Research On Intelligent Monitoring Of Laser Assisted Machining Based On Machine Vision

Laser assisted machining cutting technology has great potential in processing difficult-to-machine materials.The technology uses laser heating to soften highhardness materials,improve the plasticity of brittle materials,and improve the machinability of local workpieces,thereby improving processing quality and processing efficiency.However,the laser assisted machining cutting technology has encountered problems in engineering applications such as poor automation,high difficulty in matching the laser and processing parameters,and difficult control of the processing process due to the addition of lasers.This paper proposes the use of machining state monitoring methods to improve the degree of automation,avoid machining failures,analyze the tool state,and conduct systematic theoretical and experimental research.The main research work carried out is as follows:Based on machine vision technology,the internal and external parameters of the camera are calibrated for the CCD industrial camera and the self-made checkerboard calibration block;the CCD industrial camera is used to capture the image,and the geometric edge of the cylindrical workpiece is fitted on the image;the laser pointer is collected to incident on the cylinder The image on the surface of the workpiece uses various laser spot recognition algorithms and the improved spot fitting algorithm in this paper to identify and locate the collected images,measure the distance between the geometric centroid of the indicated spot and the axial and radial edges of the workpiece,and test different algorithms The obtained geometrical quantity;by controlling the mechanical arm to clamp the laser transmitter to move the initial incident position of the indicating spot to the specified standard position,and testing the spot fitting positioning error.The laser assisted turning experiment of two materials of superalloy GH4169 and cast iron QT500 was carried out.The image of the chip and the wear of the tool during the processing were obtained by the camera,and the images were marked and classified;the convolution in the field of deep learning image processing The basic structure of the neural network and the technology related to network parameter optimization,use Python language and Tensor Flow deep learning library to build the Yolov3-TC network model designed in this article,and input the above-mentioned chip image and turning tool wear image into the network for training;finally,with Faster-RCNN,SSD,Yolov3 and other networks compare training and testing to evaluate and verify the effectiveness of the improved network model in this paper.Based on the MFC framework of Visual Studio 2015 development tools,the laser assisted turning intelligent detection software was developed to integrate the above detection functions;including camera parameter setting,camera calibration setting,robotic arm control,spot monitoring,chip monitoring and tool wear detection Six modules;design,implement and optimize the functions required by each module of the software.