Research And Application Of Intelligent Detection Method For Surface Quality Of Textile Flat Knitting Machine Inserts

As a needle bed operated by many types of knitting needles,the flat knitting machine insert is one of the important accessories on the flat knitting machine,and its quality determines the quality of the knitted products produced.At present,most textile enterprises still use manual sampling inspection to ensure the quality of inserts.The workload is heavy and complex,and it is impossible to carry out full-scale inspection according to the drawings,and cannot meet the actual production requirements of increasing demand and high quality requirements.In the context of the accelerated development of industrial intelligence and smart factories,this paper introduces machine vision technology to carry out research on the intelligent detection method and application of the surface quality of textile flat knitting machine inserts.The main contents are as follows:(1)Aiming at the problem of low measurement accuracy of insert size due to manual detection,an edge enhancement-based insert size measurement algorithm is proposed.Firstly,the image random noise is removed from the insert preprocessing,and the measurement area is segmented with the adaptive threshold algorithm;then,the extraction of insert size measurement points is enhanced based on the four-direction Canny edge detection method,and then the dimension measurement algorithm is used to detect inserts size parameters.Through experimental analysis,the measurement accuracy of the algorithm for the two sizes reached 0.022 mm and0.023 mm respectively,which is accurate and fast compared with the traditional size measurement method.(2)Aiming at the weak generalization ability and poor robustness of traditional feature extraction methods,a defect detection algorithm based on Res Net34 deep features is proposed.First,the deep features of image defects are extracted by the Res Net34 network;secondly,a multi-level SVM classification model is constructed based on the deep features;finally,it is verified on the NEU-DE test set.The results show that the recognition accuracy of the proposed algorithm reaches 97.67%,which meets the needs of practical applications.(3)According to the actual testing needs of textile machinery manufacturers,the intelligent testing equipment and system for the surface quality of textile flat knitting machine inserts are designed.It mainly includes system scheme design,software and hardware development and test verification,and realizes automatic intelligent detection of chip insertion.The test results show that the accuracy of the system has reached 99.23%,the false detection rate of the system is 0.77%,and the single-chip detection time is less than 5s.Compared with the manual sampling detection method,the system has been greatly improved in real-time and accuracy,which meets the actual needs production requirements.