基于图像分析的粘接质量评价系统设计

Design of bonding quality evaluation system based on image analysis technology

摘要:

目前对发动机防热结构件的粘接质量的判断采用电磁超声波进行无损检测，检测人员只能通过超声波返回的A波波形进行粘接缺陷的评估。这种检测方法效率低，准确度低，对人员要求高且无法实现在线检测，严重影响产品量产。针对此类情况，采用数据可视化、智能图像分析的方法，建立一套粘接质量评价系统。系统包含了对测量结果数据进行过滤处理；根据工件尺寸按照法线函数展开成二维模型；为二维模型填充缺陷信息从而实现粘接效果可视化；采用二值法建立缺陷矩阵；通过粒子法求解缺陷的面积；采用缺陷矩阵扩张的方法，解决相邻缺陷的识别问题；再通过特征值数组的方法解决从3D展开至2D时展开边界切割缺陷区域的问题，从而完整地描述发动机防热结构件粘接质量的全面特征，为粘接质量评价提供了直接依据。

Abstract:

At present, electromagnetic ultrasonic wave is used to judge the bonding quality of engine thermal protection structure parts, and the inspection personnel can only evaluate the bonding defects through a-wave shape returned by ultrasonic wave. This detection method has low efficiency, low accuracy, high requirements for personnel and can not realize online detection, which seriously affects the mass production of products. In view of this kind of situation, a set of adhesive quality evaluation system is established by means of data visualization and intelligent image analysis. The system includes filtering the measured data; expanding the two-dimensional model according to the normal function according to the size of the workpiece; filling the defect information for the two-dimensional model to realize the visualization of the bonding effect; using the binary method to establish the defect matrix; using the particle method to solve the defect area product; using the defect matrix expansion method to solve the problem of adjacent defects identification; and then through the feature The method of value array can solve the problem of cutting the defect area from 3D to 2D, so as to fully describe the overall characteristics of bonding quality of engine thermal protection structure parts, and provide a direct basis for bonding quality evaluation.