓离散点进行自动分段处理,确定分段点,然后提出分段点区间的概念,以各个分段点为中心确定分段点区间,并对各个分段点区间的数据点进行最小二乘曲线拟合,最后采用拉格朗日乘数法将每条分段曲线平滑连接成一条完整的轮对外形轮廓线,并采用现场实测数据对该拟合方法进行了检验.实验结果表明,该方法拟合的轮廓曲线在分段点处平滑连续,且平均绝对误差为0.045 mm,最大偏移幅度小于0.05%,相较于基于手动分段的曲线拟合方法具有更高的拟合效率和拟合精度.该方法已在轮对尺寸在线检测系统中得到了成功的应用,提高了轮对尺寸结果的计算精度.
      Abstract:
      A curve fitting method based on automatically extracting subsection points from tread profile is proposed to fit the wheel tread profile accurately.Firstly, the subsection points are determined by segmenting discrete points of the wheel tread profile based on given errors and threshold.Secondly, the concept of subsection point interval is proposed, and each subsection point is taken as the interval center to extract a corresponding interval.Thirdly, the least square curve fitting method is adopted to fit all data points in the interval.Finally, all curves are connected to a smooth one by adopting the Lagrange multiplier method.The data obtained in the field were used to verify the fitting curve.The results show that the wheel tread profile fitted by this method is smooth at subsection points.The mean absolute error is 0.045 mm, and the largest offset is 0.05%.Compared to the manual segmentation method, the automatic segmentation method is more accuracy and effective.The method has been applied to the auto wheel-size detection system, which increases the accuracy of the detected results.
      Author: WANG Lu XING Zong-yi CHEN Shuang ZHANG Yong
      作者单位: 南京理工大学自动化学院,江苏南京,210094
      年,卷(期): 2017, 42(2)
      分类号: U270.331+.1
      在线出版日期: 2017年5月16日
      基金项目: 国家重点研发计划项目