Rapid Intelligent Detection Method of Micro Defects of Injection Molding Products Based on Deep Learning
孟雨涵 MENG Yu-han；沈天成 SHEN Tian-cheng；谭立 TAN Li；
薄康莹 BO Kang-ying；徐小青 XU Xiao-qing
（常州机电职业技术学院，常州 213164）
（Changzhou Vocational Institute of Mechatronic Technology，Changzhou 213164，China）
摘要：在注射成型过程中，由于各种因素的影响，注塑制品可能会出现短射、飞边、熔接痕、气泡、裂纹等表面微缺陷，这些微缺陷仅凭人工方法是无法检测和识别出的。针对此问题，本文以注塑制品微缺陷为研究对象，构建注塑制品表面微缺陷数据集，设计适用于注塑制品表面微缺陷的深度卷积神经网络，利用深度学习方法训练数据集，建立卷积神经网络学习评价标准和机制，获得注塑制品表面微缺陷智能识别模型，进而构建出注塑制品微缺陷快速智能无损检测方法，促进了注塑制品检测的智能化、精准化、快速化发展。
Abstract: In the process of injection molding, due to the influence of various factors, injection molding products may have surface micro defects such as short shot, flash, weld mark, bubble and crack. These micro defects cannot be detected and identified only by manual methods. To solve this problem, this paper takes the micro defects of injection molded products as the research object, constructs the surface micro defects data set of injection molded products, and designs the deep convolution neural network suitable for the surface micro defects of injection molded products, and uses the deep learning method to train the data set, and establishes the convolution neural network learning evaluation standard and mechanism, and obtains the intelligent recognition model of surface micro defects of injection molded products, so a rapid intelligent nondestructive testing method for micro defects of injection molded products is constructed, which promotes the intelligent, accurate and rapid development of injection molded products testing.
关键词：智能检测；注塑制品；微缺陷检测；卷积神经网络
Key words: intelligent detection；injection molding products；micro defect detection；convolution neural network.
中图分类号：TH74 文献标识码：A 文章编号：1006-4311（2022）20-111-03
DOI:10.3969/j.issn.1006-4311.2022.20.037.
文章出处：孟雨涵,沈天成,谭立,等. 基于深度学习的注塑制品微缺陷快速智能检测方法初探[J]. 价值工程,2022,41(20):111-113.
|
