Font Size: a A A

Research On Optimization Model Of Sewing Stitch Quality Based On Deep Learning

Posted on:2022-02-19Degree:MasterType:Thesis
Country:ChinaCandidate:Q X HuangFull Text:PDF
GTID:2481306494475544Subject:Electronic Science and Technology
Abstract/Summary:PDF Full Text Request
The quality of sewing stitch depends on the setting of sewing process parameters,and its quality affects the appearance of the final product.The setting of sewing process parameters depends on the experience of employees,and the traditional stitch quality is detected manually,which is difficult to meet the requirements of timeliness and accuracy in industry.Aiming at the difficulty of stitch quality detection,deep learning technology is used to realize the intelligent judgment of stitch quality,and regression analysis method is used to analyze the relationship between the top-thread tension data and the stitch quality data,and thus the optimization of the stitch quality is realized.The main research contents are as follows:(1)Under the premise that other sewing parameters are consistent,adjust the topthread tension value to sew the stitch sample.The image acquisition system is built to collect the stitch image,and the image enhancement technology is used to highlight the stitch features in the images.(2)The two indicators of stitch quality,stitch uniformity and stitch skew,are respectively quantified using two stitch characteristic parameters,namely the standard deviation of the stitch corner data and the slope of the straight line fitting of the stitch corner.The stitch corner is extracted from the stitch image to calculate the stitch characteristic parameters and the quality label of the stitch.(3)A data set composed of stitch image data and stitch quality label data is used for supervised training of the constructed convolutional neural network.The network parameter optimization model is updated continuously by back propagation,which realizes the intelligent judgment of the stitch quality.(4)The regression method is used to analyze the relationship between the topthread tension data and the stitch quality label data,and the corresponding top-thread tension value is obtained when the stitch quality is optimal,which realizes the optimization of the stitch quality.(5)Sewing experiments are carried out by using the top-thread tension value and its nearby value obtained from the regression model.The quality label of the stitches sewn under the condition of each top-thread tension value is predicted by the intelligent judgment model of stitch quality.By comparison,the top-thread tension value with the optimal stitch quality is consistent with the result obtained from the regression model,which verifies the feasibility and effectiveness of the method proposed in this paper.
Keywords/Search Tags:Stitch quality, Image processing, Feature extraction, Convolution neural network
PDF Full Text Request
Related items