基于RBF的汽车纵梁加工质量检测系统研究  

Machining quality study of the auto rack girder detection system based on RBF neural network

作　　者：王华[1] 高金刚[1] 张爽[1] 

机构地区：[1]长春工程学院机电工程学院,长春130012

出　　处：《长春工程学院学报（自然科学版）》2010年第3期63-67,共5页Journal of Changchun Institute of Technology：Natural Sciences Edition

基　　金：吉林省科技发展计划项目(20060534)

摘　　要：针对大尺寸的汽车车架纵梁装配孔检测方式落后、效率低的问题,提出了一种基于RBF神经网络和线形结构光的汽车纵梁装配孔的机器视觉在线检测系统。利用RBF可以逼近高阶输入输出非线性系统的特性,校正摄像机成像过程中的高阶畸变,提高了系统的检测精度。该检测系统经生产现场测试,在汽车车架尺寸范围内其装配孔尺寸检测精度达到±0.1 mm,位置检测精度不大于±0.3 mm,检测时间不超过1.5 min/根,能够满足汽车车架纵梁在线检测的要求。On account of out-of-date and low efficiency detection mode for large dimension auto rack girder,it is presented that is the machine vision online detection system for the auto rack girder based on RBF and linear structured light.Taken advantage of RBF characteristic that approximating high order input and output nonlinear system,camera image error correction high order distortion model is constructed and the precision of camera detection is raised.After tested at production field,the precision of detection diameter is ±0.1mm,the precision of detection position is less than±0.3mm,and the detection time is less than 1.5 min per piece in the range of the auto rack girder dimension.These meet the demand of online detection of the auto rack girders.

关 键 词：RBF神经网络 线阵CCD 大尺度 镜头畸变校正 汽车纵梁 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]