Research On Intelligent Error-proof System Of Valve Parts Assembly Based On Machine Vision

In order to solve the problem of quality hidden danger caused by wrong and missing parts in the manual assembly process of the valve parts and the problem of manual error detection after assembly that reduces the production efficiency of the valve products,this paper relies on the research topic of the intelligent error prevention system for the assembly of the valve parts to study the error prevention technology in the assembly process of the valve parts.Based on the theory of error prevention and combined with the actual working condition analysis of valve parts assembly,this paper designs a ' double insurance ' error prevention scheme for real-time detection of error prevention after assembly with visual guidance before assembly.Among them,the visual guide error prevention adopts the parametric three-dimensional modeling method to model the parts.According to the existing two-dimensional written assembly manual and process regulations of the factory,the guide video of the assembly process of the valve is designed and produced,which improves the intuition and convenience of assembly guidance,reduces the burden of operators,and improves the quality and efficiency of valve assembly.The real-time detection and error prevention after assembly adopts the method of machine vision detection.The task is transformed into the acquisition and classification of assembly state images.The VGG16 model is migrated to the image classification task in this paper by using the migration learning method,and the accurate classification of assembly state images is realized.By replacing the first two full connection layers of the VGG16 migration model with the global average pooling layer,the parameters of the model are reduced and the network performance is improved.The software development platform under Python environment is used to realize the intelligent error prevention system of valve parts assembly.The accuracy and stability of the system are tested by simulating the working environment,and the system is improved and adjusted according to the experimental results.