摘要
小波分析具有良好的时频局部性,在工程表面计量中取得了广泛的运用。在分析小波滤波提取粗糙度评定基准稳健性时,发现小波滤波方式存在与标准高斯滤波方式相同的稳健性差问题。为此,提出一种基于M估计理论的稳健小波滤波方法,列举了四种备选稳健估计权函数,并给出了一种稳健估计权函数调节参数的设置方法。仿真实验和实测实验结果表明,该方法提高了小波滤波方式的稳健性,抑制了异常点对表面评定基准提取的影响。
The wavelet analysis with better time-frequency localization has a wide range of application in measurement of engineering surface. The poor robustness problem of wavelet filter has been found when extract the mean line for the roughness profile and this problem is the same to the Gaussian filter. To solve it,a robustness wavelet filtering method based on M-estimate theories is put forward. In the same time,four kinds of alternative robust weight functions are listed and a setting method of adjustable parameters is given. The results of simulation experiment and the practical test show that this method can improve robustness of the wavelet filter. And the influence of anomalous points to the extraction of the surface evaluation reference is inhibited by this way.
出处
《机械设计与研究》
CSCD
北大核心
2016年第2期60-64,共5页
Machine Design And Research
基金
国家自然科学基金资助项目(51365009)
广西制造系统与先进制造技术重点实验室主任课题(13-051-09-009Z)
桂林电子科技大学研究生教育创新计划资助项目(YJCXS201501)
关键词
表面计量
小波分析
稳健滤波
M估计理论
surface metrology
the wavelet filter
robust filter
M-estimate theories
