Method and device for classifying an event in an elongate textile test specimen

Abstract

In this computer-implemented method for classifying an event in an elongate textile test specimen, at least one measured value of at least one event parameter for the event is measured (901). At least one reference dataset, containing, for a respective reference event, at least one reference value of the at least one event parameter and a classification of the reference event, is provided (902). The at least one measured value is compared (904) with the at least one reference value of all of the reference datasets. The event is classified (908) in accordance with the classification of the reference event whose at least one reference value is most similar to the at least one measured value in accordance with a predefined similarity criterion. The method achieves high quality and, at the same time, high productivity.

Description

[0001] Method and device for classifying an event in an elongated textile test piece

[0002] FIELD OF EXPERTISE

[0003] The present invention lies in the field of textile quality control. It relates to a computer-implemented method and a device for classifying an event in an elongated textile test specimen. The invention also relates to a yarn clearing system and a yarn processing machine. It is preferably, but not exclusively, used in yarn clearers on spinning or dishwashing machines.

[0004] Furthermore, the invention relates to a computer-implemented method for providing at least one reference data set for the method or device according to the invention for classifying an event. Finally, the invention also relates to a computer-readable medium on which at least one reference data set provided according to the method is stored.

[0005] STATE OF THE ART

[0006] To ensure yarn quality, so-called yarn clearers are used on spinning or dishwashing machines. A yarn clearer contains a measuring head with at least one sensor that scans the moving yarn. Commonly used sensor principles are capacitive and optical, both of which are described in WO-2012 / 051730 A1. The purpose of scanning is to detect events in the yarn.

[0007] In this document, "events" in an elongated textile test piece are defined as locations along the longitudinal direction (i.e., locations with a finite length of usually less than 1 m) where at least one specific measured value deviates from a corresponding target value. When the elongated textile test piece is moved along its longitudinal direction, an event passes a fixed point, e.g., a sensor, within a finite time interval (usually less than 1 s). Examples of events include thick spots, thin spots, or foreign matter in a yarn.

[0008] Measurement results obtained from the sensor signal are continuously evaluated against predefined criteria, such as a cleaning limit. If a yarn defect lies below the cleaning limit, it is tolerable; if it lies above the cleaning limit, it is an intolerable yarn defect that must be removed from the yarn or at least recorded. The basis of yarn clearing is therefore a classification system with two classes: tolerable and intolerable yarn defects.

[0009] A method for setting a cleaning limit is disclosed in WO-2011 / 038524 A1. First, a statistical representation of the yarn is determined by measurements on the yarn. Based on the statistical representation, the cleaning limit is calculated and suggested for application. A length-related number of inadmissible events expected with this cleaning limit is calculated and output. An operator can comment on the expected number of inadmissible events, after which the cleaning limit is automatically set according to the comment.

[0010] The yarn events can be plotted as points in a two-dimensional classification scheme, in which the defect length is typically plotted along the abscissa and the defect size (deviation of the mass per yarn length, the yarn cross-section, the yarn reflectivity, etc. from a target value) is plotted along the ordinate. Examples of a cleaning limit and a classification scheme are also given in WO-2011 / 038524 A1.

[0011] According to EP-0 685 580 A1, yarn defects are recorded, entered into the classification scheme, and counted in each class. This creates a defect pattern in the classification scheme. This defect pattern is compared with predefined model reference patterns that allow conclusions to be drawn about the causes of the detected defects. The predefined reference patterns can be determined through prior testing or from experience. According to WO-2010 / 078665 A1, event densities in the classification scheme are determined from the defect size and length. In the classification scheme, a yarn body is represented as a surface. The surface is bounded by the abscissa, the ordinate, and a line in the event field that follows a constant event density. The events in the yarn body can be considered statistical "noise" belonging to the actual yarn and should not be removed from the yarn.

[0012] WO-2013 / 185246 A1 also teaches the representation of a first density line in the classification scheme, which refers to a first test sample and follows a constant event density. In addition to the first density line, a reference density line is represented, which follows the same event density as the first density line but refers at least partially to a reference test sample that is different from the first test sample. This enables a comparison of the courses of the first density line and the reference density line. From the comparison of the courses of the density lines, a comparison of the qualities of the first test sample and the reference test sample can be made.

[0013] WO-00 / 73189 A1 aims to enable improved, simplified, and rapid adjustment of the cleaning limit so that its effect on the final product can be more accurately predicted. For this purpose, representations of defects in the final product, e.g., in the yarn, are to be generated based on the cleaning limit, which visualize the effect of the defects in the final product.

[0014] Spinning practice has shown that even with a supposedly optimally set cleaning limit, yarn clearing can sometimes be unsatisfactory. On the one hand, after yarn clearing, defects remain in the yarn which, although below the cleaning limit, are nevertheless perceived as disruptive. To remove these defects, the cleaning limit could be "tightened", i.e. shifted downwards. This would, however, lead to the unnecessary removal of tolerable yarn defects and thus reduce productivity. On the other hand, yarn clearing may remove defects from the yarn which, on closer inspection, are not at all disruptive. To prevent this, the cleaning limit could be "softened", i.e. shifted upwards. This would, however, result in other, disruptive yarn defects not being removed and thus reduce quality.

[0015] WO-97 / 31262 A1 deals with the electronic simulation of fabric patterns. It proposes imaging yarn sections with a video camera, digitally storing the images, and interweaving them to obtain a realistic image of the resulting fabric. The simulated image can be output via a screen or printer.

[0016] PRESENTATION OF THE INVENTION

[0017] It is an object of the present invention to provide a method and a device for classifying an event in an elongated textile test piece moved along its longitudinal direction, which reduce or eliminate the disadvantages of the prior art. In particular, as many disruptive yarn defects as possible are to be classified as such without unnecessarily classifying permissible yarn events as also disruptive, and as many permissible yarn events as possible are to be classified as such without classifying disruptive yarn defects as permissible. This is intended to achieve high quality and, at the same time, high productivity.

[0018] A further object of the invention is to provide a method and a yarn clearing system for yarn clearing that reduce or eliminate the disadvantages of the prior art. In particular, the productivity of the yarn clearing process and the quality of the cleared yarn are to be increased.

[0019] Furthermore, a method for providing at least one reference data set for the method or device according to the invention is to be specified.

[0020] These and other objects are achieved by the inventive method and device, as defined in the independent patent claims. A method and a yarn clearing system for yarn clearing, as well as advantageous embodiments, are specified in the dependent patent claims. The invention is based on the consideration that the metrological assessment of a yarn event does not always correspond to that of the human eye. The algorithms previously used for yarn clearing, which compared the measurement results of physical measurements on the yarn with a predetermined clearing limit, delivered satisfactory results in many cases, but could not handle the exceptions described above. In many cases, only a human can conclusively assess whether a yarn event is perceived as disturbing or not. After all, the final textile product made from the yarn is also viewed by humans.

[0021] Based on this consideration, the invention achieves these objectives by having the yarn event assessed and classified by an operator. Furthermore, the yarn event is measured using a sensor, as is known from the prior art. The classification is assigned to the measurement results associated with the yarn event and stored together with them as a reference data set. This process is preferably repeated for several different yarn events, thus creating and providing a database of reference data sets.

[0022] The computer-implemented method according to the invention serves to classify an event in an elongated textile test piece. It comprises the following steps: a) measuring at least one measured value of at least one event parameter for the event; b) providing at least one reference data set, which contains at least one reference value of the at least one event parameter and a classification of the reference event for each reference event; c) comparing the at least one measured value with the at least one reference value of all reference data sets; and d) classifying the event according to the classification of the reference event whose at least one reference value is most similar to the at least one measured value according to a predetermined similarity criterion. In one embodiment, the at least one event parameter is an event variable, e.g.a deviation of a length-related mass density, cross-section, or reflectivity of the test material from a target value, and the at least one reference value and the at least one measured value are each a signal consisting of a plurality of values ​​of the event size as a function of a length position or time. In this case, the similarity criterion can include a cross-correlation of the two signals.

[0023] In another embodiment, the at least one event parameter comprises an event size, e.g., a deviation of a length-related mass density, a cross-section, or a reflectivity of the test object from a target value, and an event length. In this case, the similarity criterion can include a distance between a point representing the event and a point representing the reference event in a coordinate system spanned by the event length and the event size.

[0024] In one embodiment, several different event parameters are considered. At least one first event parameter can be measured capacitively, for example, and at least one second event parameter can be measured optically, for example.

[0025] In one embodiment, the measured value of the at least one event parameter is measured while the elongated textile test material is moved along its longitudinal direction.

[0026] The at least one reference data set can be provided by measuring the at least one reference value for each reference event, visually assessing and classifying the reference event by an operator, and assigning the classification to the at least one reference value and storing it together with the reference value as a reference data set. The reference event is preferably visually assessed by taking an image of the reference event and presenting the image to the operator for visual assessment. In one embodiment, classification takes place in a classification system with exactly two classes, namely a first class with permissible events and a second class with impermissible events.

[0027] In the computer-implemented method according to the invention for cleaning yarn moving along its longitudinal direction, events in the yarn are classified into exactly the two classes described above, and events classified in the second class are removed from the yarn.

[0028] The device according to the invention for classifying an event in an elongated textile test material comprises a) a measuring device for measuring at least one measured value of at least one event parameter for the event; and a computer system with b) a memory for storing at least one reference data set, which contains at least one reference value of the at least one event parameter and a classification of the reference event for each reference event; c) a processor configured to compare the at least one measured value with the at least one reference value of all reference data sets; d) a processor configured to classify the event according to the classification of the reference event whose at least one reference value is most similar to the at least one measured value according to a predetermined similarity criterion.

[0029] In one embodiment of the device, the at least one event parameter is an event variable, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material from a target value, the at least one reference value and the at least one measured value are each a signal consisting of a plurality of values ​​of the event variable as a function of a length position or of time, and the measuring device is configured to record the signal. The at least one event parameter can comprise an event variable, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material from a target value, and an event length. In one embodiment, the measuring device is configured to measure several different event parameters. The measuring device can, for example,be configured to capacitively measure at least one first event parameter and to optically measure at least one second event parameter.

[0030] In one embodiment, the processor is configured for classifying in a classifying system with exactly two classes, namely a first class with permissible events and a second class with impermissible events.

[0031] The yarn clearing system according to the invention for cleaning yarn moving along its longitudinal direction includes a device with a processor for classifying in a classifying system with the precisely two classes described above and a cutting device that can be triggered by the computer system for cutting the yarn as a result of a classifying event in the second class.

[0032] The yarn processing machine according to the invention, for example a dishwasher or spinning machine, includes a plurality of yarn processing stations, at each of which yarn is wound onto a bobbin, and the yarn clearing system according to the invention described above.

[0033] In one embodiment, the yarn processing machine according to the invention includes at least one camera that is set up at one of the yarn processing stations to record images of events in the yarn, an output unit connected to the at least one camera for outputting an image recorded by the camera, and an input unit connected to the processor for inputting a classification of the output image.

[0034] According to the computer-implemented method according to the invention for providing at least one reference data set for the method according to the invention or the device according to the invention, at least one reference value of at least one event parameter for the reference event is measured for each reference event in an elongated textile test material, the reference event is visually assessed and classified by an operator, and the classification is assigned to the at least one reference value and stored together with the at least one reference value as a reference data set.

[0035] In one embodiment, the reference event is visually assessed by capturing an image of the reference event and presenting the image to the operator for visual assessment.

[0036] The invention also relates to a computer-readable medium on which at least one reference data set provided according to the invention is stored.

[0037] One advantage of the invention is that events in the test material are classified according to human perception rather than abstract parameters. This helps ensure that the final textile product produced from the test material is free of defects that would be perceived as disturbing by the human eye. During yarn cleaning, all defects perceived as disturbing are removed from the yarn, guaranteeing high yarn quality. At the same time, the removal of non-disturbing yarn events is avoided, increasing productivity.

[0038] LIST OF DRAWINGS

[0039] The invention is explained in detail below with reference to the drawings. Some embodiments relate to a yarn cleaner at the winding station of a dishwashing machine. This is not intended, however, to limit the generality of the invention. Yarn cleaners are used not only in dishwashing machines, but also in spinning machines. The invention is not limited to the application of yarn cleaning. It is also suitable for other elongated textile test materials such as sliver. Figure 1 shows a schematic of a yarn winding machine with a yarn cleaner system. Figure 2 shows (a) a fictitious measurement signal and (b) a fictitious reference signal in

[0040] Dependence of the length position.

[0041] Figure 3 shows (a) a fictitious mass signal and (b) a fictitious diameter signal as a function of length position. Figure 4 shows a diagram illustrating a similarity criterion for the two signals from Figure 3.

[0042] Figures 5-7 show (a) four different measurement signals for one and the same yarn event and (b) a photograph of the yarn event.

[0043] Figure 8 shows a schematic representation of an event field with a yarn body, a cleaning curve and various reference events.

[0044] Figure 9 shows a flow chart of an embodiment of the inventive method for classifying an event.

[0045] Figure 10 shows a flowchart of an embodiment of the method according to the invention for providing at least one reference data set.

[0046] Figure 11 shows a schematic table of a database with reference data sets.

[0047] IMPLEMENTATION OF THE INVENTION

[0048] Figure 1 shows a highly schematic illustration of a yarn winding machine 100 according to the invention with multiple winding stations 101. At each winding station 101, yarn 110 is rewound from a bobbin 111 onto a cheese 112 during the rewinding process. In this example, the yarn 110 moves along its longitudinal direction from bottom to top, as indicated by arrows 113. A device 120 according to the invention is installed in the dishwasher 100.

[0049] At each winding station 101, the yarn 110 is monitored by a measuring head 121 of the device 120 according to the invention. The measuring head 121 measures at least one measured value of at least one event parameter for events in the yarn 110. For this purpose, it can be equipped with a capacitive, optical, and / or other sensor as well as an evaluation unit for evaluating the sensor signals. Such measuring heads 121 are known per se and need not be explained further here.

[0050] Preferably, a cutting device is also installed at each winding station 101. This device cuts yarn sections with unacceptable defects from the yarn 110. The cutting device receives the corresponding cutting commands from the evaluation unit of the measuring head. The cutting device is preferably integrated into the measuring head 121.

[0051] All measuring heads 121 are connected via a data line 122 to a central control unit 123 of the device 120 according to the invention. The measuring heads 121 are adjusted and controlled by the control unit 123, and they transmit data such as the measured values ​​to the control unit 123. The data line 122 can be designed, for example, as a serial bus. The control unit 123 is equipped with a central computer 124, which serves, on the one hand, to collect and further process the measurement data originating from the measuring heads 121 and, on the other hand, to adjust the measuring heads 121. The central computer 124 is connected to or contains a database 125. Furthermore, the control unit 123 has an output unit 126 and an input unit 127 for an operator. The output unit 126 can be designed, for example, as a screen and / or printer. The input unit 127 can, for example, contain a keyboard or a computer mouse.The output unit 126 and the input unit 127 can be configured together as a touchscreen. The output unit 126 and the input unit 127 can be configured as a mobile terminal that is wirelessly connected to the central computer 124.

[0052] Figure 2(a) shows a fictitious measurement signal 210 for a yarn event, as it could be measured by a measuring head 121. The measurement signal 210 is composed of many measured values ​​of an event parameter AM (vertical axis 202) recorded along a length position L (horizontal axis 201) of the yarn 110. The event parameter AM can, for example, be a capacitively measured deviation of a length-related mass density M of the yarn 110 from a target value. Another example of an event parameter is an optically measured deviation of a yarn diameter from a target value. The target value is preferably determined by continuously calculating the average value over a large number of measurements. The measured values ​​can, for example, be recorded every 2 mm and combined to form the measurement signal 210 according to Figure 2(a).

[0053] In a representation analogous to Figure 2(a), Figure 2(b) shows a fictitious reference signal 220 for a reference event. The reference signal 220 is in turn composed of many reference values ​​of the same event parameter AM, recorded along the length position L of a reference yarn. The reference yarn can be different or the same as the one for which the measurement signal 210 of Figure 2(a) was recorded. If it is a different yarn, it should nevertheless have similar properties to enable a meaningful classification of the yarn events; in particular, it should be made of the same material, have the same yarn count, and be produced using the same spinning process.

[0054] Of course, the signals 210, 220 could be represented as a function of time instead of the length position L. Each of the two signals 210, 220 can easily be converted from length dependence to time dependence and vice versa, given knowledge of the instantaneous speed of the yarn.

[0055] In the method according to the invention, which is fully discussed in connection with Figure 9, the measurement signal 210 is compared with the reference signal 220. The two signals 210, 220 can be compared, for example, using the cross-correlation function, which is known per se to those skilled in the art. The maximum of the cross-correlation function is a measure of the similarity of the two signals 210, 220: the larger the value, the more similar. Furthermore, those skilled in the art will be familiar with other methods for comparing two signals 210, 220.

[0056] Another approach for comparing the measurement signal 210 with the reference signal 220 takes advantage of the relatively simple basic shape of the yarn events, which typically represents a "peak" or a "valley." It reduces the two signals 210, 220 to a value of two event parameters each. A first event parameter is an average event magnitude AMM or AMR, e.g., an average of all signal values ​​that reach at least 80% of the signal maximum. A second event parameter is an event length LM or LR, e.g., the length of the event at 50% of the average event height. Thus, each event can be represented by a two-dimensional vector (LM, AMM) or (LR, AMR). The length of the difference vector is a measure of the similarity of the two events: the shorter, the more similar. This comparison method can be generalized to more than two dimensions (event parameters).Figure 3 shows two different fictitious measurement signals 310, 320 for one and the same yarn event as a function of their length position L (horizontal axis 301). A first measurement signal 310 according to Figure 3(a) can consist of measured values ​​of the deviation AM (vertical axis 302) of a length-related mass density M of the yarn from a target value, as in Figure 2(a). A second measurement signal according to Figure 3(b) can consist of measured values ​​of the deviation AD (vertical axis 303) of a yarn cross-section D from a target value.

[0057] Each of the two measurement signals 310, 320 can be individually compared with a corresponding (not shown) reference signal, e.g., using the cross-correlation function as discussed above. As illustrated in the diagram in Figure 4, the two maxima R (e.g., normalized to the value 1) can then be ma x(AM) (horizontal axis 401) or R max(AD) (vertical axis 402) of the cross-correlation function can be entered as the coordinates of a point 410 in a two-dimensional coordinate system. A distance 411 of point 410 from the ideal corner point P(l, 1) is a measure of the similarity of the two measurement signals 310, 320 of Figure 3 with the corresponding reference signals: the shorter, the more similar.

[0058] Alternatively, each of the two measurement signals 310, 320 of Figure 3 can be reduced to a few measured values ​​of a few event parameters, as discussed above with reference to Figure 2. Each yarn event can then be represented, for example, by a four-dimensional vector whose components are the event size and event length determined from the two measurement signals 310, 320. The length of the difference vector between the measurement vector and the reference vector is a measure of the similarity of the two events: the shorter, the more similar.

[0059] As a supplement to the fictitious measurement signals 210, 310, 320 of Figures 2 and 3, Figures 5-7 refer to real game events, which can be events to be classified or reference events. Partial figures 5(a), 6(a), and 7(a) each show real measurement signals of four event variables (vertical axes 502, 602, 702) as a function of their longitudinal position or time (horizontal axes 501, 601, 701). The four measurement signals are as follows: • Solid line: measurement signal 511, 611, 711 of a capacitive sensor for mass deviations;

[0060] • finely dashed line: inverted measuring signal 512, 612, 712 of an optical sensor for diameter deviations that measures light transmitted past the yarn;

[0061] • dotted line: measuring signal 513, 613, 713 of an optical sensor for foreign matter that measures green light reflected from the yarn;

[0062] • coarsely dashed line: measuring signal 514, 614, 714 of an optical sensor for foreign matter that measures red light reflected from the yarn.

[0063] In the method according to the invention, any of the four measurement signals, a combination of any two or three measurement signals or all four measurement signals can be taken into account.

[0064] Sub-Figures 5(b), 6(b), and 7(b) each show an image 520, 620, 720, namely a photograph, of the corresponding yarn event 540, 640, 740 in a yarn 530, 630, 730, but not necessarily at the same scale as in sub-Figures 5(a), 6(a), 7(a) and possibly shifted along the length axis 501, 601, 701 relative to the measurement signals. The vertical lines in the background merely represent a scale (10 mm) on a yarn chart used.

[0065] Table 1 gives an overview of the three game events 540, 640, 740 and some of their measured values.

[0066] Table 1

[0067] For the nep 540 of Figure 5 and the short thick spot 640 of Figure 6, the mass signals 511, 611 and the diameter signals 512, 612 each show significant deviations, while the two foreign matter signals 513, 514, 613, 614 exhibit rather insignificant deflections. The opposite is true for the foreign matter fiber 740 of Figure 7.

[0068] An advantage of the present invention will now be illustrated using the example of Figure 7. Assume that the method according to the invention is carried out only with the diameter signal 712 and the other three signals 711, 713, 714 are not available. The diameter signal 712 is triangular, with the individual peaks corresponding to the turns of the foreign fiber 740. In conventional yarn clearing, they could be below the clearing limit, so that the foreign fiber 740 would not be cleared. With the method according to the invention, however, the image 720 and thus the reference event of Figure 7 are classified as inadmissible. A measurement signal with a similar triangular shape is also classified as inadmissible according to the invention and removed from the yarn during yarn clearing.

[0069] The yarn events can be represented in a known event field 800, as shown schematically in Figure 8. The event field 800 is a two-dimensional Cartesian coordinate system. Values ​​of an event variable, e.g., the deviation AM of a length-related mass density M of the yarn from a target value, are plotted along a vertical axis 802, and values ​​of an event length L are plotted along a horizontal axis 801 (see Figure 2). In this event field 800, yarn events, i.e., the measured values ​​and / or reference values, can be entered as points. The points 811, 812 entered in the example of Figure 8 are reference values.

[0070] In the event field 800 of Figure 8, a cleaning limit is entered in the form of a cleaning curve 830. The cleaning curve 830 divides the event field 800 into two areas 831, 832: a first area 831 in which deviations are tolerated, and a second area 832, which is complementary to the first area 831 and in which deviations are cut out of the yarn or at least registered as inadmissible defects. The cleaning curve 830 thus indicates how far a yarn event of a given length may deviate from the target value, or how long a yarn event with a given deviation may be in order to just be tolerated. In addition, a yarn body 840, as known from WO-2010 / 078665 A1, is shown in the event field 800 of Figure 8. To determine the yarn body 840, a sufficiently long yarn section is measured in a calibration process.From the measured values ​​of the respective associated event parameters AM and L, event densities in the event field 800 are determined. In this way, an event density can be uniquely assigned to each point in the event field 800. A sufficiently high limiting event density, e.g., 1000 events per 100 km of yarn length, is selected. The connection of all points in the event field 800 to which the limiting event density is assigned results in a density curve 841, which delimits the yarn body 840 from the rest of the event field 800. The yarn body 840 is delimited by the coordinate axes 801, 802 itself toward the two coordinate axes 801, 802. These delimitations create a coherent area 840 that is characteristic of the yarn.

[0071] The inventive calculation and representation of the yarn body 840 allows the operator to quickly and intuitively grasp the characteristics of the yarn being examined and thus determine the cleaning curve 830. The event density in the yarn body 840 is so high that it would be pointless to attempt to remove events from the yarn body 840. Therefore, the cleaning curve 830 should lie above the yarn body 840.

[0072] In the event field 800 of Figure 8, two types of reference events 811, 812 are plotted as points. A first type 811 lies in the area between the yarn package 840 and the cleaning curve 830. In this area, which lies below the cleaning curve 830, no yarn events should generally be cut out. However, it may still happen that certain events in this area are perceived as disruptive and therefore cutting them out is desired. Shifting the cleaning curve 830 downwards would lead to the unnecessary cutting out of tolerable yarn events and thus reduce productivity. The present invention takes a different approach by defining specific reference events 811 in the area. Yarn events that are similar to the reference events 811 of the first type should be cut out as exceptions to the general rule.The points corresponding to the positions of this first type 811 of reference events are marked with crosses in the example of Figure 8. To establish a connection with Figure 2(b), the coordinates LR, AMR for such a point are plotted.

[0073] A second type 812 of reference events lies above the cleaning curve 830. This is where the disturbing yarn defects should be located, which should generally be cut out. However, there may be certain yarn events 812 located above the cleaning curve 830 that, upon closer inspection, are not disturbing at all. Shifting the cleaning curve 830 upwards would result in other, disturbing yarn defects not being cut out, thus reducing the quality. In this case, too, the present invention provides exceptions to the general rule, namely certain reference events 812 above the cleaning curve 830, which are marked with circles in Figure 8. Yarn events that are similar to these reference events 812 of the second type should be spared from cutting out as exceptions to the general rule.

[0074] The computer-implemented method according to the invention for classifying an event in an elongated textile test specimen moved along its longitudinal direction is now explained with reference to the flow diagram in Figure 9.

[0075] At least one measured value of at least one event parameter for the event to be classified is measured 901. A first example of an event parameter is the mass deviation AM discussed in connection with Figure 2, of which a plurality of measured values ​​are combined to form a measurement signal 210. A second example are the two event parameters event length L and mass deviation AM, also discussed in connection with Figure 2. Further examples are the capacitive and optical measurements discussed in connection with Figure 3, of which the length-dependent signals 310, 320 and / or the deviation AM or AD and the length L can be used as event parameters.

[0076] At least one reference data set and preferably a plurality of reference data sets are provided 902, preferably in a reference database. Each reference data set relates to a reference event and contains at least one reference value of at least one event parameter and a classification of the reference event. "Classification" here refers to the assignment of exactly one class to the reference event. The provision 902 of the reference data sets will be discussed in more detail below with reference to Figure 10.

[0077] One reference data set after the other is retrieved from the reference database 903. The at least one measured value is compared with the at least one reference value of the respective reference data set 904. If the at least one measured value is most similar to the at least one reference value of all previous reference data sets 905, the classification from the respective reference data set is stored, e.g., in a buffer 906. Examples of similarity criteria were given above. The comparison 904 of the at least one measured value with the at least one reference value is repeated until all reference data sets have been checked 907.

[0078] After the comparisons 904, the currently available classification is the classification of the reference event whose at least one reference value is most similar to the at least one measured value according to the specified similarity criterion. The current event is classified according to the available classification 908.

[0079] If the method according to the invention is used for yarn clearing, classification 908 preferably takes place in a classification system with exactly two classes, namely a first class with permissible events and a second class with impermissible events. Events classified in the first class are left in the yarn, while events classified in the second class are removed from the yarn.

[0080] In other embodiments, the classification system may include more than two classes. A first example is a classification system with the following classes for yarn events:

[0081] (i) permitted for all types of further processing;

[0082] (ii) not permitted only as weft yarn for weaving;

[0083] (iii) not suitable for weaving;

[0084] (iv) inadmissible for knitting;

[0085] (v) inadmissible for all types of further processing. A second example of a classification system with more than two classes is one with the following classes for yarn events:

[0086] (i) permitted for all colours;

[0087] (ii) not permitted for white fabrics / knitted fabrics;

[0088] (iii) not permitted for dyed fabrics / knitted fabrics;

[0089] (iv) inadmissible for all colours.

[0090] The flowchart in Figure 10 shows an example of a method according to the invention for providing at least one reference data set for the method according to the invention or the device according to the invention for classifying an event. The method can be carried out in a dedicated learning phase and / or during production, e.g., during yarn cleaning.

[0091] There must be a reference event 1001 for which the reference data set is to be provided.

[0092] In one embodiment, the reference event is present in the same test material in which the other events are later to be classified. In a yarn clearing system, this can be achieved by examining a sufficiently long yarn section for reference events in a learning phase before yarn clearing and classifying these as described below, or by continuously using and classifying the yarn events found as reference events during yarn clearing. According to the first alternative, reference events below the cleaning curve 830 (see Figure 8) can also be taken into account in a simple manner, which is difficult when the yarn clearing function is switched on. The second alternative has the advantage that the yarn is cleaned from the first meter onwards and the reference database is continuously completed; for this, no reference data records are yet available at the start of the cleaning process.Both alternatives are not mutually exclusive and can be combined.

[0093] In another embodiment, the reference event is present in a different test material than the events to be classified. However, this material should have similar properties to the material in which the events to be classified are located. According to this embodiment, the provision of the at least one reference data set can take place, for example, at a specially configured winding station 101 of a yarn winding machine 100 (see Figure 1), and the reference data sets thus provided can be used to classify the yarn events at all other winding stations 101. Alternatively, the reference data sets can be provided by a supplier and made available to spinning mills.

[0094] A reference parameter is selected for the present reference event 1002. At least one reference value of the event parameter is measured 1003. This is repeated if necessary for further reference values ​​and reference parameters 1004, 1005. The same reference parameters are used for each reference event.

[0095] An image of the reference event is captured 1006. This can be done with a camera. In Figure 1, such a camera 130 is schematically shown on a specific winding unit 102, which can be referred to as the "pilot winding unit." The camera 130 can be a digital camera with a planar image sensor. Alternatively, for a moving yarn 110, it can only have a line sensor, the successive images of which can be combined to form a two-dimensional image. The camera 130 can be connected to the measuring head 121 of the pilot winding unit 120 or directly to the control unit 123 in order to transmit the images it captures to the control unit 123. The remaining winding units 101 do not require a camera. Those winding units 101 that process yarn 110 of the same type as the pilot winding unit 102 can use the reference data sets from the database 125 provided for this yarn type to classify yarn events.

[0096] The image is classified 1007. The classification 1007 is performed by an operator. The image is output to the operator on a suitable output unit 126 (see Figure 1) so that they can visually perceive it. The operator assesses the reference event represented in the image and classifies it according to their assessment. They enter the classification into the inventive device 120 via a suitable input unit 127. Appropriate instructions can be issued to the operator on the output unit, e.g., "Please enter your classification." Thus, the classification is performed by means of guided human-machine interaction. An alternative embodiment does not require capturing an image. Instead of an image, the operator assesses the event directly, e.g., by viewing it with the naked eye or through a microscope. For this purpose, the relevant winding unit must be stopped until the classification of the event is complete.

[0097] The classification is assigned to at least one reference value 1008 and stored together with the reference value as a reference data set 1009.

[0098] These steps can be repeated for further reference events 1010. In this way, a database 125 with reference data records is built.

[0099] Figure 11 schematically shows a table 1100 of a database 125 (see Figure 1) with reference data records for the method and device according to the invention. Each row 1111, 1112, ... of table 1100 contains a data tuple that refers to a specific reference data record. A first column 1121 of table 1100 can contain a sequence number as a primary key for table 1100. A second column 1122 contains the at least one reference value, e.g., the reference signal (see Figure 2(b)). A third column 1123 contains the image of the respective reference event. Finally, a fourth column 1124 contains the classification of the respective reference event.

[0100] The method according to the invention can, on the one hand, be implemented as an extension of the known yarn clearing method. A clearing limit is specified, for example, in the form of a clearing curve 830 according to Figure 8. The reference events 811, 812 define exceptions to the clearing according to the clearing limit and cancel the clearing rule specified by the clearing limit. In this case, it is advantageous to design the similarity criterion in two stages, namely with an "absolute" and a "relative" similarity criterion. The relative similarity criterion was discussed in detail above. The absolute similarity criterion defines how similar a reference value must at least be to the present measured value in order to be considered at all in the relative similarity criterion.In the example of signals 210, 220 according to Figure 2, a threshold value of the maximum of the cross-correlation function can be specified for the absolute similarity criterion. A reference signal 220 is only considered for the classification of an event if the maximum of its cross-correlation function with the measurement signal 210 of the event is greater than or equal to the threshold value.

[0101] A first graphical representation of an absolute similarity criterion s is shown in Figure 4 as a circular sector 420 around the ideal corner point P(l, 1), a second graphical representation in Figure 8 as a circular disk 820 around a reference point 811. The absolute similarity criterion is met if the measured values ​​of an event lie in one of the hatched areas 420, 820, i.e. close enough to a reference event, which applies, for example, to point 410 in Figure 4. If no reference value meets the absolute similarity criterion for the current measured value, i.e. the event lies too far away from all reference events, the event is assessed as usual according to the cleaning rule given by the cleaning limit. Only when at least one reference event is close enough to the event can it cancel the cleaning rule given by the cleaning limit.This embodiment has the advantage that it also works with only a few reference data sets, or even with a single reference data set.

[0102] On the other hand, the method according to the invention can be carried out autonomously, without a conventional cleaning limit. Yarn cleaning is then carried out exclusively according to the available reference data sets. The advantage of this embodiment is that the yarn is cleaned exclusively according to a manually performed classification rather than an abstract cleaning limit. The disadvantage is that a larger number of reference data sets are required to cover all possible yarn events as accurately as possible.

[0103] Of course, the present invention is not limited to the embodiments discussed above. With knowledge of the invention, the skilled person will be able to derive further variants that also fall within the scope of the present invention, as defined in the independent patent claims.

[0104] 100 Goat winding machine

[0105] 101 Setting the spool

[0106] 102 pilot winding unit

[0107] 110 test material, yarn

[0108] 111 heads

[0109] 112 cross-wound bobbin

[0110] 113 Direction of movement of the yarn

[0111] 120 device according to the invention

[0112] 121 measuring heads

[0113] 122 data line

[0114] 123 Control unit

[0115] 124 central computers

[0116] 125 Database

[0117] 126 Output unit

[0118] 127 Input unit

[0119] 130 Camera

[0120] 201 horizontal axes

[0121] 202 vertical axes

[0122] 210 fictitious measurement signal

[0123] 220 fictitious reference signal

[0124] 301 horizontal axes

[0125] 302, 303 vertical axes

[0126] 310, 320 fictitious measurement signals

[0127] 401 horizontal axis

[0128] 402 vertical axis

[0129] 410 points

[0130] 411 Distance of the point from the vertex

[0131] 420 Circle sector for absolute similarity criterion

[0132] 501, 601, 701 horizontal axes

[0133] 502, 602, 702 vertical axes

[0134] 511, 611, 711 Measurement signals of a capacitive sensor

[0135] 512, 612, 712 Measurement signals of an optical sensor for transmitted light

[0136] 513, 613, 713 Measurement signals of an optical sensor for reflected green light

[0137] 514, 614, 714 Measurement signals of an optical sensor for reflected red light

[0138] 520, 620, 720 pictures of yarn events

[0139] 530, 630, 730 yarns

[0140] 540, 640, 740 game events

[0141] 800 Event field

[0142] 801 horizontal axis

[0143] 802 vertical axis

[0144] 811, 812 Points representing reference events

[0145] 820 Circular disk for absolute similarity criterion 830 Cleaning curve

[0146] 831 Area below the cleaning curve

[0147] 832 Area above the cleaning curve

[0148] 840 yarn bodies

[0149] 841 Density curve

[0150] 901 Measuring a measured value

[0151] 902 Providing at least one reference data set

[0152] 903 Get the next reference record

[0153] 904 Compare the measured value with the reference value

[0154] 905 Decision: Is the reference value most similar to the measured value?

[0155] 906 Preparing the classification from the reference data set

[0156] 907 Decision: Have all reference records been checked?

[0157] 908 Classify the event according to the provided classification

[0158] 1001 new reference events

[0159] 1002 Get a reference parameter

[0160] 1003 Measuring a reference value

[0161] 1004 Decision: Is there another reference value?

[0162] 1005 Decision; Is there another reference parameter?

[0163] 1006 Taking an image of the reference event

[0164] 1007 Classifying the image

[0165] 1008 Assigning the classification to the reference value

[0166] 1009 Saving the reference value and classification as a reference data set

[0167] 1010 Decision: Is there another reference event?

[0168] 1100 Table of a database

[0169] 1111, 1112, ... rows of the database

[0170] 1121, 1122, ... columns of the database

Claims

PATENT CLAIMS 1. A computer-implemented method for classifying an event (540, 640, 740) in an elongated textile test piece (110), comprising the following steps: a) measuring (901) at least one measured value of at least one event parameter for the event (540, 640, 740); b) providing (902) at least one reference data set, which contains, for each reference event, at least one reference value of the at least one event parameter and a classification of the reference event; c) comparing (904) the at least one measured value with the at least one reference value of all reference data sets; and d) classifying (908) the event (540, 640, 740) according to the classification of the reference event whose at least one reference value is most similar to the at least one measured value according to a predetermined similarity criterion.

2. The method according to claim 1, wherein the at least one event parameter is an event variable, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material (110) from a target value, and the at least one reference value and the at least one measured value are each a signal (220, 210) which consists of a plurality of values ​​of the event variable as a function of a length position or of time.

3. The method of claim 2, wherein the similarity criterion includes a cross-correlation of the two signals.

4. The method according to claim 1, wherein the at least one event parameter comprises an event size, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material (110) from a target value, and an event length.

5. The method according to claim 4, wherein the similarity criterion includes a distance between a point representing the event and a point representing the reference event (811, 812) in a coordinate system spanned by the event length and the event size.

6. Method according to one of the preceding claims, wherein several different event parameters are taken into account.

7. The method according to claim 6, wherein at least a first event parameter is measured capacitively and at least a second event parameter is measured optically.

8. Method according to one of the preceding claims, wherein the measured value of the at least one event parameter is measured (901) while the elongated textile test material (110) is moved along its longitudinal direction.

9. Method according to one of the preceding claims, wherein the at least one reference data set is provided (1003) by measuring the at least one reference value for each reference event (1003), the reference event is visually assessed and classified by an operator (1007), and the classification is assigned to the at least one reference value (1008) and stored together with the reference value as a reference data set (1009).

10. The method of claim 9, wherein the reference event is visually assessed by capturing (1006) an image (520, 620, 720) of the reference event and presenting the image (520, 620, 720) to the operator for visual assessment.

11. Method according to one of the preceding claims, wherein the classification (908) is carried out in a classification system with exactly two classes, namely a first class with permissible events and a second class with impermissible events.

12. A computer-implemented method for cleaning yarn (110) moved along its longitudinal direction, wherein events (540, 640, 740) in the yarn (110) are classified according to claim 11 and events classified in the second class are removed from the yarn (110).

13. A device (120) for classifying an event (540, 640, 740) in an elongated textile test piece (110), comprising: a) a measuring device (121) for measuring at least one measured value of at least one event parameter for the event (540, 640, 740); and a computer system (123) with b) a memory (125) for storing at least one reference data set, which contains, for each reference event, at least one reference value of the at least one event parameter and a classification of the reference event; c) a processor configured to compare the at least one measured value with the at least one reference value of all reference data sets; and d) a processor configured to classify the event (540, 640, 740) according to the classification of the reference event whose at least one reference value is most similar to the at least one measured value according to a predetermined similarity criterion.

14. Device (120) according to claim 13, wherein the at least one event parameter is an event variable, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material from a target value, the at least one reference value and the at least one measured value are each a signal (220, 210) which consists of a plurality of values ​​of the event variable as a function of a length position or of time, and the measuring device (121) is set up to record the signal (210).

15. Device (120) according to claim 13, wherein the at least one event parameter comprises an event size, for example a deviation of a length-related mass density, a cross-section or a reflectivity of the test material (110) from a target value, and an event length.

16. Device (120) according to one of claims 13-15, wherein the measuring device (121) is arranged to measure a plurality of different event parameters.

17. Device (120) according to claim 16, wherein the measuring device (121) is configured to capacitively measure at least one first event parameter and to optically measure at least one second event parameter.

18. The device (120) according to any one of claims 13-17, wherein the processor is configured for classifying in a classifying system having exactly two classes, namely a first class with permissible events and a second class with impermissible events.

19. Yarn clearing system for cleaning yarn (110) moving along its longitudinal direction, comprising a device (120) according to claim 18 and a cutting device triggerable by the computer system (123) for cutting the yarn (110) as a result of classifying an event in the second class.

20. Yarn processing machine, e.g., a dishwashing machine (100) or spinning machine, having a plurality of yarn processing stations (101, 102), at each of which yarn (110) is wound onto a bobbin (112), comprising a yarn clearing system according to claim 19.

21. Yarn processing machine according to claim 20, comprising at least one camera (130) which is set up at one of the yarn processing stations (102) for recording images (520, 620, 720) of events (540, 640, 740) in the yarn (110), an output unit (126) connected to the at least one camera (130) for outputting an image (520, 620, 720) recorded by the camera (130), and an input unit (127) connected to the processor for inputting a classification of the output image (520, 620, 720).

22. Computer-implemented method for providing at least one reference data set for the method according to any one of claims 1-12 or the Device according to one of claims 13-21, wherein for each reference event in an elongated textile test material (110) at least one reference value of at least one event parameter for the reference event is measured (1003), the reference event is visually assessed and classified (1007) by an operator, and the classification is assigned to the at least one reference value (1008) and stored together with the at least one reference value as a reference data set (1009).

23. The method according to claim 22, wherein the reference event is visually assessed by capturing an image (520, 620, 720) of the reference event and presenting the image (520, 620, 720) to the operator for visual assessment.

24. A computer-readable medium on which at least one reference data set provided according to claim 22 or 23 is stored.

Images