Research On Key Technologies Of Visual Inspection And Identification Of Assembly Elementray Unit For Chassis

Chassis is one of the key accessories of high-tech products such as computers and intelligent electronic equipment.At present,the inspection standards and inspection methods of chassis assembly quality seriously lag behind the development and market demand of the chassis equipment industry.This paper is titled “Research on the Key Technologies of Visual Inspection and Identification of Assembly Elementray Unit for Chassis”,systematically analyzing the image feature extraction and classification mechanism of Chassis Assembly Elementary Unit(AEU),focusing on the deepening of model depthization chassis AEU image classification network technology,Based on model lightweight AEU fast image classification network technology and intelligent identification technology based on image classification network chassis assembly quality,developed an intelligent identification system for chassis assembly quality optimized image classification network.The project has great academic value and practical significance to promote the development of intelligent inspection and instrumentation of manufacturing engineering.The research work was supported by the Guangzhou City Science and Technology Plan Project(No.201802030006).The paper studies the visual inspection and identification system of chassis assembly elementary unit,and summarizes the research progress at home and abroad from three aspects: classical image classification method,visual inspection image classification method based on deep learning,and component quality identification technology for visual inspection.The main point of this paper are as follows:? Analyze the requirements of chassis AEU intelligent visual inspection and identification system,covering image feature extraction,image classification,size extraction,AEU assembly identification and other categories.Design the overall architecture and process of the intelligent visual inspection and identification system,including data and production,image classification network,chassis standard dictionary,coordinate positioning,quality identification,monitoring data collection and other modules.Complete chassis AEU rapid and accurate classification,chassis assembly quality identification,Chassis detection information management and other work.? Design a model depthization chassis AEU image classification network structure,analyze chassis AEU image feature extraction and classification mechanism,make chassis AEU dataset,and solve the problem of chassis AEU data category imbalance through oversampling.Analyze the chassis feature extraction process and characteristics,design based on the deep network chassis AEU image classification method,integrate high-dimensional and low-dimensional image features,improve image classification effects,use Exponential Linear Units(ELU)activation function,identity mapping,modular convolutional structure design chassis AEU image classification network structure which can effectively solve the problems of model degradation and gradient descent,achieve accurate classification of chassis AEU images,and improve the accuracy of Top-1 classification.? Design a model lightweight chassis AEU image classification network structure,according to deep network image feature extraction mechanism,further study a variety of chassis AEU fast image classification mechanism,according to(37)(37)? convolution acceleration principle,through the Winograd algorithm to accelerate the convolution And the use of 1×1 small convolution kernel structure further reduces the amount of model parameters,which effectively reduces the time and space complexity of the model,realizes the rapid and accurate classification of chassis AEU images,reduces the calculation amount,and reduces the detect time.? According to the requirements of chassis AEU intelligent visual detection and identification system,discuss the ideas and standards of chassis assembly quality intelligent identification,and according to the chassis assembly quality detection standards,put forward the chassis assembly quality identification basis table and identification process.Study the method of extracting AEU size information of the coordinate positioning chassis,use Matlab to complete the calibration,correct the camera distortion,realize the conversion of AEU image coordinates to world coordinates,and correct the errors caused by the shooting angle,design and optimize the intelligent identification method of the chassis assembly quality of the image classification network,effectively reduce the detection error,and improve the detection rate and classification accuracy of chassis AEU.? Carry out the construction of chassis AEU visual inspection and identification platform to test and verify the platform detection function and versatility.Experiments were conducted on the two aspects of chassis AEU image classification and chassis assembly quality detection and identification that optimize the image classification network,and comprehensively evaluate the application effect of chassis AEU visual detection and identification platform.