Method for computer-aided adjustment of the configuration of different textile productions

Abstract

The invention relates to a method for computer-aided adjustment of a configuration of different textile productions, wherein production specifications are changed for different productions and a textile material is processed in at least one production step by at least one machine based on the production specifications, the method comprising the following steps: - starting the implementation of at least one or more production experiments, in which at least one production step is implemented with a setting configured and based on production specifications, - determining the influence of the configuration based on the setting in the respective production experiment, - adjusting the setting based on the determined influence in order to determine an optimized setting in terms of production specifications, - starting the implementation of a textile production, in which at least one production step is implemented with the optimized setting and based on production specifications, wherein at least one machine learning method is used in order to implement the adjustment based on the determined influence, wherein the at least one machine learning method is implemented as at least one kernel method.

Description

[0001] This invention relates to a method for computer-aided adjustment of configurations in the production of different textiles. The invention also relates to an apparatus, a system, and a computer program for this purpose.

[0002] According to existing technology, various machines are used in textile production and must be configured to process textile materials. The characteristics of the results of these production processes depend on a variety of factors. Therefore, it is impossible to find a single configuration that can provide the same quality of production for different processes. Factors affecting each production process, such as hardware wear and tear, indoor climate, fiber blends, and quality, are all different, thus the same configuration may lead to different production results.

[0003] In view of this, existing technologies have also revealed the manual determination of production configurations and inputting them into the corresponding machines. However, this is time-consuming, and the degree of optimization of the configuration may depend heavily on the experience of the operators performing the configuration.

[0004] According to document DE 10 2006 014 475 A1, the configuration can be found during shutdown testing. However, this would generate a large amount of waste. Therefore, this paper proposes a method to control the textile machine using a neural network, thereby preventing waste. This method is only applicable to the production of textile machines.

[0005] Other similar methods are known from WO 2020 / 100092 A1 and EP 3 760 772 A1.

[0006] In view of this, the object of the present invention is to overcome at least partially the above-mentioned disadvantages. In particular, the object of the present invention is to provide an improved solution for adjusting configurations for different production processes.

[0007] The features and details described in conjunction with the method of the present invention also apply to the apparatus, system and computer program of the present invention, and vice versa; therefore, the disclosure of each aspect of the invention can always be used interchangeably with reference to these inventions.

[0008] The solution of the present invention to achieve the above-mentioned objectives is, in particular, a computer-aided method for adjusting the configuration of different textile production processes, wherein, in at least one production step, textile materials, such as fibers, are processed by at least one machine, preferably used for manufacturing textile products (such as yarn), based on predetermined production values. The method can be implemented as a computer-implemented method or only partially implemented by a computer. The latter may refer to a method with several steps implemented by the at least one machine.

[0009] Furthermore, production targets can be changed for different production processes. Production targets, for example, are specifications for textile products, i.e., the corresponding production outcomes. Production targets can vary for different production processes, even though the production is still carried out using the same machinery. In this case, the textile materials and / or the operating mode of at least one machine can be changed during the production process based on the production targets. Environmental factors in the production process can also be changed. Additionally, production targets can define the composition of the textile materials, such as fiber blends.

[0010] The method includes the following steps for automating the production preparation of a corresponding textile product using modified production pre-set values, wherein these steps are preferably performed sequentially and / or repeatedly in a given order: - Initiate the implementation of at least one or more production experiments, wherein in the respective production experiments, the at least one production step is configured using settings and implemented based on the predetermined production values. - The impact of the configuration based on the stated settings was determined in the corresponding production experiments. - Adjust the settings based on the effects of the measurements to determine the settings optimized for the production target.

[0011] The settings can be adjusted for each subsequent production experiment after the first of multiple production experiments has been conducted. This allows the adjusted settings to be used to configure at least one production step in the corresponding subsequent production experiments. Thus, the final adjusted settings can only be determined as the optimized settings after the last production experiment has been conducted. This is illustrated below for at least three production experiments, wherein the following steps are performed sequentially in a given order: - First, initiate the implementation of the first production experiment out of multiple production experiments, where at least one production step is configured with initial (predefined as appropriate) settings and implemented based on predetermined production values. -The impact of the configuration based on the initial settings was then determined in the first production test. -Then the initial settings are adjusted based on the measured effects. - This initiates the implementation of the second production experiment in a series of production experiments, in which at least one production step is configured using the previously adjusted settings and implemented based on predetermined production values. -The impact of the configuration based on the previously adjusted settings was then determined in the second production test. -Then the settings are adjusted again based on the measured effects. - This initiates the implementation of the last of a series of production trials, in which at least one production step is executed using the previously adjusted settings and based on predetermined production values. -Then, in the final production test, the impact of the configuration based on the previously readjusted settings was determined. -Then the settings are adjusted again based on the measured effects. - The adjusted settings will be confirmed as optimized settings.

[0012] The method may further include a next step of (automating) production initiation of the corresponding textile production using the modified production preset values, preferably implemented after the optimized settings regarding the production preset values ​​have been determined: - Initiating the implementation of textile production, wherein at least one production step is configured with optimized settings and implemented based on predetermined production values.

[0013] The implementation of starting textile production may include the following steps: - Use optimized settings to automate the configuration of at least one machine.

[0014] Furthermore, at least one machine learning method can be used to implement adjustments based on the measured impact, preferably implemented as at least one kernel method. This allows for the automated improvement or creation of production setpoints through artificial intelligence. Subsequently, the settings adjusted by the at least one machine learning method can be applied to the next production experiment to configure the at least one machine, or ultimately determined as the optimized settings. In this way, the at least one machine learning method can control the production experiment and ultimately production.

[0015] These production experiments can be initiated and / or implemented fully automatically. Accordingly, the at least one machine learning method can autonomously control the production experiments, such as spinning trials, by adjusting settings. It can also automatically measure effects. These effects are, for example, implemented as at least one measurement result of at least one detection device (such as a sensor). Measuring effects may include, for example, evaluating the production experiment and / or the textile materials processed in the process and / or the production results and / or other factors (such as power consumption). The production experiment can be implemented using the same at least one machine (i.e., the same hardware used in production), particularly using the same textile materials (such as fiber composites) corresponding to predetermined production values ​​and / or the same environmental conditions (such as climate) in the environment of the at least one machine; therefore, the at least one machine learning method can determine optimal settings for a specific production.

[0016] The at least one machine may include a rotor spinning machine or an air-jet spinning machine. The at least one machine may also include only a single machine, such that processing is carried out in the production process, for example, solely by that machine, such as a rotor spinning machine or an air-jet spinning machine.

[0017] The corresponding production can be set to produce only a limited quantity using predetermined production values. This allows for frequent changes to the predetermined production values, for example, at most once a week, or even every few days or hours. The production preparation step can also be repeated after each change to the predetermined production value. The corresponding production can also be limited in terms of production volume and / or production time, causing production to end after exceeding predetermined quantity limits and / or production time, thus requiring another production preparation step. The advantage of this is that the settings can be optimized again when the predetermined production values ​​and / or influencing factors (such as the lifespan of at least one machine) change.

[0018] Furthermore, according to the invention, the settings include at least one or more setting values ​​for the at least one machine, preferably a spinning machine and / or a winding machine, to control the processing of the textile material in the at least one production step, preferably the processing of fibers into yarn (also called yarn) and / or the rewinding and / or winding of the yarn. For this purpose, the setting values ​​can be transmitted to the machine, for example, via a data connection (such as a network), thereby configuring the machine and the production steps it performs.

[0019] Furthermore, the at least one or more set values ​​may include at least one of the following: draw-out speed, rotational speed of the machine's rotor, negative pressure, stress of the textile material and / or production result during winding, type and / or option of the textile material, particularly the winding speed of the yarn winding device, settings for the cleaner, and particularly settings for the limit values ​​that the cleaner can use to cut the textile material. The cleaner may preferably be implemented as or include sensors for performing quality inspection measurements on the yarn. The cleaner may also compare the measured values ​​with at least one limit value to identify limit violations when they are exceeded and / or fall below the limit value. This allows for ensuring sufficient yarn quality based on comparison. When a limit value is violated, the cleaner may, as appropriate, send a signal to initiate the cutting of the erroneous yarn segment.

[0020] Optionally, determining the effect can be implemented by receiving at least one detection result by at least one detection device. The detection device may be part of the machine or arranged on the machine. The detection device may be implemented to detect the measured variable affected by the setting (i.e., the effect of the setting). Accordingly, the detection result may be generated from the detection, particularly measurement and / or analysis, performed on the machine and / or the textile material and / or in the environment of the machine and / or on the production result (preferably yarn) of the corresponding production experiment.

[0021] The at least one detection device may include a capacitive sensor to measure the capacitance of the volume through which the textile material and / or production result (such as yarn) passes. This allows the detection of the amount passing through. This also allows inference of whether details appear on the yarn and / or the size of those details. Alternatively or supplementarily, the at least one detection device may also include an optical sensor. The sensor's light-emitting diode can illuminate the textile material and / or production result (such as yarn), and the sensor measures the arriving light, thereby determining, for example, the yarn diameter. Similarly, the at least one detection device may also include analytical equipment for detailed analysis of the textile material and / or production result. Furthermore, the at least one detection device may be implemented to perform length and / or weight measurements to determine the length-to-weight ratio. The at least one detection device may also have at least one optical sensor for detecting foreign fibers. Furthermore, the at least one detection device may be implemented to detect production information, such as the frequency of yarn breakage during manufacturing and / or the frequency at which yarn must be re-started. Therefore, this production information may also be provided as a result of the at least one detection. The at least one detection device may also be implemented to determine the yarn twist and / or tension.

[0022] Preferably, the plurality of production experiments can be performed sequentially, wherein the effect is measured for at least one of the production experiments, and the settings are subsequently adjusted based on the measured effects so that the configuration is performed with the adjusted settings in the next production experiment. This allows the production experiments to be organized sequentially and controlled by adjustment steps. Alternatively, the configuration can be performed fully automatically, for example, by transmitting the setting data to the corresponding machine. The necessary data can also be collected by the machine itself for configuration.

[0023] For example, the plurality of production experiments can be carried out sequentially, wherein in each production experiment, the settings are changed by the adjustments and the effects measured in the process are used as input values ​​for the machine learning method, so as to preferably optimize the settings specifically by means of at least one target predetermined value. In the plurality of production experiments carried out for production preparation for each production, the same production predetermined value is used, which is also applied to subsequent production; therefore, production-specific optimization is involved here. The at least one target predetermined value may also include, for example, a predefined optimization objective. Advantageously, the at least one target predetermined value may include at least one of the following: maximizing energy efficiency, preferably maximizing energy efficiency by generating negative pressure and / or overpressure in the machine, maximizing the production result, particularly the quality of the yarn, maximizing production speed, minimizing yarn breakage, optimizing the efficiency of the machine, and checking the effect of fancy yarn.

[0024] Preferably, the plurality of production experiments can be performed sequentially, wherein the effect is measured in the last of the production experiments for the production preparation of the corresponding production, and the settings are subsequently adjusted based on the measured effect so that the adjusted settings are used as the optimized settings. This concludes the configuration adjustment.

[0025] In another embodiment, the textile material is processed in the corresponding production experiment and the corresponding production to produce a production result, particularly yarn. The production result of the corresponding production experiment may at least partially or completely result in waste, thus differing from the production result of the corresponding production. In other words, the purpose of the production experiment may be to produce waste, i.e., production results that must be discarded, such as yarn. This is acceptable in order to find the optimal settings for production. Advantageously, the waste, i.e., the production result of the production experiment, includes at most 1% or at most 0.1% of the production result of the production.

[0026] Furthermore, advantageously, the textile production is carried out using at least one of the same machines, but differs in the predetermined production values, preferably in the production results and / or the requirements for the production results and / or the parameterization and / or formulation of the textile materials and / or the parameters of the textile materials, preferably in the fiber number and / or fiber strength and / or fiber blend. Moreover, in the first production experiment of the production preparation for carrying out the corresponding production, a predefined initial setting is used as the setting, which is only changed after the first production experiment is completed. Preferably, correspondingly specific optimized settings for other productions are not considered. This allows optimization to be limited to the corresponding production. Thus, it ensures that settings can be optimized even for drastically different productions.

[0027] Optionally, the at least one or exactly one machine learning method can be implemented as a kernel method, and preferably, the kernel method is implemented as one of the following: support vector machine, Gaussian process, kernel principal component analysis, kernel perceptron, canonical correlation analysis, ridge regression, spectral clustering, or linear adaptive filter. The advantage of the above techniques is their ability to quickly and accurately estimate error sources. A key advantage of the above techniques and algorithms is their stable optimization even with incomplete information. Therefore, optimal results can be determined with fewer overall iterative experiments.

[0028] Furthermore, a specific proportion of training data can be produced for the training settings of the at least one machine learning method. This training data is determined solely from the at least one or more production experiments in the production preparation for the corresponding production. Preferably, the training data used for the production includes both defined initial training data, particularly predefined initial settings, and training data that changes based on the measured effects. This ensures that the settings can be optimized even for drastically different productions.

[0029] Furthermore, optionally in this invention, the at least one or exactly one machine learning method is implemented as a Gaussian process. This allows for optimization of settings even when many different production processes are involved, using limited training data.

[0030] Another advantage of this invention is that the steps for automated production preparation and / or production startup are implemented in a fully automated manner, and / or the methods are at least partially or entirely computer-implemented and preferably implemented in real time. This allows for configuration adjustments without the need for intervention from experienced users.

[0031] Furthermore, according to the invention, in the corresponding production experiment, the at least one production step is implemented in the form of at least one or exactly one spinning experiment, wherein yarn is spun based on the settings so that the settings for the corresponding production are optimized by adjustment based on the production experiment.

[0032] The at least one machine, particularly a textile machine, advantageously includes at least one of the following: a spinning machine, a textile machine for fiber preparation, a textile machine for spinning preparation, a ring spinning machine, a compact spinning machine, a rotor spinning machine, an air-jet spinning machine, an automated machine, a winding machine, and a texturing machine. Furthermore, the textile machine may have multiple adjacently arranged workstations. These workstations may be, for example, spinning workstations or similar. These workstations may also be configured as winding workstations for winding yarn onto yarn bobbins (such as cross bobbins). This is, for example, producing yarn from an upstream fiber composite material on a spinning machine, which, after leaving the spinning unit, sequentially passes through a take-up device and a yarn storer along the yarn path, and finally is wound onto a yarn winding device, i.e., the winding workstation.

[0033] At least one detection device, such as at least one sensor, can be used to monitor production. This detection device can, for example, determine at least one detection result by detecting textile materials and / or production results on a spinning machine and / or at least another machine used for production preparation and / or production. The at least one detection result is implemented, for example, as a measurement value or similar. The detection can be repeated by the detection device during production preparation and / or production, and thus this detection can also be referred to as monitoring of production preparation and / or production. For example, in a known scheme, to monitor the yarn path at each station of a spinning machine, a so-called yarn monitor is arranged within the area of ​​the yarn path, for example, between the spinning unit and the yarn winding device on a spinning machine. This yarn monitor can be a yarn sensor connected to the control device of that station, which identifies accidental yarn breakage and errors in the spun yarn. In both cases, the corresponding work process can be interrupted by the control device of that station. Furthermore, when the yarn is damaged, yarn breaking can be implemented to remove the damaged yarn area. When the yarn breaks, a section of the yarn end can be removed. In both cases, the yarn end can then be reconnected. Subsequently, on the spinning machine, for example, the yarn end originating from the yarn winding device is fed to the spinning unit and the yarn start-up operation is repeated, after which the spinning process can be restarted. During the operation of the at least one spinning machine and / or at least one machine used for production preparation and / or production, i.e., during production or production preparation, the at least one detection device can measure the at least one detection result. Furthermore, the connection position can also be generated at the location where the yarn start-up operation is repeated, which can also be an optimization target.

[0034] The at least one test result can be at least one of the following: physical and / or chemical parameters of the at least one textile machine, such as rotational speed, power consumption, and the chemical properties of the lubricant. Furthermore, the at least one test result also includes at least one physical and / or chemical parameter of the textile material, such as thickness, density, fiber chemical properties, and the like. Testing the parameters of the textile machine and / or textile material can include sampling from the textile machine and / or textile material and examining the samples in a laboratory, particularly to determine the mechanical / physical and / or chemical properties of the samples.

[0035] The production preparation may use the same production line and / or at least one textile machine or part thereof as the production. Therefore, the characteristics described for production also apply to production preparation. The production and the production preparation may include one or more production steps. Particularly when processing fibers (such as natural fibers and chemical fibers and their mixtures) into yarn, the production steps may include at least one of the following: converting the fibers into sliver, drafting, spinning, automatically conveying from one textile machine to the next, rewinding, texturing, and winding. To achieve the desired production capacity along the serial production line, the textile machines or components thereof may be arranged in parallel. This allows one production step to follow another, where the previous production step may require fewer or more parallel textile machines or components to process the textile material compared to the next production step. For example, the fiber preparation step providing the sliver may require fewer parallel textile machines or components compared to the step of spinning the sliver into yarn.

[0036] The at least one test result may include at least one of the following: machine power consumption, lubricant quality, diameter of textile material, fiber density of textile material, results of laboratory analysis of textile material, fiber moisture, fiber fineness, fiber length, yarn path, yarn error, fiber length uniformity, fiber strength, fiber knots, fiber maturity, fiber color, fiber waste, sliver / roving count / ply count, sliver / roving weight deviation, sliver / roving uniformity, thick places, thin places, twist, yarn strength, yarn elongation characteristics, strength, twist and count, yarn hairiness, yarn abrasion resistance, wear characteristics, yarn color, machine environmental parameters (especially climatic conditions), frequency of textile material and / or production results (especially yarn) breakage and / or yarn error, and others. The test results may, for example, be obtained by sampling the lubricant to determine lubricant quality. Test results for textile machines and / or textile materials may also be based on external information providers, such as quality data from textile material suppliers (e.g., raw cotton quality information), model names / product numbers and / or material characteristics and / or expiration dates of consumables such as lubricants. Environmental parameters, particularly those relating to climatic conditions, such as temperature, humidity, solar radiation, solar altitude, and / or date and / or time, preferably measured or based on information from external providers, can also be set as detection results. This can refer to climatic conditions in raw material warehouses and / or intermediate product warehouses and / or finished product warehouses. For example, climatic condition parameters can also be detected in at least one area of ​​a textile factory.

[0037] The present invention also relates to an apparatus for performing data processing, comprising components for carrying out the steps of the method of the present invention. In view of this, the apparatus of the present invention has the advantages described in detail with respect to the method of the present invention. The apparatus may be at least one of the at least one machine or part of exactly one machine.

[0038] The present invention also relates to a system for carrying out textile production, comprising: - At least one machine for processing textile materials in at least one production step. -The apparatus for performing data processing according to the present invention - At least one detection device for determining the influence by detecting at least one detection result on the machine and / or the textile material and / or the environment of the machine and / or the production results.

[0039] In view of this, the system of the present invention has the advantages described in detail with respect to the method of the present invention.

[0040] The present invention also relates to a computer program, and more particularly to a computer program product, comprising a plurality of instructions which, when executed by a computer, cause the computer to implement the method of the present invention. In view of this, the computer program of the present invention has the advantages described in detail with respect to the method of the present invention.

[0041] The computer may refer to a data processing device, such as the device of the present invention, which executes the computer program. The computer may have at least one processor for executing the computer program. It may also include a non-volatile data memory in which the computer program can be stored, from which the processor can read and execute the computer program. The computer program may also be directly integrated into the machine software of the machine.

[0042] The computer may include at least one integrated circuit, such as a microprocessor, application-specific integrated circuit (ASIC), application-specific standard product (ASSP), digital signal processor (DSP), field-programmable gate array (FPGA), or similar. The computer may also have at least one interface for data exchange, such as an Ethernet interface, a LAN (local area network) interface, a WLAN (wireless local area network) interface, a system-on-a-chip (SoC) interface, or other radio interfaces such as Bluetooth or near-field communication (NFC). The computer may also be implemented as one or more controllers, i.e., as a system composed of several controllers. The computer may also be located in the cloud and / or configured as a server to provide data processing for local applications via an interface. The computer may also be implemented as a mobile device, such as a smartphone.

[0043] The present invention also relates to a computer-readable storage medium comprising the computer program of the present invention. The storage medium is configured, for example, as a data storage device, such as a hard disk and / or non-volatile memory and / or a memory card. The storage medium may, for example, be integrated into the computer.

[0044] Further advantages, features, and details of the present invention are described in detail below with reference to the accompanying drawings. Within the scope of the present invention, the features mentioned in the claims and the specification can be applied individually or in any combination. (See figures:) Figure 1 This is a schematic diagram illustrating an embodiment of the method of the present invention. Figure 2 This is a schematic diagram illustrating an embodiment of the system, apparatus, and computer program of the present invention.

[0045] In the figures below, the same technical features in different embodiments are indicated by the same reference numerals.

[0046] Figure 1 An embodiment of a method 100 for computer-aided adjustment of the configuration of different textile production processes 210 is shown. Multiple production processes 210 can be sequentially implemented, indicated by a time axis t, wherein a predetermined production value 230 is changed for each production process 210 (as an example, the changed predetermined production value 230 is schematically shown as an arrow on the time axis t). Furthermore, in these production processes 210, textile materials, such as fibers, are processed at least once in at least one production step 220 based on the predetermined production value 230 by at least one machine 10 for manufacturing textile products (such as yarn). Further details regarding the processing of textile materials and the machine 10 can be found in [reference needed]. Figure 2 .

[0047] In addition, such as Figure 1As shown, method steps 101, 102, 103, and 104 are provided to prepare for and start automated production of a corresponding textile production 210 using a modified production predetermined value 230. According to the first method step 101, at least one or more production experiments 215 are initiated. In the corresponding production experiment 215, the at least one production step 220 can be configured with a setting 240 and implemented based on the production predetermined value 230. Subsequently, according to the second method step 102, the effect 250 of the configuration based on the setting 240 in the corresponding production experiment 215 is measured. According to the third method step 103, the setting 240 is adjusted based on the measured effect 250 to determine an optimized setting 240 in terms of the production predetermined value 230. Furthermore, dashed arrows indicate that method steps 101 to 103 can be implemented iteratively. That is, after the first production experiment 215 is completed, the effect 250 is first measured to adjust the setting 240 accordingly. The adjusted setting 240 can then be applied to implement the second production experiment 215. The influence 250 can be remeasured so that the setting 240 can be readjusted and applied to the third production experiment 215 to configure the configuration of the at least one production step 220. Optionally, this process can be repeated for more production experiments 215. Thus, these production experiments 215 are performed sequentially, with the influence 250 measured and the setting 240 adjusted between them, including the last production experiment 215 for the production preparation of the corresponding production 210. Finally, the influence 250 is measured and the setting 240 is subsequently adjusted based on the measured influence 250 so that the adjusted setting 240 is used as the optimized setting 240.

[0048] With the final iteration of step 103 completed, production preparation is complete, and the determined optimized settings 240 can be applied to the fourth method step 104 when textile production 210 is started. In textile production 210, the at least one production step 220 is also implemented based on the production predetermined value 230, but configured with optimized settings 240.

[0049] Furthermore, as shown in the figure, at least one or exactly one machine learning method 300 can be used to implement adjustment 103 based on the measured effect 250. Preferably, the machine learning method 300 is implemented as a kernel method 300.

[0050] Figure 2The method 100 of the present invention is illustrated using a specific textile machine 10 as an example. The textile machine 10 may be configured as part of a system 1, which also includes a data processing device 20, a computer program 30, and at least one detection device 15. This detection device determines the influence 250 by detecting results on the machine 10 and / or the textile material 2 and / or the environment of the machine 10 and / or the production result 3. In this specific example, the textile machine 2 may include fibers, particularly fiber composite materials, and the production result 3 may include yarn. Figure 1 In the corresponding production experiments 215 and 210 shown, the textile material 2 can be processed to produce production result 3, wherein the production result 3 of production experiment 215 is at least partially or completely waste, and therefore differs from the production result 3 of production experiment 210. The same [material name] is used in both production experiment 215 and production 210. Figure 2 The machine 10 shown is optimized under the actual conditions present in production 210. Furthermore, more textile production 210 can be processed using the same machine 10, but with respect to the production predetermined value 230, preferably differing in terms of the production result 3 and / or the requirements for the production result 3 and / or the parameterization and / or formulation of the textile material 2 and / or the parameters of the textile material 2, preferably in terms of the fiber number and / or fiber strength.

[0051] Specifically, the machine learning method 300 (KI) is first provided with initial settings 240 for production 210. The adjusted settings 240 can then be transmitted to connected spinning or winding stations via KI to spin and / or wind yarn 3. Subsequently, detection devices 15, particularly sensors, can detect the results of these production experiments 215, such as the yarn 3, as influences 250 and feed them back to KI. KI can derive new settings 240 from this dataset and implement another production experiment 215. Over time, KI continuously improves the settings. KI aims at various standards, such as improving yarn quality, minimizing yarn breaks, improving energy efficiency, or increasing production speed.

[0052] These method steps or computer programs 30 used to execute these method steps can be implemented in real time. This allows for the fully automated, real-time application of the aforementioned KI. Thus, for example, on the user side, KI can be used before each new production run 210 to determine the optimized settings 240. That is, the settings 240 with relevant settings are limitedly optimized for the specific production run 210 and are no longer needed after production run 210 is completed (i.e., after several hours). Therefore, the training time for KI is short, requiring only a small amount of training data from the production experiment 215. This provides an optimization process suitable for determining the optimized settings 240 in a very short time using a small amount of training data for each production preparation. When this occurs on the user side, it also facilitates the automated implementation of the necessary production experiments 215 on the machine 10. Another advantage is the automated transfer of samples from the production experiment 215 to laboratory equipment for measuring the effects 250.

[0053] Various methods for implementing KI (Kernel Instruction) using neural networks have been disclosed. However, due to the limited availability of training data, the present invention preferably employs kernel methods. Therefore, for example, a Gaussian process can be used as KI. Unlike neural networks, these conventional machine learning models can be reliably optimized with less training data. This limits the use of optimization methods to shorter production processes and the user side. Similarly, the present invention does not require the use of a neural network comprising at least one machine learning method 300, i.e., any neural network.

[0054] The apparatus 20 may be part of one or more machines 10 for producing textiles, in order to perform method steps 101-104 on the user side and to be initiated by the user. During production 210, a large number of textile machines 10 may be used to convert, for example, natural fibers and chemical fibers and their mixtures into yarn of a desired type, quantity, and quality. The desired type and / or quantity and / or quality and / or mixture may be defined, for example, by a predetermined production value 230 and vary for different production runs 210. Different types of textile machines 10 may be provided for production, such as winding machines or spinning machines. In particular, air-jet spinning machines and rotor spinning machines as disclosed in principle by the prior art.

[0055] It has also been disclosed that measurements are performed on the textile machine 10 and the processed textile material 2 to monitor production. For this purpose, a detection device 15 can be used, such as... Figure 2 The two sensors 15 are shown. Figure 2The spinning station 11 is schematically shown, having yarn 3 extending from spinning unit 21 to yarn winding device 13 during the spinning process. This spinning unit is, for example, an air-jet spinning unit or a rotor spinning unit. After passing through a drafting device (not shown) in the case of an air-jet spinning machine or through a carding roller unit in the case of a rotor spinning machine, the fiber composite material 2 input to spinning unit 21 is twisted inside spinning unit 21. With respect to the yarn running direction 22, a drawing device (indicated by drawing roller 19) is provided downstream of spinning unit 21, which draws the yarn 3 discharged from spinning unit 21 from spinning unit 21 by means of drawing roller pair 19 and conveying it along direction 22 to yarn winding device 13. Downstream of the yarn feeding device, a yarn storage device 17 with a yarn storage tube (dashed line) is provided. First and second yarn sensors 15 are arranged in the area of ​​the yarn inlet (not shown). The yarn storage tube 17 is arranged such that the yarn sensors 15 are positioned within the area of ​​the yarn path. Therefore, the yarn sensors 15 form a sensing mechanism for detecting the yarn path. This allows yarn errors and / or yarn breakage to be identified as detection results. Loop-shaped yarn segments arranged inside the yarn storage tube are generated by suction air directed from the yarn inlet into the yarn storage tube. This suction air is provided by a negative pressure source 18 connected to the end of the yarn storage tube 17 opposite the yarn inlet. The yarn sensors 15 thus detect the extent of the loop-shaped yarn segments within the yarn storage tube. When a yarn breakage occurs or the output sensor 15 detects a yarn error, the spinning process is interrupted via a connected control device. Subsequently, a defined free yarn end originating from the yarn winding device 13 is generated using the separation device 16, which is then fed into the spinning unit 21 via the return member 14 in the next step. Here, the yarn end is regrowth, even though it is connected to the yarn 3 originating from the spinning unit 21. Figure 2 The diagram also shows a driver 12, which is used to specify the winding speed of the yarn winding device 13.

[0056] To adjust 103 using the detection results provided by the detection device 15, the at least one detection result and / or the at least one setpoint and / or the at least one target predetermined value can be transferred as input values ​​to the at least one machine learning method 300, thereby obtaining at least one optimized setpoint as the output value of the machine learning method 300. To use the machine learning method 300, a dataset can be prepared from the detection results and used as input values ​​for the method 300. To prepare the dataset, for example, the at least one setpoint and the at least one target predetermined value can be stored in a file first. Preferably, a target predetermined value is predefined for each detection result, indicating which value of the detection result or which target of the detection result the at least one initial setpoint should be optimized for (e.g., maximized). Specifically, this can, for example, improve energy efficiency, improve yarn quality, or increase production speed, or combine the advantages of both. The target predetermined value for production speed, for example, defines a duration or a minimization of the duration.

[0057] The following describes the application of method 300 to multiple setpoints and test results, using a Gaussian process model (hereinafter referred to as Gaussian process) as an example. Like linear regression models, tree-based models, or perceptron-based models, the Gaussian process model is a machine learning method. Its advantage lies in its analytical and problem-solving capabilities. It is then shown that optimizing the setpoints of the textile machine 10 can surprisingly improve production 210.

[0058] During production preparation, this dataset can always contain (adjusted as needed) the current setpoint of setting 240 and the test results related to it and repeatedly measured according to influence 250. Therefore, the setpoint can be determined by the function f(x) n The target value y is defined. n , where x n The input values ​​x = [x1, x2, ... can be formed from the detection results.] ] T e n It is independent Gaussian noise:

[0059] Furthermore, it can be seen from f(x) n ) Observational conditional probability y n Used to determine the normal distribution:

[0060] Where σ is e n The standard deviation, σ = σ| is a variable The diagonal matrix is ​​given. To predict y, the marginal probability distribution p(y) can be determined. This probability distribution can be obtained by marginalizing the conditional distribution p(y|f(x)) through the distribution p(f(x)):

[0061] The distribution p(f(x)) is defined as having a mean of 0 and a covariance kernel matrix K that is a variable. Gaussian distribution:

[0062] in

[0063] Therefore, we can conclude that:

[0064] Each element in C can be obtained using the following formula: .

[0065] Different covariance kernel functions can be applied to K or k(x) n , x m Examples of kernel types include constant kernels, quadratic exponential kernels, and periodic kernels. A quadratic exponential kernel can be calculated from paired samples (xn, xm) in x:

[0066] The initial sample can be obtained from the initial settings.

[0067] To optimize the settings, existing ones can be used. Input values ​​x = [x1, x2, ... ] T And the known target value sy = [y1, y2, ... ] T To predict +1 The value, which is equivalent to the new input value. +1 This new input value can be equivalent to the target predetermined value. The parameter p(y') can be determined based on the distribution p(y'). +1 |y), where y' = [y1, y2,... , +1 ] T It is length A vector. This produces p(y') = N(y'|0, C'), where the new covariance matrix C' has the structure of variables Where C = K + σ| is the initial Covariance matrix, k is the length A vector whose elements are given by the following equation: k[n] = k(x n , +1 ), c is the containment +1 The scalar of its own covariance: c = k( +1 , +1 )+σ. Then, based on the value predicted in the above manner, 103 can be set to 240.

[0068] The above description of the embodiments is only to illustrate the invention within the scope of a few examples. Where technically significant, the features of these embodiments can be freely combined without departing from the scope of the invention.

[0069] List of icon symbols

[0070] 1 system

[0071] 2. Textile materials and fiber composite materials

[0072] 3. Production results, yarn

[0073] 10 machines

[0074] 11 workstations, spinning workstations

[0075] 12 drives

[0076] 13 Yarn winding device

[0077] 14 Return Components

[0078] 15. Detection device, yarn sensor

[0079] 16 Separation Device

[0080] 17 Yarn storage tube

[0081] 18 negative pressure sources

[0082] 19 lead-out rollers

[0083] 20 devices

[0084] 21 Spinning Unit

[0085] 22. Yarn running direction

[0086] 30 Computer Programs

[0087] 100 methods

[0088] 101 First method step, start

[0089] 102. Second method step: determination

[0090] 103 Third method step, adjustment

[0091] 104. Fourth method step: Start

[0092] Production 210

[0093] 215 Production Experiment

[0094] 220 Production Steps

[0095] 230 production target

[0096] 240 settings

[0097] 250 impact

[0098] 300 machine learning methods, kernel methods

[0099] t time

Claims

1. A method (100) for computer-aided adjustment of the configuration of different textile productions (210), wherein a production predetermined value (230) is changed for the different textile productions (210), and wherein textile materials (2) are processed by at least one machine (10) based on the production predetermined value (230) in at least one production step (220), the method comprising the steps of automating production preparation and production start-up of the corresponding textile productions (210) with the changed production predetermined value (230): - Initiate (101) the implementation of one or more production experiments (215), wherein in the corresponding production experiment (215), the at least one production step (220) is configured with settings (240) and implemented based on the production predetermined value (230). -The effect of the configuration based on the settings (240) was determined (250) in the corresponding production experiment (215) (102). - Adjust (103) the setting (240) based on the measured effect (250) to determine the setting (240) optimized in terms of the production predetermined value (230). - Initiate (104) the implementation of the textile production (210), wherein the at least one production step (220) is configured with the optimized settings (240) and implemented based on the production predetermined value (230). At least one machine learning method (300) is used to implement the adjustment (103) based on the measured effect (250).

2. The method (100) according to claim 1. Its features are, The settings (240) include one or more settings for the at least one machine (10) in the form of at least one textile machine, so as to control the processing of the textile material (2) in the at least one production step (220).

3. The method (100) according to claim 1 or 2. Its features are, The determination (102) of the effect (250) is carried out by receiving at least one detection result by at least one detection device (15), wherein the detection result is generated by the detection of the corresponding production experiment (215) on the machine (10) and / or the textile material (2) and / or in the environment of the machine (10) and / or on the production result (3).

4. The method (100) according to claim 1. Its features are, The plurality of production experiments (215) are carried out in sequence, wherein the effect (250) is measured (102) for at least one of the production experiments (215), and the setting (240) is then adjusted (103) based on the measured effect (250) so that the configuration is carried out with the adjusted setting in the next production experiment of the production experiment (215).

5. The method (100) according to claim 1. Its features are, The plurality of production experiments (215) are carried out in sequence, wherein in the production experiments (215), the settings (240) are changed by the adjustment (103) and the effects (250) measured in the process are used as input values ​​for the machine learning method (300) so as to produce specifically optimized settings (240) with the aid of at least one target predetermined value, wherein the at least one target predetermined value includes at least one of the following: energy efficiency, quality of production results (3), and production speed.

6. The method (100) according to claim 1. Its features are, The plurality of production experiments (215) are carried out in sequence, wherein the effect (250) is measured (102) in the last of the production experiments (215) for the production preparation of the corresponding textile production (210), and the setting (240) is then adjusted (103) based on the measured effect (250) so that the adjusted setting (240) is used as the optimized setting (240).

7. The method (100) according to claim 1. Its features are, In the corresponding production experiment (215) and the corresponding textile production (210), the textile material (2) is processed to produce a production result (3), wherein the production result (3) of the corresponding production experiment (215) is at least partially or completely waste, and thus different from the production result (3) of the corresponding textile production (210).

8. The method (100) according to claim 1. Its features are, The textile production (210) is carried out using at least one of the same machines (10), but differs in terms of the production predetermined value (230), wherein a predefined initial setting (240) is used as the first production experiment in the production experiment (215) for production preparation of the corresponding textile production (210), and the initial setting is changed only after the first production experiment (215) is completed, wherein the corresponding specific optimized settings (240) of other productions (210) are not considered.

9. The method (100) according to claim 1. Its features are, At least one of the machine learning methods (300) is implemented as a kernel method (300), and the kernel method (300) is implemented as one of the following: support vector machine, Gaussian process, kernel principal component analysis, kernel perceptron, canonical correlation analysis, ridge regression, spectral clustering, linear adaptive filter, and / or For the training settings of the at least one machine learning method (300), a specific proportion of training data is produced, the proportion being determined only from the one or more production experiments (215) of the production preparation for the corresponding textile production (210), wherein the training data for the textile production (210) includes both defined initial training data and training data that changes based on the measured influence (250).

10. The method (100) according to claim 1. Its features are, At least one of the machine learning methods (300) is implemented as a Gaussian process.

11. The method (100) according to claim 1. Its features are, The steps and / or methods (100) for automated production preparation and start-up are implemented in a fully automated manner and are partially or entirely computer-based.

12. The method (100) according to claim 1. Its features are, In the corresponding production experiment (215), the at least one production step (220) is implemented in the form of at least one spinning experiment, wherein the yarn is spun based on the settings (240) so that the settings for the production of the corresponding textile (210) are optimized by the adjustment (103) based on the production experiment (215).

13. The method (100) according to claim 2. in, The one or more settings include at least one of the following: draw-out speed, rotational speed of the machine's rotor, negative pressure, stress of the textile material (2) and / or production result (3) during roll-up, type and / or option of the textile material (2), and settings for the cleaner.

14. The method (100) according to claim 3. in, The at least one test result includes at least one of the following: the rotational speed of the rotor of the machine (10), the power consumption of the machine (10), the quality of the lubricant, the diameter of the textile material (2), the fiber density of the textile material (2), the results of laboratory analysis of the textile material (2), the environmental parameters of the machine (10), the frequency of breakage of the textile material (2) and / or the frequency of yarn errors in the production results (3).

15. The method (100) according to claim 1. Its features are, The steps and / or methods (100) for automated production preparation and start-up are implemented in full automation, at least partially or completely in real time.

16. An apparatus (20) for performing data processing, comprising components for implementing the steps of the method (100) according to any one of claims 1 to 15.

17. A system (1) for carrying out textile production (210), comprising: - At least one machine (10) for processing textile material (2) in at least one production step (220). - The apparatus (20) for performing data processing according to claim 16. - At least one detection device (15) is used to determine (102) the effect (250) by detecting at least one detection result on the machine (10) and / or the textile material (2) and / or the environment and / or production results (3) of the machine (10).

18. A computer program product comprising a plurality of instructions, which, when executed by a computer, cause the computer to perform the steps of the method (100) according to any one of claims 1 to 15 for automated production preparation and production initiation.

Images