Research On Detection System Of Melt Direct Spinning Filament Broken Based On Machine Vision

As an important raw material for the production of industrial textiles,melt directspun filaments are widely used in production and life.In order to facilitate the storage and transportation of melt direct-spun filaments,chemical fiber companies usually wind them into chemical fiber cakes.During the winding process,due to the unstable tension of the godet roller and the uneven local force at the hook,the melt direct spinning filament will break and the filament will break locally.The company regards the severity of local breakage of melt direct-spun filaments as one of the main indicators for judging the quality of chemical fiber cakes.At present,due to the lack of relevant research on the automatic detection of filament breakage in melt direct spinning,the chemical fiber production line mainly relies on manual method for filament breakage detection.In recent years,with the continuous increase of labor costs and the continuous expansion of the scale of chemical fiber production,the disadvantages of manual detection methods in terms of detection efficiency and detection accuracy have become increasingly prominent.In order to realize the automatic inspection of melt direct spinning filament breakage and improve the production efficiency of enterprises,this paper researches and designs a set of special melt direct spinning filament breakage automatic detection system based on the theoretical knowledge of machine vision.The main work content of the subject includes the following aspects:(1)Design of visual inspection device for melt direct spinning filament breakage.According to the actual production environment of the workshop,the sliding track-type traveling trolley is selected as the basic platform of the detection device;considering the characteristics of the slender and difficult to directly collect and identify the broken ends of the melt direct-spinning filament,a method with two different views is proposed.The dual-camera structure of the industrial cameras working in cooperation with the industrial camera.This structure realizes the efficient inspection and classification of the broken ends of the melt direct-spun filament through the detection process of collection,positioning,re-collection,recognition and classification;Finally,refer to the visual environment in the actual environment.Field parameters complete the selection of servo system hardware and visual inspection system hardware.(2)A collaborative optimization algorithm based on template matching algorithm and K-means unsupervised clustering algorithm is proposed to realize fast statistics and precise positioning of melt direct-spun filament hooks.After acquiring the overall image of the filament hook of a single station through the short focal length and large field of view industrial camera,the collected image is first divided into blocks and the hook matching template is selected,and the template matching algorithm is used to quickly match the points with a higher degree of matching.Coordinate retention;then the Kmeans unsupervised clustering algorithm is used to obtain the cluster nucleus coordinates and the number of nuclei;Finally the cluster nucleus coordinates and the number of cluster nuclei are respectively output as the coordinates and number of the melt directspun filament hooks.(3)A new method for detecting and classifying filament breakage in melt direct spinning based on Hough transform and RBF neural network is proposed to further detect and classify filament breakage.After obtaining a clear image of melt direct-spun filaments,the algorithm first uses wavelet soft threshold smoothing and threshold segmentation algorithms to perform image enhancement processing to eliminate the interference of background noise points on the filament body and improve the credibility of the extracted feature information;Then through the Hough transform to achieve the shape segmentation of the filament body and the broken part,as a pavement for the subsequent shape feature extraction;Finally based on the extracted multiple shape features,the radial basis function neural network model is established and trained to achieve the melt straight detection and classification of spun filament broken ends.In order to verify the feasibility of the system algorithm in this paper,Intel Core i7-10750 H CPU,industrial computer under Win10 operating system is used as the hardware platform,and the algorithm calculation and parameter analysis are carried out based on the MATLAB R2018 b software platform.The test results show that the system can quickly and accurately realize the image collection of melt direct-spinning filaments,remove image background noise,realize the segmentation of filament trunk and broken ends,and efficiently identify and classify melt direct-spinning filament broken ends.The real-time performance and accuracy provide a theoretical reference for future detection of filament breakage in melt direct spinning.