IT202000027360A1 - APPARATUS AND METHOD FOR THE CONTINUOUS DETECTION OF DEFECTS IN A THREAD IN A SPINNING MACHINE - Google Patents

Description

DESCRIPTION

The present invention relates to the sector of textile fiber processing, and in particular to the sector of instruments and methods for detecting defects in the products of the preparation for spinning processes. In particular, the object of the present invention is a method and an apparatus for detecting tangles or knots, usually called "neps", in the thread produced in a ring spinning machine.

How ? known, a spinning wheel ? able to work spools of roving to obtain spools of thread, after having performed an ironing and a twisting on the roving. To this end, the spinning wheel ? consisting of a frame, which extends along a longitudinal axis and which supports a creel from which the roving bobbins are suspended, an ironing device supported by the frame, formed by a set of coupled cylinders having longitudinal extension, among which the roving passes to be stretched, and a platform carrying the spindles, lined up along the longitudinal axis, rotating around their own vertical axis, around which the stretched and twisted yarn is wound.

AND? It is well known that the imperfections present in the yarn cause unpleasant effects in the appearance of the fabric, especially in the presence of colored fabrics. For this reason, in the sector ? strongly felt? the need to monitor the? entity? and the frequency of imperfections, and in particular of tangles of fiber.

Currently, there are two industrial methodologies for detecting wire defects.

A first methodology involves analyzing in the laboratory a sample consisting of the thread of some spools stolen from the spinning machine, generally using capacitive sensors capable of detecting the variation in mass along the thread, so as to defining the type and frequency of tangles; this methodology? generally reliable, but does not allow you to intervene on the parameters of the processes to improve the quality? of the thread n? to understand the causes of the defects found, why? performed after finishing making the yarn. Eg, ? often used the USTER tester? 5-S800 tester, manufactured and marketed by Uster Technologies AG.

A second method involves the use of detection modules, based on capacitive or optical sensors, arranged in correspondence with the spindles for winding the yarn, equipped with knives capable of physically eliminating tangles from the yarn. Even then, isn't it? possible to trace the causes that generated the defect and intervene accordingly.

Finally, other methods are applied in the thread winding phase, a process that takes place downstream of the spinning machines; eg, ? popular Uster tester? Quantum 3, based on capacitive sensors, manufactured and marketed by Uster Technologies AG, and the YarnMaster Prisma tester, based on optical sensors, manufactured and marketed by Loepfe Brothers Ltd.

A first aim of the present invention? instead that of detecting the defects of a thread in a spinning frame, in order to modify the processing parameters of the spinning frame or of the other upstream machines or proceed with maintenance, in order to obtain a better quality thread.

The aforementioned methodologies are not suitable for this purpose, also because? use capacitive or optical sensors that require regular wire feed.

Between the drawing device and the winding device of the spinning frame, on the other hand, the thread undergoes strong oscillations, due to the winding and twisting actions which take place downstream.

Some studies also predict to use image capture via a video camera, and then process the images to detect tangles. For example, some methodologies are described in the following articles:

- Li Z, Pan R and Gao W. Formation of digital yarn black board using sequence images. Textile Research Journal.

2016; 86: 593-603 ;

- Eldessouki M, Ibrahim S and Militky J. A dynamic and robust image processing based method for measuring the yarn diameter and its variation. Textile Research Journal. 2014; 84: 1948-60;

- Ling C, Lianying Z, Li C and Xuanli Z. Digital image processing of cotton yarn seriplane. 2010 International Conference on Computer and Information Applications.

2010, p. 274-7;

- Li Z, Xiong N, Wang J, Pan R, Gao W and Zhang N. An intelligent computer method for automatic mosaic of sequential slub yarn images based on image processing. Textile Research Journal. 2018; 88:2854-66.

- Carvalho V, Soares F and Vasconcelos R. Artificial intelligence and image processing based techniques: A tool for yarns parameterization and fabrics prediction.

2009 IEEE Conference on Emerging Technologies & Factory Automation. 2009, p. 1-4.

However, it is a question of methodologies that are not suitable for the pre-established industrial purposes, since expect to use a pre-tensioned wire, which has a smooth feed.

Finally, some solutions are illustrated, for example, in the patent documents DE102018111648A1, WO2019130209A3, JP2018178282A and DE102016121662A1.

Furthermore, the solutions known today are not satisfactory also as regards reliability? of results. Often, in fact, some slight irregularities? of fibers, such as a slight enlargement of the fibers in one region, are incorrectly evaluated as tangles or other defects.

Purpose of the present invention? that of realizing a method and an apparatus for the detection of defects in a thread being processed in a spinning frame, which satisfies the needs of the sector and overcomes the drawbacks of which said with reference to the prior art.

That purpose? achieved by a method according to claim 1 and by an apparatus according to claim 18. The dependent claims identify further advantageous embodiments of the invention.

The characteristics and advantages of the method and apparatus according to the present invention will be evident from the following description, given by way of non-limiting example, in accordance with the figures of the attached tables, in which:

- figure 1 shows a spinning machine provided with a detection apparatus according to the present invention;

- figure 2 ? a diagram of the spinning frame of figure 1; - figures 3a and 3b show positive images of original neps with relative pixel profile;

- figure 4 illustrates a set of positive images of synthetic neps.

For clarity of exposition, in the following we will? reference to a method and apparatus for the detection of neps; however, it is understood that the invention? applicable to the detection of defects in general.

With reference to the attached tables, with 1 yes ? indicated as a whole a spinning machine of a spinning line for obtaining spools of thread from roving bobbins, extending along a longitudinal axis X.

The spinning frame 1 comprises a frame 2 for supporting the components, made in one or more? side by side parts, and a creel 4 supported by the frame 2.

The creel 4 comprises vertical columns 6 and a plurality? of longitudinal crosspieces 8, supported by the small columns 6 at a predefined height. Are the crosspieces 8 intended to support, when hung, a plurality? of coils of wick B.

Below the crosspieces 8, i.e. downstream of the bobbins B, the spinning machine 1 comprises a drawing device 10, supported by the frame 2.

The ironing device 10 comprises a plurality of lower drawing cylinders 12a-12d, generally three or four in number, motorized, which extend longitudinally, made in one piece or from several? traits, structurally distinct, side by side and aligned.

The ironing device 10 further comprises a plurality of of pressure arms 14, placed side by side longitudinally. Each pressure arm carries upper, idle pressure rollers 16.

By coupling with the pressure cylinders 12a-12d, the pressure rollers 16 form drawing pairs, through which the roving passes, which is drawn by effect of the speed? peripheral of each drafting pair, growing from upstream to downstream.

The spinning machine 1 also comprises a winding device 17, arranged below the drawing device 10, immediately downstream of the first drawing cylinder 12a.

The winding device 17 comprises a thread guide unit 18, comprising a support 20 connected to the frame 2, and a plurality of thread guide 22, supported by the support 20 and side by side longitudinally.

The winding device 17 also comprises a platform 23, located below the thread guide assembly 18, i.e. downstream of the thread guides 22. The platform 23 is? supported by frame 2 and ? vertically translatable with reciprocating motion.

The winding device 17 further comprises a plurality of spindles 24 placed side by side along the platform 23, each one rotatable about a respective vertical axis.

In normal operation of the spinning frame 1, the roving wound in a predefined bobbin B travels a first portion of the path to pass into the drafting device 10, from which it emerges stretched; the drawn sliver R travels a second stretch of path S2 between the drawing device and a respective spindle 24, passing through a respective yarn guide 22. The yarn obtained by drawing and twisting the sliver is wound on a tube fitted on the spindle, for formation of a spool of thread.

According to the present invention, a detection region 50 is defined, between the drafting device 10 and the winding device 17, crossed by the second stretch of path S2 of the drawn roving R. The drawn roving which passes through the detection region 50 ? subjected to a process for the detection of defects, and in particular for the detection of tangles or knots, generally called ?neps?.

The detection process comprises an acquisition phase during which images of at least one portion R* of the roving R in transit in the detection region 50 are acquired.

For this purpose, the spinning machine 1 comprises acquisition means 60, for example digital, for example comprising a video camera, suitable for acquiring images I of at least one section R* of the roving R in transit in the detection region 50.

Furthermore, the spinning machine 1 preferably comprises lighting means 62 suitable for illuminating, for example by means of a warm or cold light or, in a variant embodiment, by means of infrared rays, at least one zone of the detection region 50, comprising the portion R* of the R wick.

The detection process also comprises a processing step of the images I acquired by said acquisition means 60, performed by processing means. Said processing means 70 are configured to detect the neps through an adaptive or automatic learning detection algorithm, in particular of the Haar-cascade type preferably based on the Viola-Jones method. The latter ? illustrated in the article "Rapid object detection using a boosted cascade of simple features", presented at the Conference on Computer Vision and Pattern Recognition (2001) by Paul Viola and Michael Jones, whose teaching on the implementation of the algorithm ? expressly incorporated herein.

The learning of the detection algorithm ? based on a set of positive images Ip, ie images in which the stretch R* of drawn roving R has neps, and a set of negative images In, in which the stretch R* of drawn roving R does not have neps.

According to a first embodiment (defined ?with original neps?), starting from images of a learning drawn roving, for example acquired during the normal use of a predefined spinning machine, and therefore depicting sections with neps and sections without neps, the positive images Ip of the set of positive images are determined by processing a pixel profile for each image, in which said pixel profile ? obtained by summing the luminous pixels on each row of the image and selecting as positive images those for which at least one peak of the pixel profile exceeds a threshold value, for example determined by means of an average value and a standard deviation (figures 3a and 3b).

Furthermore, the positive images are preferably subjected to a further selection, eliminating the images in which the peak is ? given by single frayed fibers (hair); said further selection ? performed through an additional selection algorithm or manually.

According to a further embodiment (defined ?with synthetic neps?), the positive images Ip are digitally constructed and constituted by images of semicircles or semiellipses, preferably vertical (figure 4), for example different from each other in terms of the length of the minor axis and of the major axis.

According to an even further embodiment (called ?hybrid?), the positive images Ip comprise the positive images with original neps and the positive images with synthetic neps, ie they combine the two aforementioned embodiments.

According to a further embodiment of the invention, said processing means are configured to detect the neps through an adaptive or automatic learning detection algorithm of the Convolutional Neural Networks (CNN) and/or Recurrent Neural Networks (RNN) type.

In the normal functioning of the spinning machine, while the roving R ? in transit through the detection area 50, the acquisition means 60 continuously acquire the images I of a portion R* of the roving R.

Said images I are processed by the processing means 70 to detect the frequency of the neps and preferably their shape, as well as classifying them by type, through the automatic learning detection algorithm, in particular of the Haar-cascade type preferably based on the Viola-Jones method or through Convolutional Neural Networks (CNN) and/or Recurrent Neural Networks (RNN).

On the basis of these surveys, in the context of a management method of the spinning frame or of a spinning line including the spinning frame and machines upstream of this, such as carding machines, combing machines, draw-combiners and roving frames, one operates on the working of the spinning frame, such as for example twisting of the thread, drawing and/or pre-drawing of the roving, type and weight of the ring, hardness of the rubbers of the axles, gauge of the cylinders, pressure on the axles, speed? of production, in order to improve the quality? of the roving, or on the parameters of the processes carried out upstream of the spinning machine, such as for example torsion, stretching and/or pre-stretching, hardness of the rubbers of the axes, gauge of the cylinders, pressure on the axes, speed? of production in a roving frame, speed? of production and drafting group gauges in a drawing frame, waste percentage, number of strokes and drafting group gauges in a combing machine, speed? of production, neps and trash removal in a carding machine and in a battery machine, or maintenance interventions are carried out on the spinning frame or on said machines upstream of it.

Innovatively, the apparatus and the method for detecting defects in a spinning frame according to the present invention satisfy the needs of the sector and overcome the drawbacks of which said.

In fact, since the detection ? performed continuously on the roving being processed in the spinning frame, ? Is it possible to intervene to modify the processing parameters on the spinning frame or on the upstream machines or intervene with maintenance interventions, in order to improve the quality? of the thread.

Furthermore, advantageously, the present invention ensures good reliability? of the results, as it allows to discern in an industrially acceptable way between major defects and other irregularities? minors.

Furthermore, the tests carried out have shown a good correspondence between the detections obtained by means of the apparatus of the present invention and the tests based on the testers usually used today of which said in the introduction.

Advantageously, moreover, the processing of the images according to the invention ? very fast and allows you to detect defects continuously and intervene quickly to improve production.

AND? it is clear that a person skilled in the art, in order to satisfy contingent needs, could make modifications to the method and to the apparatus described above, all contained in the scope of protection as defined by the following claims.

Claims (20)

CLAIM 1. Method for detecting defects in a drawn roving (R) being processed in a spinning machine, comprising the step of acquiring images (I) of the drawn roving (R) in transit in a section (Sr) of a path (S2 ) and the digital processing step of said images (I) to detect said defects. 2. Method according to claim 1, wherein the processing step includes the step of detecting the frequency of defects along the roving (R). 3. Method according to claim 1 or 2, wherein the processing step includes the step of detecting the type of defects. 4. Method according to any one of the preceding claims, wherein the image processing step (I) provides for detecting the defects by means of an automatic learning detection algorithm. 5. Method according to claim 4, in which the learning of the detection algorithm ? based on a set of positive images (Ip) in which a portion of roving depicted has defects, and a set of negative images (In), in which a portion of stretched roving depicted does not present said defects. The method according to claim 5, wherein, starting from images of a learning drawn roving having flawed and flawless features, the positive images (Ip) are determined by processing a pixel profile for each image, wherein said pixel profile ? obtained by summing the bright pixels on each row of the image and selecting as positive images those for which at least one peak of the pixel profile exceeds a threshold value. 7. Method according to claim 6, wherein said threshold value ? determined by means of an average value of the profile and a standard deviation. The method according to claim 6 or 7, wherein said images of the learning roving are acquired during the normal use of a predefined spinning machine. 9. Method according to any one of claims 6 to 8, in which the positive images are subjected to a further selection, eliminating the images in which the peak is? given by individual frayed fibers (hair). 10. A method according to claim 9, wherein said further selection is performed through a further selection algorithm. 11. A method according to claim 9, wherein said further selection is performed manually. 12. Method according to any one of claims from 5 to 11, wherein the positive images (Ip) are digitally constructed and constituted by images of semicircles or semiellipses, preferably vertical, for example different from each other in the length of the minor axis and the? major axis. The method according to any one of claims 5 to 11, wherein - a part of the positive images (Ip) ? obtained through a self-learning detection algorithm; And - a remaining part of the positive images (Ip) ? obtained through digital construction. 14. Method according to any one of claims from 4 to 13, wherein the detection algorithm ? of the Haar-cascade type. 15. A method according to claim 14, wherein the detection algorithm of the Haar-cascade type ? based on the Viola-Jones method. 16. Method according to any one of claims from 4 to 13, wherein the algorithm ? of the type Convolutional Neural Networks (CNN) and/or Recurrent Neural Networks (RNN). 17. Method of operating a spinning frame (1) of a spinning line including: - a method for detecting defects in a drawn roving (R) according to any one of the preceding claims, e - a subsequent step of varying the processing parameters of the spinning machine or carrying out maintenance interventions on said spinning machine. 18. Management method of a spinning line comprising a spinning frame (1) and textile processing machines upstream of the spinning frame, comprising - a method for detecting defects on the spinning frame (1) according to any one of claims from 1 to 16, and - a subsequent step of varying the processing parameters of at least one of said machines upstream of the spinning frame or performed on at least one of said machines upstream of the spinning frame maintenance work. 19. Apparatus for detecting defects in a drawn roving (R) being processed in a spinning frame, comprising: - a sensing region (50) placed between a drawing device (10) and a winding device (17) of the spinning machine, crossed by a portion (Sr) of a path (S2) of the roving (R); - acquisition means (60) suitable for acquiring images (I) of the wick (R); - processing means (70), operatively connected to the acquisition means, for the digital processing of said images (I) and the detection of defects. 20. Apparatus for detecting defects in a drawn roving (R) being processed in a spinning frame, comprising: - acquisition means (60) suitable for acquiring images (I) of the drawn roving (R) from a detection region (50) placed between a drawing device (10) and a winding device (17) of the spinning machine, said region detector (50) being crossed by a section (Sr) of a path (S2) of the roving (R); - processing means (70), operatively connected to the acquisition means, for the digital processing of said images (I) and the detection of defects; said processing means (70) being configured to detect defects by means of an automatic learning detection algorithm.