一种可降低汽车白车身装配偏差的改进DBSCAN算法  

An Improved DBSCAN Algorithm for Deviation Reduction of Automobile BIW Assembly

作　　者：林静 康竞然 张天侃 张晨曦 高艳俊 Lin Jing;Kang Jingran;Zhang Tiankan;Zhang Chenxi;GaoYanjun(Geely Automobile Research Institute(Ningbo)Co.,Ltd.,Ningbo 315336)

机构地区：[1]吉利汽车研究院(宁波)有限公司,宁波315336

出　　处：《汽车工艺与材料》2023年第1期56-60,共5页Automobile Technology & Material

摘　　要：为了提高白车身装配质量,结合白车身特征点分布不均匀的特性、特征点的非空间属性,以及基于构建的有权无向网络图,提出了一种参数自适应的DBSCAN算法,对连续离线白车身进行区域装配偏差识别,实现偏差区域诊断和自动报警。试验表明,本研究方法能有效监控误差源,及时解决各种问题引起的偏差,提高整车制造质量。Considering the uneven distribution of Body in White(BIW) feature points, combined with the nonspatial attributes of feature points and based on the undigraph with weight, this paper proposed a parameter adaptive DBSCAN algorithm to improve the assembly quality of BIW which can identify the regional assembly deviation of continuous off-line BIW and realize the deviation area diagnosis and automatic alarm. Experiments show that this method can effectively monitor the error source, solve various faults in time and improve the quality of vehicle manufacturing.

关 键 词：DBSCAN 白车身 装配偏差识别 有权无向网络图 

分 类 号：TP301[自动化与计算机技术—计算机系统结构] TP305[自动化与计算机技术—计算机科学与技术]