表面粗糙度测量中稳健滤波算法的研究

发布时间：2018-04-12 14:20

本文选题：表面粗糙度 + 稳健估计　；参考：《哈尔滨理工大学》2017年硕士论文

【摘要】：现代社会,不仅工业上对工件表面的形貌特征的精细程度提出了越来越高的要求,日常生活中人们对自己使用的消费品的表面的加工精细度也有了一定的要求。粗糙度、波纹度、形状共同组成了物体的表面特征,表面的粗糙度形容的是零件表面的形貌特征,具有波长短(频率高)的特点,对它进行准确的测量与评定意义深远,对于零件的加工技术有非常重要的意义。本研究中使用的表面形貌的评定方法是建立在高斯滤波法的基础上的,实验数据是触针式表面粗糙度测量仪测得的。针对传统高斯滤波器在使用中的局限性,尤其是粗大误差及数据中特殊特征值的存在对受正态分布束缚的高斯滤波器造成的误导而导致的表面基准的畸形,引入了稳健理论对其进行改进。在明确稳健滤波算法需要达到的目标之后,对稳健性度量标准及经典的稳健估计方法进行了分析,并选定最常用的M估计作为稳健估计方法。接下来对M估计进行了详细的分析。稳健高斯滤波算法的实现选用了修正权法,具体的实现采用了迭代算法。针对不同的数据的特征,选择对应的稳健估计权函数进行处理从而达到最好的处理效果。算法实现的工具使用的是Matlab2015版,首先对数据进行预处理并作出数据图形,根据特征选用了QC权函数进行处理,并将稳健算法引入前后的滤波效果进行了对比,证明了适当稳健算法的引入能很好的抑制数据中异常特征对滤波基准造成的影响,能够更加准确的得到表面粗糙度滤波中线,满足了研究提出的要求。
[Abstract]:Roughness, corrugation and shape together make up the surface features of the object. The roughness of the surface describes the features of the surface morphology of the parts, and has the characteristics of short wavelength (high frequency). It is of profound significance to accurately measure and evaluate the roughness, corrugation and shape of the object.It is very important for the machining technology of parts.The evaluation method of surface topography used in this study is based on the Gao Si filter method and the experimental data are measured by a touch needle surface roughness measuring instrument.In view of the limitation of traditional Gao Si filter in use, especially the surface datum deformity caused by the error of gross error and the existence of special characteristic value in the data, which is caused by the misdirection of Gao Si filter bound by normal distribution.The robust theory is introduced to improve it.After defining the target of robust filtering algorithm, the robustness metrics and classical robust estimation methods are analyzed, and the most commonly used M estimators are selected as robust estimation methods.Then the M estimation is analyzed in detail.The robust Gao Si filtering algorithm adopts the modified weight method, and the iterative algorithm is used in the concrete implementation.According to the characteristics of different data, the corresponding robust estimation weight function is selected for processing to achieve the best processing effect.The tool to realize the algorithm is the Matlab2015 version. Firstly, the data is preprocessed and the data graph is made. According to the characteristics, QC weight function is selected to deal with it, and the filtering effect before and after the introduction of the robust algorithm is compared.It is proved that the introduction of the appropriate robust algorithm can restrain the influence of abnormal features on the filtering datum and obtain the surface roughness filtering midline more accurately, which meets the requirements of the research.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN713

【相似文献】