【作者】 张伟；

【导师】 何彦；

【作者基本信息】 重庆大学 ， 机械工程， 2022， 硕士

【摘要】 表面粗糙度作为衡量工件表面质量的一个重要指标,与工件的工作性能和使用寿命紧密相关。随着目前机器视觉技术的快速发展,大量的基于机器视觉的粗糙度检测方法被研究出来,以实现工件粗糙度快速、非接触的检测。然而目前基于视觉的表面粗糙度检测研究大多针对规整的方形或圆柱形工件,而对广泛使用的螺旋曲面工件等复杂形状工件的研究较少,导致现有的粗糙度视觉检测方法难以实现该类工件准确、有效的粗糙度检测。因此,本文以螺旋曲面工件为研究对象,对其基于机器视觉的表面粗糙度检测方法进行研究。主要研究内容如下:首先,对螺旋曲面工件表面粗糙度的视觉检测进行了分析,从成像原理说明利用视觉评估其粗糙度的可行性,同时研究该类工件结构和与成像对其粗糙度检测的影响。在此基础上,针对性的设计一种基于机器视觉的螺旋曲面工件粗糙度检测方案。其次,针对所捕获的螺旋曲面工件表面图像中存在大量非检测区域的干扰信息的问题,开展螺旋曲面工件感兴趣区域(ROI)提取方法研究。通过图像分割技术与形态学处理技术对原始图像进行处理,生成平滑的精确的待测目标区域边界。在此基础上,利用掩膜图像处理方式提取待测目标区域信息,实现ROI提取过程。最后,考虑螺旋曲面工件表面粗糙度评估过程存在特征信息冗余、高维度、回归性、数据量少等特点,开展了其特征处理与评估模型相关研究,并提出了一种基于主成分分析-支持向量机(PCA-SVR)的粗糙度评估模型。利用基于纹理和基于统计的描述方法对ROI提取后的图像进行描述并提取量化指标,综合考虑高维特征参数间的有效信息,采用PCA方法对原始图像特征进行降维处理并去除特征参数中大量的冗余信息。基于降维后得到的主成分特征数据与工件表面粗糙度数据,采用SVR建模方法,建立了特征数据与粗糙度之间的关系映射模型。基于上述研究,针对一种具有代表性的螺旋曲面工件(斜齿轮)建立粗糙度视觉检测系统。结合斜齿轮的复杂结构特点,对系统的硬件平台进行设计;基于所提出的螺旋曲面工件粗糙度视觉检测方法对系统的软件系统进行开发。在上述基础上,利用高速干切滚齿加工后的斜齿轮对检测系统的功能进行测试,验证所建立的粗糙度视觉检测系统的可行性。更多还原

【Abstract】 As an important index to measure the surface quality of workpiece,surface roughness is closely related to the working performance and service life of workpiece.With the rapid development of machine vision technology,a large number of roughness detection methods based on machine vision have been studied to achieve rapid and non-contact detection of workpiece roughness.However,most of the current research on surface roughness detection based on vision is aimed at regular square or cylindrical workpieces,and the research on the widely used complex shape workpieces such as helical surface workpieces is less,resulting in the existing roughness visual detection methods are difficult to achieve accurate and effective roughness detection of such workpieces.Therefore,this paper takes the helical surface workpiece as the research object,and studies its surface roughness detection method based on machine vision.The main research contents are as follows:Firstly,the visual detection of the surface roughness of the helical surface workpiece is analyzed,and the feasibility of using vision to evaluate its roughness is explained from the imaging principle.At the same time,the influence of the workpiece structure and imaging on its roughness detection is studied.On this basis,a roughness detection scheme of helical surface workpiece based on machine vision is designed.Secondly,in order to solve the problem that there are a large number of non-detection region interference information in the captured surface image of helical surface workpiece,the region of interest(ROI)extraction method of helical surface workpiece is studied.The original image is processed by image segmentation technology and morphological processing technology to generate smooth and accurate target area boundary.On this basis,the mask image processing method is used to extract the target area information,and the ROI extraction process is realized.Finally,considering the characteristics of feature information redundancy,high dimension,regression and less data in the process of surface roughness evaluation of helical surface workpiece,the related research on feature processing and evaluation model is carried out,and a roughness evaluation model based on principal component analysis-support vector machine(PCA-SVR)is proposed.The texture-based and statistical-based description methods are used to describe the image after ROI extraction and extract quantitative indicators.Considering the effective information between high-dimensional feature parameters,PCA method is used to reduce the dimension of the original image features and remove a large number of redundant information in the feature parameters.Based on the principal component feature data and workpiece surface roughness data obtained after dimension reduction,the relationship mapping model between feature data and roughness is established by SVR modeling method.Based on the above research,a roughness visual detection system is established for a representative helical surface workpiece(helical gear).Combined with the complex structural characteristics of helical gears,the hardware platform of the system is designed.The software system is developed based on the proposed visual detection method of workpiece roughness of helical surface.On the basis of the above,the function of the detection system is tested by using the helical gear processed by high speed dry cutting hobbing,and the feasibility of the established roughness visual detection system is verified.更多还原