Research Of Inspection Technology For Gearbox Parts Assembly Anti-mistake Based On Machine Vision

Visual inspection is an important research topic in the field of intelligent manufacturing.Compared with the traditional method of artificial detection,the machine vision detection technology has the characteristics of intelligence,flexibility and automation in the testing process of the gearbox assembly.This advanced technology meets the requirements for stability and reliability in large mass assembly manufacturing,so it has great practical application value.Based on the research of machine vision technology,this topic has developed a visual on-line inspection system,which is used in the process of assembly detection for the gearbox.This system can meet the requirements of detection in the case of intelligent manufacturing.According to the characteristics of the gearbox assembly process,the paper relies on the image feature extraction and the support vector machine to achieve the purpose of automated inspection.The system can be used for quality control during assembly process to prevent the wrong and leakage assembly of key parts.It can effectively solve the problems of different assembly defects in conventional artificial detection,and ensure the speed,accuracy and reliability of the detection.The main contents of this paper are as follows:Through the analysis of the visual inspection system,the overall plan and the detection process are determined.In order to obtain the high quality gearbox image,combined with the characteristics of the parts,a better lighting scheme is designed in this paper.Then,the subject calculates the key parameters of the industrial camera,lianr and so on in the image acquisition system,and finally built an experimental platform for the visual inspection system.This paper studies the visual detection algorithm.The system selects Gaussian filter to remove the noise contained in the target image quickly and effectively,and uses the Otsu global threshold method to segment the image.Through the optimization of the geometric feature extraction,the effective criterion features for the defects are generated.Finally,a vision detection algorithm for gearbox assembly defects based on support vector machine is proposed in this paper.Combined these algorithms with SVM multi-level binary tree classification strategy,the system realizes the multi-category quick classification detection.On the basis of the above work,the visual inspection software is developed.Then,under the condition of the laboratory,the visual inspection system is tested andgets good results.At the end of this paper was some analysis for the influencing factor relevant with the testing results,as well as directions for further improvement.