Research On The Intelligent Optimization Processing System Of Solid Wood Panels

China is a large consumer and exporter of wood products,but has limited forest resources and relies heavily on imported timber.The domestic wood processing industry depends too much on manual labor,resulting in inconsistent product quality and low utilization of timber resources,informationization and intelligence.This makes it hard to satisfy the increasing demand for highquality household and handicraft products and other necessities.Therefore,the government has proposed a plan to integrate information technology with industrialization,promoting the transformation and upgrading of the wood processing industry towards intelligent manufacturing,achieving automation and intelligence in wood processing and manufacturing.This paper integrates machine vision and artificial intelligence technology to focus on the detection for surface color characteristics and local defect characteristics of solid wood panels,a set of intelligent optimization processing system has been developed.The main research contents are as follows.(1)A reliable image acquisition system of solid wood panels was built.Through the secondary development of the DALSA industrial camera,using the line scanning way,with the trigger and frame merge method to realize online acquisition,the double-sided image acquisition of long solid wood panels was realized.The double-sided images of Chinese fir,beech,cherry,ash and pine were collected,and the foundation for the subsequent intelligent optimization processing was extablished.(2)A color sorting and color space correction model of solid wood panels was established.Aiming at the problem that the color difference between similar solid wood panels was not obvious and the sorting index was not clear,an extreme gradient boosting classification tree model for solid wood panels sorting was proposed.By screening the color features extracted in different color spaces,the first-order moments of a,B and H channels were used to classify the three colors,namely light,medium and dark,of solid wood panels.The accuracy rate reached 97.22%.Aiming at the problem of quantifying the color depth of solid wood panels,a gradient regression tree model for color depth regression of solid wood panels was proposed,which modified the existing color model,quantified the color depth value of solid wood panels,and further improved the color sorting effect.(3)An intelligent calculation method of processing coordinates based on double-sided detection of solid wood panel defects was proposed.Aiming at the problem of double-sided defect positioning during the processing of solid wood panels and the coordinate conversion between positioning and actual processing,a double-sided defect detection algorithm for solid wood panels based on deep learning YOLOv7 was established to detect the defects on the upper and lower surfaces of solid wood panels.The detection accuracy of the fully trained model reached 0.94.Taking the thickness of the saw blade and the minimum processing length into account,the processing coordinates of the solid wood panels were calculated and determined through image simulation,which improved the optimization efficiency and increased the effective yield of the solid wood panels by 12.31%.(4)A set of hardware-compatible intelligent processing system software for solid wood panels was developed.The software core of the solid wood panels processing system was built based on Qt software and Twin CAT3 software.It realized the visual human-computer interaction interface of color sorting,defect detection,double-sided processing coordinate calculation,and sent the processing coordinates to the Twin CAT3 software by controlling PLC through ADS communication;realized the lateral truncation processing motion control of solid wood panels in Twin CAT3 software,and performed the sawing action according to the intelligent optimized processing coordinates,and improved the comprehensive utilization efficiency of solid wood panels.