Method for determining quality of collection bobbin

Abstract

The invention relates to a method and a device for determining the quality of a collection bobbin (14) containing a synthetic thread (30), in which during the production of the collection bobbin (14), a total error (61) is compiled, in which the total error (61) is classified as a quality-dependent error (63) or a quality-independent error (62), in which the quality-dependent error (63) is used to determine a quality class (7) of the collection bobbin (14).

Description

[0001] The present invention relates to a method according to the features of claim 1 for determining the bobbin quality of a collecting bobbin containing synthetic yarn, and to an apparatus according to the features of claim 11 for determining the bobbin quality of a collecting bobbin containing synthetic yarn.

[0002] For example, a method for monitoring a plurality of processing points of synthetic yarn is known from WO 2019 / 137835 A1.

[0003] In the production and processing of synthetic yarn, product and / or process parameters are generally continuously monitored in order to obtain as stable a process control as possible, in particular as stable a product quality as possible.

[0004] Especially when producing textured threads, monitoring the yarn tension of the thread running at the respective processing points has proven effective in detecting process faults and / or production fluctuations at an early stage.

[0005] In known methods for monitoring production processes, the yarn tension of the thread running at the processing points is continuously measured.

[0006] The yarn tension measurement signal generated here is compared with a threshold value of the permitted yarn tension in order to identify so-called error diagrams and use them for further analysis.

[0007] By means of a suitable learning algorithm, the respective error causes in the production process are detected from the error diagrams in order to be eliminated by an operator, for example.

[0008] However, with current machine learning programs, it is not possible to determine which types of errors and qualities the program has recorded and transmitted.

[0009] Nevertheless, the first bobbin quality is classified based on compiled errors.

[0010] The problem is that errors unrelated to quality are also taken into account in the classification of the bobbin quality.

[0011] Therefore, the object of the present invention is to provide a method and an apparatus which enable the bobbin quality to be classified only based on quality-related errors.

[0012] Regarding the method, this object is achieved by a method according to the features of claim 1.

[0013] Regarding the apparatus, this object is achieved by an apparatus having the features of claim 11.

[0014] Further improvements to the method and the apparatus are claimed in the dependent claims.

[0015] According to one aspect of the invention, a method for determining the quality of a collected bobbin containing synthetic yarn is provided, wherein during the production of the collected bobbin, total errors are compiled, wherein the total errors are classified as quality-relevant errors or quality-independent errors, the quality-relevant errors being used to determine the quality grade of the collected bobbin.

[0016] The total error is to be understood in particular as the error which has been recorded during the production of the synthetic yarn.The total error can be reported to the apparatus for producing synthetic yarn, for example to a data analysis unit or a controller of the apparatus for producing synthetic yarn.

[0017] Synthetic yarns are understood here to mean in particular yarns which have been produced from a plurality of filaments by means of a synthetic melt to form pre-oriented yarns or POY, which are then processed in a post-processing step to form textured yarns which are particularly suitable for processing in the textile industry. These textured yarns are also referred to as draw-textured yarns (DTY).

[0018] With this method, it is advantageously possible to evaluate the bobbins according to their actual quality, thereby making it possible to ensure that the synthetic yarn collected on the collecting bobbin is suitable for further processing.

[0019] The quality of the yarn may be relevant, for example, for the weaving process and / or for subsequent dyeing, wherein a high-quality yarn has better properties for subsequent processing and the risk of a poor-quality end product can be minimized.

[0020] According to a development of the method for determining the package quality of a collected package, the total error is determined on the basis of data determined from a sensor device, in particular a yarn tension sensor device.

[0021] Currently, the quality of yarn can usually be determined by sensors that monitor the yarn during production.

[0022] In particular, the measurement of yarn tension with correspondingly configured measuring instruments which actively measure the yarn tension is already established in this respect.

[0023] However, there are other devices that allow the yarn tension to be determined directly or indirectly. This can be determined indirectly, for example, via the drafting system and its power consumption.

[0024] For this purpose, the work required, for example the torque of a godet or reel guide, can also be determined in order to determine the yarn tension.

[0025] According to a further development of the method for determining the bobbin quality of collected bobbins, the data determined from the sensor device are recorded in a graph which defines tolerance ranges, wherein the graph is analyzed using a first machine learning procedure for determining the total error to detect the total error.

[0026] A machine learning program shall be understood to mean a software module and / or an AI application that has been programmed to independently solve problems or independently identify patterns. This can include detecting and / or reporting possible errors from defined measurement data in production by synthetic methods. The programming can also include teach-in.

[0027] The first machine learning program for determining the total error basically reports the number of errors and may also be referred to hereinafter as the machine learning program for determining the number of errors.

[0028] To determine the total error from the data of the sensor device, the data is recorded in a time graph, where the abscissa indicates the progress of time and the ordinate indicates the corresponding measured values of the sensor device.

[0029] In the graph of yarn tension measurement, the time values are plotted on the abscissa and the tension values are plotted on the ordinate. This produces a typical graph based on which the occurring errors can be analyzed according to the processes and patterns depicted.

[0030] The graph can be analyzed accordingly by means of the first machine learning program for detecting the total error, to which AI (Artificial Intelligence) can be connected.

[0031] The machine learning program here can be programmed for this problem by teach-in. After teach-in, the machine learning program can automatically detect errors and report them.

[0032] According to a preferred improvement of the method for determining the bobbin quality of a collecting bobbin, the total error and the first machine learning program are used to determine a first quality grade, which is assigned to the collecting bobbin.

[0033] For the first classification of the collecting bobbin, the determined total faults are compiled and also assigned by means of the machine learning program and / or an additional program of the collecting bobbin currently being produced.

[0034] The first quality grade shall be understood as the first classification of the quality of the collecting bobbin based on the number of the recorded total errors.

[0035] This is stored in the database. Thus, at least the first bobbin quality can be determined based on the total errors occurring in the production of the collecting bobbin and the associated synthetic yarn.

[0036] However, in a further step, the method of the present invention allows for a more precise determination of the bobbin quality by evaluating the total error according to its correlation with the yarn quality.

[0037] According to a specific improvement of the method for determining the bobbin quality of a collection bobbin, with the aid of a second machine learning program for evaluating the error quality, the total error is classified as a quality-related error or a quality-unrelated error.

[0038] Classifying or analyzing the total error with respect to quality relevance allows for a better assessment of the bobbin quality of the associated collection bobbin around which the yarn is wound.

[0039] Quality-related errors should be understood as errors determined during the production of the synthetic yarn that reduce the quality of the yarn and its further processability. These errors can be, for example, errors that reduce the dyeing and / or weaving of the subsequently formed textile from the yarn.

[0040] Quality-unrelated errors should be understood as errors determined in the production of the synthetic yarn that do not change the quality of the yarn. Such recorded errors may be caused by false messages of single filament breakage or additional measurements with a yarn tension measuring instrument. Quality-unrelated errors do not affect the quality and processability of the yarn.

[0041] This has the following advantage: A collection bobbin that initially falls into a lower quality class based on the first classification of the total error can fall into a better quality class by reclassifying the total error as a quality-related or quality-unrelated error, because only quality-related errors are considered.

[0042] This also has the following advantage: Fewer collection bobbins are misassessed. This is because the method allows for a more accurate determination of the quality of individual collection bobbins, such that the collection bobbins can be further processed in a high-quality process. Thus, rejects and waste can also be avoided.

[0043] Furthermore, collection bobbins that are initially not suitable for further processing at all due to the total error can also be qualified for a higher quality class through reclassification, and thus unnecessary waste generation can also be avoided.

[0044] According to a specific improvement of the method for determining the bobbin quality of a collection bobbin, with the aid of a third machine learning program for defining a tolerance range, a maximum setpoint and a minimum setpoint for limiting the tolerance range for error detection are determined.

[0045] The maximum and minimum setpoints can also be defined by predetermined empirical values, for example, by an experienced operator.

[0046] However, in a specific improvement of the method, it is stipulated to monitor and further define the setpoints with the aid of the third machine learning program, where the machine learning program can also be taught here. The setpoints each define a tolerance range with a maximum and a minimum value, where exceeding or falling below the respective setpoint can indicate an error.

[0047] This error may be crucial for the quality of the bobbin. In addition, this error is also a corresponding warning that the setpoint allowed here has been exceeded or fallen below. If the respective specified setpoint is exceeded or fallen below excessively, the production process can even be interrupted.

[0048] According to another improvement of the method for determining the bobbin quality of the collected bobbins, the first machine learning program for detecting the total error, the second machine learning program for evaluating the error quality, and / or the third machine learning program for defining the tolerance range have a learning process from the operator, and the learning process of the first, second, and / or third machine learning program is requested from the operator.

[0049] To optimize the detection of errors and / or optimize the monitoring and setting of the setpoint, it is advantageous that, despite the available and recorded data, the relevant machine learning programs are further trained because these machine learning programs can also detect patterns and related errors that the operator cannot detect.

[0050] To ensure that the error detection is accurate and correct, the errors newly detected by the machine learning program should also be confirmed by an experienced operator so that unnecessary error reports do not occur.

[0051] According to another improvement of the method for determining the bobbin quality of the collected bobbins, the results of the first machine learning program, the second machine learning program, and / or the third machine learning program are evaluated by a quality algorithm for determining the second quality grade of the collected bobbins and stored in a database.

[0052] The quality algorithm is another instance that can be used to determine the bobbin quality; in particular, the role of the quality algorithm is to re - grade the first result identifying the first quality grade based on graded quality - related or quality - irrelevant errors.

[0053] The quality algorithm summarizes the results from error detection and error quality detection and re - grades the bobbin quality based on these results, so that advantageously, if most of the detected errors are not quality - related, the bobbin quality can be improved.

[0054] According to a specific improvement of the method for determining the bobbin quality of the collected bobbins, the second machine learning program has an error detection algorithm that reports quality - related errors and / or quality - irrelevant errors to the quality algorithm.

[0055] In a specific improvement, the second machine learning program for error quality detection not only grades the total error but also reports to the error evaluation of the quality algorithm what types of errors exist.

[0056] In a simplified form, only quality-related or only quality-unrelated errors can be reported to the quality algorithm here, so that the latter can evaluate which errors are reported in the total error.

[0057] Then, the quality algorithm program can improve or reduce the bobbin quality of the previously temporarily stored bobbins, depending on how much of the total error is quality-related or quality-unrelated.

[0058] According to a preferred improvement of the method for determining the bobbin quality of the collecting bobbins, the method comprises the following steps:

[0059] - Providing raw yarn from a supply bobbin,

[0060] - Processing the raw yarn into textured yarn,

[0061] - Collecting the total error recorded during the processing of the raw yarn,

[0062] - Determining a first quality grade based on the total error,

[0063] - Collecting the textured yarn on a collecting bobbin,

[0064] - Assigning a quality grade to the associated collecting bobbin with the textured yarn,

[0065] - Classifying and marking the total error as error quality with quality-related and quality-unrelated errors,

[0066] - Subtracting the quality-unrelated error from the total error to obtain a corrected error,

[0067] - Determining a second quality grade based on the error quality, and

[0068] - Assigning and storing the second quality grade of the associated collecting bobbin.

[0069] According to the improvement of the above method, the defined total error can be classified into quality-related and quality-unrelated errors by the process steps outlined above, and a new quality grading of the collecting bobbins can be carried out, where the errors are considered according to their relevance to the quality of the yarn produced.

[0070] This can cause the collecting bobbins initially assigned to a lower quality grade to move to a better quality grade.

[0071] According to one aspect of the present invention, there is provided a device for determining the bobbin quality of a collection bobbin containing synthetic yarn, in particular for an improved method as specified above. The device includes: a sensor device for collecting data of the synthetic yarn, a data analysis unit having an error detection unit and an error classification unit, which interacts with a controller of an operating point for producing the synthetic yarn, wherein the error classification unit interacts with a quality definition unit for defining a quality grade, and a database for storing the quality grade of the associated collection bobbin.

[0072] With this device, accurate quality grading of the produced collection bobbins can be carried out. In particular, by using the above device, the quality of the collection bobbins can be purposefully and precisely evaluated so as to be able to grade high-quality collection bobbins.

[0073] According to an improvement of the device for determining the bobbin quality of a collection bobbin containing synthetic yarn, a tolerance range determination unit is assigned to the quality definition unit, and the tolerance range determination unit transmits the current threshold of the tolerance range to the quality definition unit.

[0074] A favorable additional tool for improving the device is the tolerance range determination unit, which regularly controls and monitors the threshold for defining the tolerance range, enabling accurate error detection.

[0075] The present invention relates to a method and a device by means of which the determined errors are re-classified individually by sorting the compiled total errors according to their influence on the yarn quality. In this regard, at least harmless errors that do not affect the yarn quality are classified, marked, and / or identified so that they can be distinguished from errors that may have a negative impact on the yarn quality. This is also possible for quality-related errors.

[0076] The detection of harmless errors can be performed by pattern recognition of previously determined sensor data. For this purpose, the pattern recognition has a machine algorithm, similar to AI, which can be trained by teaching harmless errors. This is also possible for quality-related errors.

[0077] This enables the quality of the collection bobbin to be re-classified in the case where a first quality grading is performed based on the detected total errors and a second quality grading is performed by weighting the detected total errors at least according to their quality relevance and irrelevance to the yarn quality.

[0078] Hereinafter, further exemplary embodiments will be described with reference to the accompanying drawings and with reference to the drawings to more specifically explain the method of the present invention for determining the bobbin quality of a collection bobbin and the associated device for determining the bobbin quality of a collection bobbin. This results in further embodiments and advantages of the method and the device.

[0079] In the figure:

[0080] Figure 1 An exemplary embodiment of a processing point of a textile machine for producing synthetic crimped yarns is schematically shown, which has an associated data analysis unit,

[0081] Figure 2 A first machine learning program for defining and detecting the total error is schematically shown, by means of which a first quality grade of the collection bobbin and the synthetic yarn collected thereon can be defined,

[0082] Figure 3 A third machine learning program for defining a tolerance range is schematically shown, which is defined by a maximum and a minimum threshold to be defined,

[0083] Figure 4 A third machine learning program for detecting the error quality is shown, in particular the error quality affecting the bobbin quality and the errors not affecting the bobbin quality, and

[0084] Figure 5 An overall view of the data analysis unit and its associated first, second and third machine learning programs is shown, and the first, second and third machine learning programs can be used to grade the bobbin quality by means of a quality algorithm.

[0085] Figure 1 An exemplary embodiment of the processing point 1 of the textile machine is schematically shown, in this case a texturing machine for producing textured yarn 30 (DTY = drawn textured yarn) from non-textured yarn 3 (also called POY (pre-oriented yarn)). Such a texturing machine has a plurality of such processing points 1 identical in design. For example, usually more than 100 processing points are arranged side by side along the longitudinal side of the machine. Here, only one processing point 1 of the textile machine will be described in more detail.

[0086] Figure 1 The processing point 1 and the winding point 2 of the textile machine are schematically shown. The processing point 1 has a bobbin creel 4 in which a supply bobbin 5 and a spare bobbin 6 are held. The supply bobbin 5 supplies the yarn 3, which is transferred for drafting and texturing at the processing point 1. The supply bobbin 5 is also called a so-called POY bobbin because it was previously produced in a melt spinning process and wound with newly spun POY yarn. The yarn end of the supply bobbin 5 is connected to the yarn head of the spare bobbin 6 by a knot. Thus, after the supply bobbin 5 is completely unwound, the yarn 3 is continuously pulled out. Then the yarn end of the spare bobbin 6 is connected to the yarn head of the new supply bobbin 5.

[0087] The yarn 3 is pulled off from the supply bobbin 5 by a first conveying mechanism 7.1. The conveying mechanism 7.1 is driven by a drive device 8. In this exemplary embodiment, the conveying mechanism 7.1 is formed by a driven godet and a freely rotatable roller, and the yarn is wound around the godet and the roller a plurality of times. Further downstream of the conveying mechanism 7.1, a heating device 9, a cooling device 10 and a texturing unit 11 are arranged. The texturing unit 11 is preferably designed as a false-twist friction unit in order to generate false twists in the multifilament yarn, which causes the crimping of the individual filaments of the yarn. In order to draw the yarn, a second conveying mechanism 7.2 is arranged downstream of the texturing unit 11 and is driven by the drive device 8. The design of the conveying mechanism 7.2 is the same as that of the first conveying mechanism 7.1, and the second conveying mechanism 7.2 operates at a higher circumferential speed for drawing the yarn. In the processing point 1, the synthetic yarn 3 is thus deformed and simultaneously drawn.

[0088] After the yarn 3 has been crimped, it is guided to the winding point 2 by a third conveying mechanism 7.3. The conveying mechanism 7.3 is driven by the drive device 8. The conveying mechanism 7.3 is designed as a so-called clamping conveying mechanism, which has a drive shaft and a pressure roller. The yarn 3 is guided in a clamping slot on the circumference of the shaft.

[0089] The winding point 2 has a bobbin holder 13, which holds the collecting bobbin 14. The collecting bobbin holder 13 is designed to be pivotable and can be operated manually or automatically for replacing the collecting bobbin 14. A drive roller 15 is assigned to the collecting bobbin holder 13, and the drive roller 15 is driven by a roller drive device 15.1. In order to lay the yarn on the circumference of the collecting bobbin 14, a traversing unit 12 is assigned to the winding point 2, and the traversing unit 12 has a drivable traversing guide. For this purpose, the traversing guide is oscillatingly driven by a traversing drive device 12.1. The traversing drive device 12.1 and the roller drive device 15.1 of the winding point 2 are designed as separate drive devices and are connected to the machine control unit 16. Similarly, the drive devices 8 of the conveying mechanisms 7.1, 7.2 and 7.3 and the texturing drive device 11.1 of the texturing unit 11 in the processing point 1 are designed as separate drive devices and are connected in parallel to the machine control unit 16.

[0090] In order to monitor the production process, the yarn tension of the running line is continuously recorded at each processing point 1 of the textile machine. For this purpose, in the exemplary processing point 1, a sensor device 17 is arranged in the yarn path upstream of the conveying mechanism 7.2.

[0091] The sensor device 17 has a yarn tension sensor 17.1 and a measurement signal converter 17.2. The sensor device 17 is connected to a data analysis unit 18. The data analysis unit 18 is coupled to a transmitter 19, which is connected to a suitable receiving system using wired or wireless transmission technology to transmit information and signals. The data analysis unit 18 has a plurality of software modules or machine learning programs 35, 36, 37 to analyze the measurement signals from the sensor device 17 and the data obtained therefrom for monitoring the production process.

[0092] The data collected and evaluated by the data analysis unit 18 can be transmitted to the database 66 and / or the machine control unit 16 via the transmitter 19. The transmitter 19, which acts as a transmitter and a receiver, can transmit data wirelessly and / or via a cable.

[0093] At the start of the process, which requires a product change in the supply bobbin or after a certain operating time of the process unit, the state of the yarn tension is usually checked by manual measurement. For this purpose, the operator uses a yarn tension measuring instrument 25, which the operator holds in the yarn path of the processing point 1 to be inspected.

[0094] In Figure 1 this, the yarn tension measuring instrument 25 guided by the operator is shown in dashed lines. In this case, the yarn tension measuring instrument 25 is preferably held in the yarn path upstream or downstream of the sensor device 17. The yarn tension measured value displayed by the yarn tension measuring instrument 25 is checked by the operator to see if it is within the predetermined tolerance range of the production process. However, the intervention of the yarn tension instrument 25 creates an event at the sensor device 17, resulting in a deviation from the standard state of the yarn tension in the production process. This event is fed by the yarn tension sensor 17.1 and the measurement signal converter 17.2 to the data analysis unit 18.

[0095] Inside the data analysis unit 18, a yarn tension analysis module is provided, which continuously receives the measurement signals of the measurement signal converter 17.2 and generates a so-called error chart.

[0096] The chart 351 defines an abnormal measurement signal process of the yarn tension. Such a chart 351 is always generated in data technology when the yarn tension measured online in the production process deviates from the range 352 defined by the specified limit values OG, UG. If the yarn tension of the thread significantly moves outside the defined range 352, this has an impact on the quality of the thread and thus on the quality of the product produced. Therefore, each event for changing the yarn tension is recorded and generated as an error chart.

[0097] Figure 2Schematically shows an error chart 351, which is assigned as a cause of the error to the manual yarn tension measurement during the analysis. The determination and analysis of the error chart are clear from EP 3 737 943 A1 and are described in more detail therein.

[0098] There are multiple error charts 351, which are analyzed by means of a machine learning program to identify the source of the fault.

[0099] Figure 2 The error chart 351 shown in also shows the typical measurement signal course of the yarn tension T that occurs when using the yarn tension measuring instrument 25 for manual measurement of the yarn tension. For this purpose, the yarn tension T is plotted on the ordinate and the time t is plotted on the abscissa.

[0100] The intervention of the yarn tension measuring instrument 25 in the yarn path causes an increase in the yarn tension T, which remains substantially constant during the duration of the manual measurement of the yarn tension. In this regard, the manual yarn tension measurement can be clearly assigned to the error chart using a machine learning program.

[0101] From Figure 2 it can also be seen that the error chart 351 has a threshold, which indicates a maximum value OG and a minimum value UG, which also define a tolerance range 352.

[0102] If the measured yarn tension remains within the tolerance range 352, no total error 61 is displayed. However, if the yarn tension exceeds the tolerance range 352, a total error 61 is generated, which is then compiled and assigned to the associated collecting bobbin 14 by means of the machine learning program 35, thereby assigning a first quality grade 40, which is Figure 2 shown by a, b or c in.

[0103] A quality grade 40 of a means that the collecting bobbin has a very high quality and is very suitable for further processing.

[0104] For a quality grade 40 with the label b, the quality is medium, and the collecting bobbin 14 is not suitable for further processing of every product, and for a quality grade 40 of c, the further processing of the collecting bobbin 14 is only possible to a limited extent, if at all.

[0105] However, since the total error 61 can also be caused by errors unrelated to quality, for example, as previously mentioned, the yarn tension measurement, where even an overshoot or undershoot of the threshold of the tolerance range 352 and the corresponding error classification can occur, according to the invention, a second machine learning program 36 for grading the error quality is provided, by means of which the quality relevance of the total error 61 can be analyzed.

[0106] First, refer toFigure 3 which shows a third machine learning program 37 having a tolerance range definition unit 70 by means of which a threshold value for the tolerance range for the diagram 351 can be defined or taught.

[0107] This can be carried out, for example, by means of a teaching 64 and / or the operator has specified the threshold values OG and UG based on empirical values of the operating position 1.

[0108] Figure 4 shows a second machine learning program 36 for detecting a quality-related error 63 which is used to evaluate the collecting bobbin 14 in the second quality class 200.

[0109] In a first step, the total error 61 detected by means of the first machine learning program 35 is fed to the second machine learning program 36. The second machine learning program 36 performs an error classification 60 with a corresponding algorithm and then classifies the total error 61 accordingly as a quality-unrelated error 62 or a quality-related error 63.

[0110] Here, an intermediate stage can also be established which further weights the relevance or irrelevance of the error accordingly, for example by a numerical value from 1 to 10, where 10 is a very high quality-related error, 1 is rather low, and at 0 the error has no influence on the quality of the thread of the collecting bobbin 14 at all.

[0111] In order to refine the error classification unit 60, the knowledge of the operator can also be introduced by means of a teaching 64.

[0112] The teaching 64 refers to a method for programming the machine learning program. For this purpose, the programming is supported by the operator who introduces his experience into the error messages of the machine learning program.

[0113] The pattern of the errors defined in this way is stored in the controllers of the respective machine learning programs 35, 36. This step can be repeated regularly as required to better control and / or check the response of the machine learning program.

[0114] The program sequence can include the machine learning program displaying new or unknown errors to the operator, possibly with solution suggestions or evaluation suggestions, and the operator confirming them and / or modifying them when these suggestions are correct. In addition, when additional corresponding evaluation by the operator or teaching by the operator is required, for example if a new, unknown error occurs, the error classification unit 60 can send a warning signal or an alarm signal to the operator.

[0115] In addition, the error classification unit 60 can interact with the quality definition unit 100 and can communicate its quality assessment of each total error 61 as a quality-related error or a quality-unrelated error 63, so that the quality definition unit 100 and its quality algorithm 101 can perform a new classification on the collecting bobbin 14 and can assign a second quality level 200 to the latter, which second quality level 200 then has priority for the further life of the collecting bobbin 14.

[0116] In addition, the error classification unit 60 is connected to a database 66 in which the corresponding evaluations or also the recorded total errors 61 and their evaluations as quality-related errors 63 or quality-unrelated errors 62 are stored, where a corresponding error level 65 - indicated by I, II or III - can also be assigned to the total error 61 or the quality-related error 63.

[0117] This is always stored in the database 66 linked to the associated collecting bobbin 14 so that it can be retrieved during and after the entire life of the collecting bobbin 14.

[0118] Figure 5 A general view of a possible embodiment of the data analysis unit 18 for the method according to the invention and for the device for determining the bobbin quality of the collecting bobbin 14 according to the invention is shown.

[0119] In the exemplary embodiment shown, the data analysis unit 18 can have three machine learning programs 35, 36 and 37.

[0120] Each machine learning program 35, 36, 37 has a different function. Their results are ultimately checked using the quality definition unit 100 with the quality algorithm 101. Then the quality definition unit 100 and its quality algorithm 101 are used to determine the second quality level 200, which is then assigned to each collecting bobbin 14.

[0121] For example, this can lead to an improvement of the previously defined first quality level 40 of the collecting bobbin 14 based on the results or definitions of the second machine learning program 36 and the third machine learning program 37.

[0122] The first machine learning program 35 has an error detection unit 50 which can detect the total error 61 with the aid of the previously presented diagram 351. Based on the total error 61, a first quality level 40 is assigned to the collecting bobbin 14. However, these total errors 61 can also have quality-unrelated errors, such as a measurement by a yarn tension measuring instrument which may exceed the setpoint of the tolerance range 352 but does not affect the quality of the thread 30.

[0123] These quality - independent errors may also include filament breakage, which may lead to changes in the yarn tension, but do not affect the quality processing or processability of the collecting bobbin 14 and the synthetic yarn wound thereon.

[0124] The total error 61 is evaluated by a second machine - learning program 36, which has an error classification unit 60 for detecting and defining quality - related errors 62 and quality - independent errors 63.

[0125] The teaching of the error classification unit 60 and the teaching of the error detection unit 50 can be carried out, for example, according to a teaching method 64, in which the operator at least confirms the corresponding error information of each unit.

[0126] In addition, the data analysis unit 18 can have a third machine - learning program 37, which at least defines the tolerance range for error detection.

[0127] The third machine - learning program 37 has a tolerance range determination unit 70, which defines the thresholds UG, OG of the error detection chart 351 or adjusts them accordingly in the corresponding analysis.

[0128] Then the results of the first, second, and / or third machine - learning programs 35, 36, 37 are fed to a quality definition unit 100, which then, by means of a quality algorithm 101, re - grades the first quality grade based on the results of the tolerance range determination unit 70, the error classification unit 60, and the error detection unit 50. Thus, a new second quality grade 200 is then provided for the collecting bobbin 14 and the latter is stored together with the associated collecting bobbin 14 in a database 66.

[0129] In another exemplary embodiment of the method for determining the bobbin quality, the data analysis unit 18 has the first and second machine - learning programs 35, 36 or the first and third machine - learning programs 35, 37.

[0130] In another embodiment, the second machine - learning program 36 can be programmed to be only used for error detection and error quality classification and used in the data analysis unit 18.

[0131] According to one embodiment of the method for determining the bobbin quality of a collecting bobbin, the method may include the following steps:

[0132] Providing the unprocessed yarn 3 from a supply bobbin 5, for example as Figure 1 shown.

[0133] The unprocessed yarn 3 can be provided to a creel 4.

[0134] To allow unrestricted processing of the yarn 3 to be processed, the raw yarn 3 is provided to the supply bobbin 5 and the reserve bobbin 6 respectively. The raw yarns 3 of the supply bobbin 6 and the reserve bobbin 5 are linked to each other respectively, as Figure 1 shown.

[0135] The raw yarn 3 is processed into the textured yarn 30.

[0136] Preferably, as Figure 1 shown, the raw yarn 3 is guided to the heating device 9 and the cooling device 10 via the conveying mechanisms 7.1, 7.2 and 7.3. In addition, a deformation is introduced into the raw yarn 3 by means of the deformation unit 11, thereby forming the textured yarn 30. The textured yarn is collected on the collecting bobbin 40.

[0137] During the production of the textured yarn 3, the total detected error 61 recorded during the processing of the raw yarn 3 is collected. The total error 61 can be recorded, for example, by means of the sensor device 17.

[0138] Then the total error 61 is evaluated by the data analysis unit 18 and its machine learning programs 35, 36 and 37.

[0139] Using the first machine learning program 35 for detecting the number of errors, the total error 61 and its frequency can be determined. The first quality grade 40 can be determined based on the frequency, where only the number of the total errors 61 is considered. The first quality grade 40 is assigned to the associated collecting bobbin 14 with the textured yarn 30.

[0140] Using the second machine learning program 36 for detecting and evaluating the error quality, the total error 61 is classified and marked as the quality-related error 63 and the quality-unrelated error 62.

[0141] With the quality definition unit 100 having the quality algorithm 101, the second quality grade 200 of the collecting bobbin can be determined, where, for example, the quality-unrelated error 62 is subtracted from the total error 61 to obtain the corrected error, so that the second quality grade 200 can be defined based on the corrected error.

[0142] The second quality grade 200 of the associated collecting bobbin 14 is also assigned and stored, for example, stored in the database 66.

[0143] As Figure 1 and Figure 5Schematically shown, according to one embodiment, an apparatus for determining the bobbin mass of a collection bobbin 4 may include: a sensor device 19 for collecting data of a synthetic yarn 30, a data analysis unit 18 having an error detection unit and an error classification unit 60, which interacts with a controller 16 of an operating point 1 for producing the synthetic yarn 30. The error classification unit 60 interacts with a quality definition unit 100 for defining a quality level 200. The collected data and results are stored and maintained in a database 66 for storing the quality level 200 of the associated collection bobbin 14.

[0144] In addition, a tolerance range determination unit 70 is assigned to the quality definition unit 200, and the tolerance range determination unit 70 transmits updated thresholds OG, UG of the tolerance range to the quality definition unit.

[0145] List of reference numerals:

[0146] 1 Processing point / position

[0147] 3 Undeformed yarn, pre-oriented yarn, POY

[0148] 4 Bobbin rack

[0149] 5 Supply bobbin

[0150] 6 Spare bobbin

[0151] 7.1, 7.2, 7.3 First conveying mechanism, second conveying mechanism, third conveying mechanism

[0152] 8 Driving device

[0153] 9 Heating device

[0154] 10 Cooling device

[0155] 11 Deformation unit

[0156] 11.1 Deformation driving device

[0157] 12 Traversing unit

[0158] 13 Collection bobbin holder

[0159] 14 Collection bobbin

[0160] 15 Driving roller

[0161] 15.1 Roller driving device

[0162] 16 Machine control unit

[0163] 17 Sensor device

[0164] 17.1 Yarn tension sensor

[0165] 17.1 Measurement signal converter

[0166] 18 Data analysis unit

[0167] 19 Transmitter

[0168] 25 Yarn tension measuring instrument

[0169] 30 Textured yarn, DTY, synthetic yarn

[0170] 35 First machine learning program

[0171] 36 Second machine learning program

[0172] 37 Third machine learning program

[0173] 351 Chart

[0174] 352 Tolerance range

[0175] 40 First quality grade

[0176] 50 Error detection unit

[0177] 60 Error classification unit

[0178] 61 Total error

[0179] 62 Quality-independent error

[0180] 63 Quality-related error

[0181] 64 Teaching

[0182] 65 Error grade

[0183] 66 Database

[0184] 70 Tolerance range determination unit

[0185] 100 Quality definition unit

[0186] 101 Quality algorithm

[0187] 200 Second quality grade

[0188] OG, UG threshold

[0189] T Yarn tension

Claims

1. A method for determining the bobbin quality of a bobbin (14) containing synthetic yarn (30), wherein, During the production of the collecting bobbin (14), a total error (61) is compiled, wherein the total error (61) is classified as a quality-related error (63) or a quality-unrelated error (62), and the quality-related error (63) is used to determine the quality grade (200) of the collecting bobbin (14).

2. The method for determining the bobbin mass of a collecting bobbin (14) according to claim 1, characterized in that, The total error (61) is determined from data determined from a sensor device (17), in particular a yarn tension sensor device (17.1).

3. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The data determined from the sensor device (17) is recorded in a graph (351), the graph defining a tolerance range (352), wherein a first machine learning program (35) for determining the total error is used to analyze the graph (352) to detect the total error (61).

4. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The first quality grade (40) is determined using the total error (61) and the first machine learning program (35), and the first quality grade is assigned to the collecting bobbin (14).

5. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The total error (61) is classified as a quality-related error (63) or a quality-unrelated error (62) by means of a second machine learning program (36) for evaluating the error quality.

6. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The definition of the tolerance range (352) for error detection is determined by means of a third machine learning program (37) for defining the tolerance range.

7. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The first machine learning program (35) for detecting the total error, the second machine learning program (36) for evaluating the error quality, and / or the third machine learning program (37) for defining the tolerance range have a learning process (64) from an operator, and the learning process (64) of the first machine learning program, the second machine learning program, and / or the third machine learning program is requested from the operator.

8. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The results of the first machine learning program (35), the second machine learning program (36), and / or the third machine learning program are evaluated by a quality algorithm (101) for defining the second quality grade (200) of the collecting bobbin (14), and the results are stored in a database (66).

9. A method for determining the bobbin mass of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that, The second machine learning program (37) has an error detection algorithm (60), and the error detection algorithm (60) reports the quality-related error (63) and / or the quality-unrelated error (62) to the quality algorithm (101).

10. A method for determining the bobbin quality of a collecting bobbin (14) according to at least one of the preceding claims, characterized in that: - providing raw yarn (3) from a supply bobbin (5), - processing the raw yarn (3) into textured yarn (30), - collecting the total error (61) recorded during the processing of the raw yarn (3), - defining the first quality grade (40) based on the total error, - collecting the textured yarn (30) on a collecting bobbin (14), - assigning the quality grade (40) to the associated collecting bobbin (14) having the textured yarn (30), - Classify the total error (61) and label it with an error quality having a quality-related error (63) and a quality-unrelated error (62), - Subtract the quality-unrelated error (62) from the total error (61) to obtain a corrected error, - Define a second quality grade (200) based on the corrected error, and - Assign and store the second quality grade (200) of the associated collecting bobbin (14).

11. An apparatus for determining the bobbin mass of a bobbin (14) containing synthetic yarn (30) for use in the method according to the preceding claims, the apparatus comprising: A sensor device (19) for collecting data of the synthetic yarn (30); A data analysis unit (18) having an error detection unit and an error classification unit (60); and a database (66) for storing the quality grade (200) of the associated collecting bobbin (14), the data analysis unit (18) interacting with a controller (16) of an operating point (1) for producing the synthetic yarn (30), wherein the error classification unit (60) interacts with a quality definition unit (100) for defining the quality grade (200).

12. The apparatus for determining the bobbin mass of a bobbin (14) containing synthetic yarn (30) according to claim 11, characterized in that, Assign a tolerance range determination unit (70) to the quality definition unit (200), and the tolerance range determination unit (70) transmits updated thresholds (OG, UG) of the tolerance range to the quality definition unit.

Images