Research On Intelligent Detection Technology Of Flat Workpiece Based On Deep Learning

As the foundation of the founding of the country,the instrument of rejuvenating the country,and the foundation of a strong country,the manufacturing industry plays an important role in China's economic construction.With the rise of artificial intelligence,the combination of artificial intelligence and manufacturing has become more and more close,which has promoted the transformation of traditional manufacturing to intelligent manufacturing.China's proposal of "Made in China 2025" is the only way to make a big country from manufacturing to a strong country.It will speed up the deep integration of the new generation of information detection technology and manufacturing,and help solve the problem of flexible inspection of workpieces for intelligent manufacturing.The traditional manual detection adaptability is capable of detecting various types of workpiece features,but the detection speed is slow,relying on manual experience,it is difficult to meet the requirements of online rapid detection;while the machine vision detection speed is fast,but it is difficult to adapt to various types.Problems with workpieces and different feature detection.If the advantages of manual detection and the advantages of machine vision rapid detection are combined,it is possible to realize the parallel collaborative processing of classification,positioning,feature extraction and surface defect detection of various types of workpieces,thereby realizing intelligent detection of workpieces.Due to its high flexibility and adaptability,machine vision inspection technology has been widely used in the field of intelligent manufacturing inspection.Based on the related literatures at home and abroad,this paper takes the flat workpiece of the explorer as the detection target,and constructs a deep detection model to design a new detection algorithm to realize the classification,positioning,feature extraction and surface defect detection of the workpiece.Provide technical support for intelligent inspection of workpieces.The main contents of its research are as follows:1.Based on the theory of machine vision acquisition system,the workpiece visual acquisition system is designed.The camera model is constructed,the design of the acquisition system is completed,and the hardware of the acquisition system is selected to provide a hardware support platform for workpiece detection image acquisition.2.In order to realize the automatic detection of workpieces,based on the research of deep learning model,this paper introduces the mainstream YOLO and Faster-RCNN models into the classification of workpieces and the detection of surface defects.Among them,the YOLO algorithm is used to learn the feature information in the input data for classification and localization.The Faster-RCNN algorithm is used to analyze and learn the input data layer by layer to realize the surface defect detection of the workpiece.On this basis,the problems existing in the workpiece detection of these two models are analyzed,which provides a model basis for the collaborative detection of workpiece classification,location,feature extraction and surface defects.3.In view of the shortcomings of the above two depth learning algorithms for detecting workpiece position information and workpiece feature extraction,this paper combines the advantages of Object-detection API and Mask-RCNN model to construct a new deep learning API-MASK model.And algorithm.The tensorflow deep learning framework is used to analyze and learn the input data hierarchically,and the high-level features obtained by the learning are used for the classification,positioning,feature extraction and surface defect detection of the workpiece,and the parallel collaborative processing of each detection task is realized to meet the intelligent detection of the workpiece.Demand.4.Apply the API-MASK model and algorithm to the flat workpiece for experimental verification.The experimental results show that the classification accuracy of the algorithm is 100%,and the defect detection accuracy is between 91% and 96%.Compared with other algorithms,the accuracy of the workpiece is relatively high in the classification,positioning,feature extraction and surface defect detection.It can realize parallel collaborative processing of each inspection task and has certain versatility.Figure 52,table 9,and 78 references.