基于深度学习算法的汽车装配件缺陷智能检测  

Intelligent detection of automotive assembly defects based on deep learning algorithms

作　　者：潘永清 杨雨 王秦 Pan Yongqing;Yang Yu;Wang Tao(Zhejiang Geely Automobile Engineering Technology Development Co.,Ltd.,Cixi 315300,China;Zhejiang Geely Automobile Parts&Components Co.,Ltd.,Cixi 315300,China;Zhejiang ZEEKR Automobile Research and Development Co.,Ltd.,Cixi 315300,China)

机构地区：[1]浙江吉利汽车工程技术开发有限公司,浙江慈溪315300 [2]浙江吉利汽车零部件采购有限公司,浙江慈溪315300 [3]浙江极氪汽车研究开发有限公司,浙江慈溪315300

出　　处：《汽车知识》2024年第12期51-53,共3页AUTOMOTIVE KNOWLEDGE

摘　　要：常规的汽车装配件缺陷智能检测方法主要使用AdaBoost Genetic Algorithm遗传决策树算法采集分类输入Hausdorff值,易受自适应阈值分割作用影响,导致检测性能指标不佳。本文提出了一种基于深度学习算法的汽车装配件缺陷智能检测方法,即利用深度学习算法提取汽车装配件缺陷特征,生成汽车装配件缺陷智能检测SE-ResUnit中心,从而完成对汽车装配件缺陷的智能检测。实验结果表明,设计的汽车装配件深度学习算法缺陷智能检测方法的各项检测指标良好,具有可靠性,有一定的应用价值,有助于提高产品质量、降低生产成本浪费。The conventional intelligent detection method for automotive assembly defects mainly uses the AdaBoost Genetic Algorithm to collect classification input Hausdorff values,which is easily affected by the adaptive threshold segmentation effect,resulting in poor detection performance indicators.This paper,an intelligent defect detection method for automobile assembly parts based on deep learning algorithm is proposed.By utilizing deep learning algorithms to extract defect features of automotive components and generate the SE ResUnit center for intelligent detection of automotive component defects,intelligent detection of automotive component defects has been achieved.The experimental results show that the intelligent defect detection method of the designed deep learning algorithm for automotive parts has good detection indicators,reliability,and certain application value.It has made certain contributions to improving product quality and reducing production cost waste.

关 键 词：深度学习算法 汽车 装配件 缺陷 智能检测 

分 类 号：U468.2[机械工程—车辆工程]