Information processing device, estimation method, estimation program, and program product

Abstract

This invention provides a technique for estimating the dyeing quality of false-twisted yarn using the raw yarn before the false-twisting process. [Solution] An information processing device capable of estimating the dyeing quality of false-twisted yarn, which is manufactured by false-twisting a raw yarn, comprises a control device. The control device performs a process to acquire an estimation model generated by machine learning the correlation between explanatory variables and target variables defined in each of a plurality of training data sets, the explanatory variables include a plurality of physical properties measured with respect to the training yarn. The target variable includes the dyeing quality measured with respect to the training false-twisted yarn manufactured by false-twisting the training yarn. The device performs a process to acquire a plurality of physical properties with respect to the target yarn, and a process to input the plurality of physical properties acquired with respect to the target yarn into the estimation model and output the dyeing quality obtained from the estimation model.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus, an estimation method, and an estimation program.

Background Art

[0002] Japanese Patent Application Laid-Open No. 2024-125173 (Patent Document 1) discloses an abnormality determination apparatus that can accurately determine abnormalities in POY (Partially Oriented Yarn) manufactured by a spinning and take-up machine.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Draw textured yarn (DTY) is manufactured by texturing POY as a raw yarn. At the manufacturing site, various quality tests are conducted to check the quality of the draw textured yarn. As an example, the dyeing quality of the draw textured yarn is confirmed by a dyeing test.

[0005]

[0006] In order to confirm the dyeing quality of the draw textured yarn, it is necessary to perform a raw yarn manufacturing process, a texturing process, a sample knitting process, a dyeing process, and a quality confirmation process. Thus, a large amount of man-hours is required to confirm the dyeing quality from the actually manufactured draw textured yarn. As a result, the cost for storing the yarn to be quality inspected has become enormous.

Means for Solving the Problems

[0007] One example of this disclosure provides an information processing device capable of estimating the dyeing quality of false-twisted yarn, which is produced by false-twisting a raw yarn. The information processing device includes a control device. The control device performs a process to obtain an estimation model generated by machine learning the correlation between explanatory variables and a target variable defined in each of a plurality of training data. The explanatory variables include a plurality of physical properties measured with respect to the training yarn. The target variable includes the dyeing quality measured with respect to the training false-twisted yarn produced by false-twisting the training yarn. The device performs a process to obtain a plurality of physical properties with respect to the target yarn, and a process to input the plurality of physical properties obtained with respect to the target yarn into the estimation model and output the dyeing quality obtained from the estimation model.

[0008] This allows the information processing device to estimate the dye quality of the false-twisted yarn from the physical properties of the raw yarn. In other words, the dye quality of the false-twisted yarn can be estimated without manufacturing the false-twisted yarn from the raw yarn. As a result, the amount of work required to verify dye quality is significantly reduced. In addition, the amount of false-twisted yarn stored for quality inspection is reduced, lowering storage costs. Furthermore, by using multiple different physical properties of the raw yarn as explanatory variables in the estimation of dye quality, the accuracy of the dye quality estimation is further improved.

[0009] In one example of this disclosure, the above-mentioned multiple physical properties include the thermal stress relating to the learning filament and the boiling water shrinkage rate relating to the learning filament. The above-mentioned multiple physical properties obtained with respect to the filament being estimated include physical properties of the same type as the explanatory variables.

[0010] By learning the thermal stress and boiling water shrinkage rate of the raw yarn as explanatory variables, the accuracy of estimating dyeing quality is further improved.

[0011] In one example of this disclosure, the explanatory variables further include physical properties measured for the learning false-twist yarn during the false-twist process.

[0012] By learning the physical properties related to false-twisted yarn as explanatory variables, the accuracy of estimating dyeing quality is further improved.

[0013] In one example of this disclosure, the false twisting process includes the steps of heating the conveyed yarn, stretching the conveyed yarn, and twisting the conveyed yarn. The physical properties of the false twisted yarn for learning include the tension measured on the false twisted yarn while it is being conveyed after the twisting step.

[0014] By learning the tension measured for false-twisted yarn during transport as an explanatory variable, the accuracy of estimating dyeing quality is further improved.

[0015] In one example of this disclosure, the machine learning process includes selecting a physical property that correlates with the objective variable from among a plurality of candidate physical properties relating to the yarn used for learning as the explanatory variable.

[0016] This eliminates explanatory variables that are not correlated with the target variable during training, further improving the accuracy of staining quality estimation.

[0017] In one example of this disclosure, the machine learning further includes a process of learning the correlation between the explanatory variables selected by the selection process and the target variable using multiple regression analysis.

[0018] This allows for a simpler construction of the estimation model, as the multiple regression analysis is performed after excluding explanatory variables that are not correlated with the dependent variable.

[0019] Another example of this disclosure provides an estimation method for estimating dye quality for false-twisted yarn produced by false-twisting a yarn. The estimation method comprises the step of obtaining an estimation model generated by machine learning the correlation between explanatory variables and a target variable, each defined in a set of training data. The explanatory variables include a set of physical properties measured with respect to the training yarn. The target variable includes dye quality measured with respect to the training false-twisted yarn produced by false-twisting the training yarn. The estimation method further comprises the step of obtaining a set of physical properties with respect to the yarn to be estimated, and the step of inputting the set of physical properties obtained with respect to the yarn to be estimated into the estimation model to output the dye quality obtained from the estimation model.

[0020] The above estimation method allows the dye quality of false-twisted yarn to be estimated from the physical properties of the raw yarn. In other words, the dye quality of false-twisted yarn can be estimated without manufacturing false-twisted yarn from the raw yarn. As a result, the man-hours required to confirm dye quality are significantly reduced. In addition, the amount of false-twisted yarn stored for quality inspection is reduced, thus lowering storage costs. Furthermore, by using multiple different physical properties of the raw yarn as explanatory variables in the estimation of dye quality, the accuracy of the estimation is further improved.

[0021] Another example of this disclosure provides an estimation program for estimating the dye quality of false-twisted yarn produced by false-twisting a yarn. The estimation program causes an information processing device to perform a process to obtain an estimation model generated by machine learning the correlation between explanatory variables and a target variable defined in each of several training data sets. The explanatory variables include several physical properties measured with respect to the training yarn. The target variable includes the dye quality measured with respect to the training false-twisted yarn produced by false-twisting the training yarn. The estimation program further causes the information processing device to perform a process to obtain several physical properties with respect to the yarn to be estimated, and a process to input the obtained physical properties with respect to the yarn to be estimated into the estimation model and output the dye quality obtained from the estimation model.

[0022] Based on the above estimation program, the dyeing quality of the false-twisted yarn is estimated from the physical properties of the raw yarn. That is, the dyeing quality of the false-twisted yarn is estimated without manufacturing the false-twisted yarn from the raw yarn. As a result, the man-hours for confirming the dyeing quality are significantly reduced. Also, the amount of false-twisted yarn stored for quality inspection is reduced, and the cost for storage is reduced. Furthermore, by using a plurality of different physical properties related to the raw yarn as explanatory variables for estimating the dyeing quality, the estimation accuracy of the dyeing quality is further improved.

[0023] The above and other objects, features, aspects and advantages of the present invention will become apparent from the following detailed description of the present invention understood in connection with the accompanying drawings.

Brief Description of the Drawings

[0024] [Figure 1] It is a diagram showing an example of the device configuration of the information processing system. [Figure 2] It is a schematic diagram showing an example of the device configuration of the spinning take-up device. [Figure 3] It is a schematic diagram showing an example of the device configuration of the false-twisting machine. [Figure 4] It is a diagram schematically showing the processing process for estimating the dyeing quality of the false-twisted yarn. [Figure 5] It is a diagram schematically showing the processing process for estimating the dyeing quality of the false-twisted yarn. [Figure 6] It is a diagram showing an example of the learning data. [Figure 7] It is a diagram showing an example of the configuration of the thermal stress measuring device. [Figure 8] It is a diagram showing a flowchart related to the learning process. [Figure 9] It is a diagram showing an example of the correlation table generated during the learning process. [Figure 10] It is a diagram showing another example of the correlation table generated during the learning process. <000011​​​A diagram showing another example of the correlation table generated during the learning process. [Figure 13] A diagram showing another example of the reverse table generated during the learning process. [Figure 14] A diagram showing an example of the output result of the analysis tool. [Figure 15] A diagram showing the output result output during the calculation process of the regression equation. [Figure 16] A diagram showing the correction coefficient of each regression equation. [Figure 17] A diagram showing a flowchart related to the verification process. [Figure 18] A diagram schematically showing the process of the verification process. [Figure 19] A diagram showing an example of the hardware configuration of the information processing apparatus. [Figure 20] A diagram showing an example of the functional configuration of the information processing apparatus.

Embodiments for Carrying Out the Invention

[0025] Hereinafter, each embodiment according to the present invention will be described with reference to the drawings. In the following description, the same parts and components are denoted by the same reference numerals. Their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated. In addition, each embodiment and each modification example described below may be selectively combined as appropriate. [[ID=3​​​​​​​​​​​The spinning and winding device 1, the false twisting machine 30, and the information processing device 100 are configured to be able to exchange data by some means. As an example, the spinning and winding device 1, the false twisting machine 30, and the information processing device 100 are connected to a network, and data is exchanged through the network. In other aspects, data exchange may be performed via a storage medium such as a USB (Universal Serial Bus), or may be performed via an external terminal such as a cloud service or a server.

[0029] The information processing device 100 is, for example, a computer. The information processing device 100 may be integrally configured with the above-described spinning and winding device 1, may be integrally configured with the above-described false twisting machine 30, or may be a server configured to be communicable with the spinning and winding device 1 or the false twisting machine 30.

[0030] The spinning and winding device 1 is a device for manufacturing POY which is the raw yarn of the false twisted yarn. Details of the spinning and winding device 1 will be described later.

[0031] The false twisting machine 30 is a device for manufacturing a false twisted yarn by false twisting the POY which is the raw yarn. Details of the false twisting machine 30 will be described later.

[0032] <B. Spinning and Winding Device 1> Next, referring to FIG. 2, the spinning and winding device 1 shown in FIG. 1 will be described. FIG. 2 is a schematic diagram showing an example of the device configuration of the spinning and winding device 1.

[0033] As shown in FIG. 2, the spinning and winding device 1 respectively takes up yarns Y each composed of a plurality of filaments F spun from the spinning device 2, and winds them around a plurality of bobbins B respectively to form a plurality of packages P. Hereinafter, the vertical direction, the front-rear direction, and the left-right direction shown in FIG. 2 will be defined as the vertical direction of the spinning and winding device 1, the front-rear direction of the spinning and winding device 1, and the left-right direction of the spinning and winding device 1, respectively, for the description.

[0034] Furthermore, in the following, the direction in which multiple filaments F are fed within the spinning take-up device 1 is defined as the downstream side, and the direction opposite to the direction in which multiple filaments F are fed within the spinning take-up device 1 is defined as the upstream side.

[0035] The spinning and taking device 1 includes a cooling unit 3, a lubrication unit 4, taking rollers 6 and 7, a yarn guide 8, and a winding device 9. First, in the spinning device 2, polymer supplied from a polymer supply device (not shown) consisting of a gear pump or the like is pushed downward from a plurality of die heads 2a arranged in the left-right direction, and each group of filaments F is spun out in an arranged left-right direction.

[0036] Subsequently, each group of filament F is sent to the cooling section 3 and the lubrication section 4. In the lubrication section 4, the filaments F are bundled into one thread Y for each group. Then, the multiple threads Y are transported along the thread path, which is aligned in the left-right direction, along the take-up roller 6, thread guide 8, and take-up roller 7. Furthermore, the multiple threads Y are distributed in the front-rear direction from the take-up roller 7 and then wound onto multiple bobbins B in the winding device 9.

[0037] The cooling unit 3 has a plurality of cylindrical cooling tubes 10, each of which is positioned below a plurality of die heads 2a provided on the spinning apparatus 2. The plurality of filaments F spun from the die heads 2a of the spinning apparatus 2 are sent through the internal space 10a of each cooling tube 10 from top to bottom along the axial direction of the cooling tube 10. A flow straightening section 10b is provided around the internal space 10a, and cooling air supplied from a compressed air supply device (not shown) flows into the internal space 10a while being straightened by the flow straightening section 10b. The flow straightening section 10b mainly straightens the flow rate of the cooling air flowing into the internal space 10a so that it is generally uniform in the circumferential direction of the cooling tube 10.

[0038] The lubrication unit 4 has a plurality of lubrication guides 11, each positioned below each cooling cylinder 10. The lubrication guides 11 bundle multiple filaments F spun from the die 2a into a single thread Y, and also apply lubricant to the thread Y (multiple filaments F).

[0039] Multiple threads Y that have passed through the lubrication section 4 are sent to the winding device 9 by the take-up rollers 6 and 7. The thread guide 8 is positioned between the take-up roller 6 and the take-up roller 7 and individually guides the multiple threads Y moving from the take-up roller 6 to the take-up roller 7.

[0040] The winding device 9 comprises a machine base 13, a turret 14, two bobbin holders 15, a support frame 16, a contact roller 17, a traverse device 18, and the like. By rotating the bobbin holders 15, the winding device 9 simultaneously winds multiple threads Y sent from the take-up roller 7 onto multiple bobbins B, forming multiple packages P.

[0041] The turret 14 is a disc-shaped component and is attached to the machine base 13. The turret 14 is rotationally driven by a motor (not shown). Two bobbin holders 15 are cantilevered to the turret 14 in a position extending in the front-rear direction. Multiple cylindrical bobbins B are mounted in each bobbin holder 15 in a line along its axial direction. By rotating the turret 14, the two bobbin holders 15 can be switched between an upper winding position and a lower retracted position.

[0042] The support frame 16 is a long, frame-like member extending in the front-to-back direction. This support frame 16 is fixed to the machine base 13. A long, front-to-back roller support member 19 is attached to the lower part of the support frame 16 so as to be movable up and down relative to the support frame 16. A contact roller 17 extending along the axial direction of the bobbin holder 15 is rotatably supported on the roller support member 19. This contact roller 17 comes into contact with the package P in the process of being formed, and a predetermined contact pressure is applied to the package P, thereby shaping the package P.

[0043] The traverse device 18 has a plurality of traverse guides 18a arranged in the front-rear direction. The plurality of traverse guides 18a are driven by a motor (not shown) and reciprocate in the front-rear direction respectively. As the traverse guide 18a reciprocates with the yarn Y hung thereon, the yarn Y is wound around the corresponding bobbin B while being zigzagged back and forth about the fulcrum guide 18b.

[0044] <C. False Twisting Machine 30> Next, referring to FIG. 3, the false twisting machine 30 shown in FIG. 1 will be described. FIG. 3 is a schematic diagram showing an example of the device configuration of the false twisting machine 30.

[0045] The false twisting machine 30 manufactures a false twisted yarn Y1 using the yarn Y produced by the above-described spinning take-up device 1 as the raw yarn Y0. The false twisting machine 30 manufactures the false twisted yarn Y1 by false twisting the conveyed raw yarn Y0 while heating and stretching it.

[0046] Hereinafter, the direction in which the yarn Y is conveyed in the false twisting machine 30 is defined as the downstream side, and the direction opposite to the direction in which the yarn Y is conveyed in the false twisting machine 30 is defined as the upstream side.

[0047] As shown in FIG. 3, the false twisting machine 30 includes a pair of yarn feed rollers 32, a heater 34, a tension sensor 35, a twister 36, a tension sensor 37, a pair of yarn feed rollers 38, and a winding device 40.

[0048] The pair of yarn feed rollers 32 is configured to draw the raw yarn Y0. The raw yarn Y0 is spanned by the pair of yarn feed rollers 32 and the pair of yarn feed rollers 38. The pair of yarn feed rollers 32 is rotationally driven by a drive source (not shown) such as a motor. When the pair of yarn feed rollers 32 is rotationally driven, the raw yarn Y0 is conveyed from the upstream side to the downstream side.

[0049] The heater 34 is disposed between the pair of yarn feed rollers 32 and the pair of yarn feed rollers 38 and extends along the conveying direction of the yarn Y. The heater 34 is configured to heat the yarn Y conveyed in the false twisting machine 30.

[0050] The tension sensor 35 is disposed between the heater 34 and the twister 36, and measures the tension of the yarn Y before passing through the twister 36. The tension sensor 35 has, for example, a load cell (not shown). The load cell is a sensor for converting the applied force into an electrical signal. The tension sensor 35 measures the tension of the yarn Y before passing through the twister 36 by detecting the force acting on the load cell from the yarn Y.

[0051] The twister 36 is a device for twisting the yarn Y heated by the heater 34. The twister 36 is disposed between the heater 34 and the tension sensor 37.

[0052] The tension sensor 37 is disposed between the twister 36 and the yarn feed roller pair 38, and measures the tension of the yarn Y after passing through the twister 36. As an example, the tension sensor 37 has a load cell (not shown). The load cell is a sensor for converting the applied force into an electrical signal. The tension sensor 37 measures the tension of the yarn Y after passing through the twister 36 by detecting the force acting on the load cell from the yarn Y.

[0053] The yarn feed roller pair 38 is disposed on the downstream side of the tension sensor 35. The yarn feed roller pair 38 rotates by being applied with a driving force from a motor (not shown), and feeds the yarn Y from the upstream side to the downstream side. The rotation speed of the yarn feed roller pair 38 is faster than the rotation speed of the yarn feed roller pair 32. The yarn Y is stretched by the rotation speed ratio between the yarn feed roller pair 32 and the yarn feed roller pair 38.

[0054] The winding device 40 is disposed on the downstream side of the yarn feed roller pair 38, and is a device for winding the false-twist processed yarn Y1 subjected to false-twist processing onto a bobbin (not shown). The false-twist processed yarn Y1 wound on the bobbin is formed as a package.

[0055] <D. Summary> Next, with reference to Figure 4, the function for estimating the dyeing quality of the false-twisted yarn Y1 will be described. Figure 4 is a schematic diagram showing the processing steps for estimating the dyeing quality of the false-twisted yarn Y1. This estimation function is implemented, for example, in the information processing device 100.

[0056] As described above, false-twisted yarn Y1 is produced by false-twisting the raw yarn Y0 produced by the spinning and drawing machine 1. To confirm the dyeing quality of false-twisted yarn Y1, it is usually necessary to go through the false-twisting process of the raw yarn Y0, the trial knitting process of the false-twisted yarn Y1, the dyeing process of the false-twisted yarn Y1, and the quality confirmation process, which is time-consuming.

[0057] Therefore, the information processing device 100 estimates the dye quality of the false-twisted yarn Y1 that can be manufactured from the raw yarn Y0 based on the physical properties of the raw yarn Y0 before false-twisting. This makes it possible to estimate the dye quality of the false-twisted yarn Y1 without actually manufacturing the false-twisted yarn Y1 from the raw yarn Y0.

[0058] More specifically, the control device 101 of the information processing device 100 first acquires an estimated model 124. The estimated model 124 is pre-generated by machine learning the correlation between explanatory variables and the target variable defined for each of the multiple training data sets. The explanatory variables include multiple different types of physical properties measured for the training yarn Y0. The target variable includes the dyeing quality measured for the training false-twist yarn Y1, which is produced by false-twisting the training yarn Y0.

[0059] During estimation, the control device 101 acquires multiple physical properties of the target yarn Y0 and inputs these properties into the estimation model 124. The estimation model 124 then outputs the dyeing quality of the false-twist yarn Y1 produced from the target yarn Y0. The control device 101 outputs the dyeing quality obtained from the estimation model 124 as the estimation result.

[0060] As described above, the dyeing quality of the false-twisted yarn Y1 is estimated without manufacturing the false-twisted yarn Y1 from the raw yarn Y0. As a result, the man-hours required to confirm the dyeing quality are significantly reduced. As a result, the amount of the false-twisted yarn Y1 stored for quality inspection is reduced, and the cost for storage is reduced. Further, by using a plurality of different physical properties related to the raw yarn Y0 as explanatory variables used for estimating the dyeing quality, the estimation accuracy of the dyeing quality is further improved.

[0061] Preferably, the physical properties of the raw yarn Y0 learned as explanatory variables include the thermal stress related to the learning raw yarn Y0 and the boiling water shrinkage rate related to the learning raw yarn Y0. By learning the relationship between these explanatory variables and the objective variable, the estimation accuracy of the dyeing quality of the false-twisted yarn Y1 is improved. The reason for this will be described later.

[0062] <E. Modified Example> Note that the explanatory variables used for estimating the dyeing quality are not limited to the physical properties of the raw yarn Y0. As an example, the explanatory variables used for estimating the dyeing quality may include the physical properties of the false-twisted yarn Y1 manufactured from the raw yarn Y0 in addition to the physical properties of the raw yarn Y0.

[0063] Referring to FIG. 5, the estimation model 124 according to the modified example will be described. FIG. 5 is a diagram schematically showing a processing procedure for estimating the dyeing quality of a false-twisted yarn.

[0064] The estimation model 124 shown in FIG. 5 is generated in advance by machine learning the correlation between the explanatory variables and the objective variable defined for each of a plurality of learning data. In this example, the explanatory variables further include not only a plurality of physical properties measured for the learning raw yarn Y0 but also physical properties measured for the false-twisted yarn Y1 manufactured from the raw yarn Y0.

[0065] At the time of estimation, the control device 101 not only acquires a plurality of physical properties regarding the raw yarn Y0 to be estimated, but also further acquires the physical properties regarding the false-twisted yarn Y1 manufactured from the raw yarn Y0. Then, the control device 101 inputs a plurality of physical properties regarding the raw yarn Y0 to be estimated and the physical properties regarding the false-twisted yarn Y1 to be estimated into the estimation model 124. Thereby, the estimation model 124 outputs the dyeing quality of the false-twisted yarn Y1 to be estimated. The control device 101 outputs the dyeing quality output from the estimation model 124 as an estimation result.

[0066] The information processing device 100 can more accurately estimate the dyeing quality of the false-twisted yarn Y1 by using not only the physical properties of the raw yarn Y0 but also the physical properties of the false-twisted yarn Y1. Also, in this example, although it is necessary to actually manufacture the false-twisted yarn Y1, it is not necessary to dye the false-twisted yarn Y1. Therefore, the man-hours for confirming the dyeing quality are reduced compared to the conventional method.

[0067] The physical properties of the false-twisted yarn Y1 learned as explanatory variables include, for example, the tension of the false-twisted yarn Y1 measured after the false-twisting process is performed. By adding the tension to the explanatory variables, the estimation accuracy of the dyeing quality of the false-twisted yarn Y1 is further improved. The reason for this will be described later.

[0068] <F. Experiment> The inventors confirmed by experiments which explanatory variables are suitable as explanatory variables for estimating the dyeing quality of the false-twisted yarn Y1. Hereinafter, the experimental content conducted by the inventors will be described.

[0069] (F1. Learning data 122) First, referring to FIG. 6, the learning data 122 used in this experiment will be described. FIG. 6 is a diagram showing the learning data 122 prepared by the inventors.

[0070] In the example of FIG. 6, 12 pieces of learning data 122 are shown. Each of the learning data 122 associates a data number NM, an explanatory variable EV, and an objective variable DV.

[0071] The explanatory variable EV includes explanatory variable EV1 relating to the physical properties of the raw yarn Y0, which is POY, and explanatory variable EV2 relating to the physical properties of the false-twisted yarn Y1, which is DTY.

[0072] In the example in Figure 6, the explanatory variable EV1 includes seven explanatory variables EV1A to EV1G. Explanatory variables EV1A to EV1G were measured using yarn Y0 sampled from package P (see Figure 2).

[0073] The explanatory variable EV1A, "POY strength," is an indicator of the durability of the yarn Y0. "POY strength" is calculated by dividing the tension at the time of breakage when the yarn Y0 is stretched by the thickness of the yarn Y0. The unit of this tension is expressed in "cN." The unit of this thickness is expressed in "dtex." The unit of "POY strength" is expressed as "cN / dtex." "POY strength" can be measured, for example, using a strength measuring device.

[0074] The explanatory variable EV1B, "POY elongation," is an indicator of the flexibility of the yarn Y0. "POY elongation" is calculated by dividing the length of the yarn Y0 at the time of breakage when the yarn Y0 is stretched by the length of the yarn Y0 before stretching. The unit of "POY elongation" is expressed as "%". "POY elongation" can be measured, for example, using an elongation measuring device.

[0075] The explanatory variable EV1C, "DR (Draw Ratio) thermal stress," is the stress applied to the POY by simulating the false twisting process in the false twisting machine 30 described above. "DR thermal stress" is measured, for example, using the thermal stress measuring device 400 shown in Figure 7.

[0076] Figure 7 shows an example of the configuration of the thermal stress measuring device 400. The thermal stress measuring device 400 comprises a thread feed roller pair 432, a tension sensor 434, a heater 436, and a thread feed roller pair 438.

[0077] In the following, the direction in which the yarn Y0 is transported within the thermal stress measuring device 400 is defined as the downstream side, and the direction opposite to the direction in which the yarn Y0 is transported within the thermal stress measuring device 400 is defined as the upstream side.

[0078] The yarn feed roller pair 432 is configured to take in the raw yarn Y0. The raw yarn Y0 is passed through the yarn feed roller pair 432 and the yarn feed roller pair 438. The yarn feed roller pair 432 is rotationally driven by a drive source (not shown), such as a motor. As the yarn feed roller pair 432 is rotationally driven, the raw yarn Y0 is conveyed from the upstream side to the downstream side.

[0079] The tension sensor 434 is positioned between the yarn feed roller pair 432 and the heater 436 and measures the tension of the yarn Y0 before it passes through the heater 436. The tension sensor 434 has, for example, a load cell (not shown). A load cell is a sensor that converts the applied force into an electrical signal. The tension sensor 434 measures the tension of the yarn Y0 before it passes through the heater 436 by detecting the force applied from the yarn Y0 to the load cell. The tension of the yarn Y0 measured by the tension sensor 434 is stored as "DR thermal stress" in the training data 122. The unit of "DR thermal stress" is "cN".

[0080] The heater 436 is positioned between the tension sensor 434 and the yarn feed roller pair 438 and extends along the direction of conveyance of the yarn Y0. The heater 436 is configured to heat the yarn Y0 as it is conveyed within the thermal stress measuring device 400. The heating temperature of the heater 436 is set to be the same as, for example, the heating temperature of the heater 34 (see Figure 3) described above.

[0081] The yarn feed roller pair 438 is located downstream of the heater 436. The yarn feed roller pair 438 rotates under the driving force from a motor (not shown), feeding the yarn Y0 from upstream to downstream. The rotational speed of the yarn feed roller pair 438 is faster than that of the yarn feed roller pair 432. The yarn Y0 is stretched by the rotational speed ratio of the yarn feed roller pair 432 and the yarn feed roller pair 438. This rotational speed ratio is called the draw ratio (DR). The draw ratio is set, for example, to be the same as the rotational speed ratio of the yarn feed roller pairs 32 and 38 (see Figure 3) described above.

[0082] Referring again to Figure 6, the explanatory variable EV1D, "DR strength," represents the tension of the yarn Y0 when the yarn Y0 is statically stretched at the draw ratio of the false twisting machine 30 described above. The unit of "DR strength" is "cN".

[0083] The explanatory variable EV1E, "BWS (Boiled Water Shrinkage)," represents the boiling water shrinkage rate of the yarn Y0. More specifically, "BWS" is calculated by dividing the length by which the yarn Y0 shrinks when immersed in boiling water by the length of the yarn Y0 before immersion. The unit of "BWS" is expressed as "%". "BWS" may be measured using a measuring instrument or by manual measurement.

[0084] The explanatory variable EV1F, represented as "U%", is an index indicating the variation in the thickness of the raw yarn Y0 measured at various points along the longitudinal direction. The measurement interval for this thickness is shorter than the measurement interval for "Uhi%". The unit of "U%" is "%". "U%" is measured, for example, using a yarn unevenness measuring machine.

[0085] The explanatory variable EV1G, "Uhi%", is an index indicating the variation in the thickness of the raw yarn Y0 measured at various points along the longitudinal direction. The measurement interval for this thickness is longer than the measurement interval for "U%". The unit of "Uhi%" is "%". "Uhi%" is measured, for example, using a yarn unevenness measuring machine.

[0086] The explanatory variable EV2 represents the explanatory variables related to the physical properties of the false-twisted yarn Y1, which is DTY. In the example in Figure 6, the explanatory variable EV2 includes four explanatory variables EV2A to EV2D. Explanatory variables EV2A to EV2D were measured using the false-twisted yarn Y1 obtained by false-twisting the raw yarn Y0 in the false-twisting machine 30.

[0087] The explanatory variable EV2A, "DTY strength," is an indicator of the durability of the false-twisted yarn Y1 after processing. "DTY strength" is calculated by dividing the tension at the time of breakage when the false-twisted yarn Y1 is stretched by the thickness of the false-twisted yarn Y1. The unit of this tension is expressed in "cN." The unit of this thickness is expressed in "dtex." The unit of "DTY strength" is expressed in "cN / dtex." "DTY strength" is measured, for example, using a strength measuring device.

[0088] The explanatory variable EV2B, "DTY elongation," is an index indicating the flexibility of the false-twisted yarn Y1 after processing. "DTY elongation" is calculated by dividing the length of the false-twisted yarn Y1 at the time of breakage when stretched by the length of the false-twisted yarn Y1 before stretching. The unit of "DTY elongation" is expressed as "%". "DTY elongation" can be measured, for example, using an elongation measuring device.

[0089] The explanatory variable EV2C, "T1," represents the tension of the false-twisted yarn Y1 measured during false-twisting by the false-twisting machine 30. The unit of "T1" is "cN." "T1" is obtained, for example, from the tension sensor 35 (see Figure 3) described above. As described above, the tension sensor 35 is located upstream of the twister 36. That is, "T1" is the tension measured before the twisting process by the twister 36.

[0090] The explanatory variable EV2D, "T2," represents the tension of the false-twisted yarn Y1 measured during false-twisting by the false-twisting machine 30. The unit of "T2" is "cN." "T2" is obtained, for example, from the tension sensor 37 (see Figure 3) described above. As described above, the tension sensor 37 is located downstream of the twister 36. That is, "T2" is the tension measured after the twisting process by the twister 36.

[0091] The "Dyeing L* value," shown as the dependent variable DV, is an index indicating the dyeing quality of the false-twist yarn Y1. The "Dyeing L* value" indicates the intensity of the color when the false-twist yarn Y1 is actually dyed. A higher "Dyeing L* value" indicates that the false-twist yarn Y1 is lighter in color, while a lower "Dyeing L* value" indicates that the false-twist yarn Y1 is darker in color. The "Dyeing L* value" is measured, for example, by a measuring device such as a spectrophotometer.

[0092] (F2. Learning process) Next, referring to Figures 8 to 16, we will explain the learning process using the aforementioned training data 122 (see Figure 6). Figure 8 is a flowchart showing the learning process.

[0093] The control unit 101 of the information processing device 100 executes the various processes shown in Figure 8 by running the learning program 126 (see Figure 19), which will be described later. In other aspects, some or all of the processes shown in Figure 8 may be executed by circuit elements or other hardware.

[0094] In step S110, the control device 101 acquires the aforementioned training data 122 (see Figure 6). The training data 122 may be acquired from a storage device within the information processing device 100, or from a device different from the information processing device 100 (for example, a server).

[0095] In step S112, the control device 101 calculates the correlation coefficient between each of the explanatory variables EV defined in the training data 122 and the target variable DV. Furthermore, the control device 101 calculates the correlation coefficients between the explanatory variables EV defined in the training data 122. Figure 9 shows the correlation table LS1 which represents the correlation coefficients calculated in step S112.

[0096] In step S114, the control device 101 selects an explanatory variable EV that is correlated with the target variable DV from among the candidate explanatory variables specified in the training data 122. The explanatory variable EV is selected based on the correlation table LS1.

[0097] As an example, the control device 101 excludes explanatory variables EV whose absolute value of the correlation coefficient with the objective variable DV is less than a predetermined threshold. This threshold is, for example, 0.3. In this case, the control device 101 excludes explanatory variables EV1A, EV1B, EV1F, EV1G, EV2A, and EV2B.

[0098] Furthermore, if the absolute value of the correlation coefficient between two explanatory variables EV is greater than or equal to a predetermined threshold, the control device 101 excludes the explanatory variable EV whose absolute value of the correlation coefficient with the dependent variable DV is lower. This threshold is, for example, 0.9. As a result, one of the explanatory variable EVs that exhibit multicollinearity is excluded. In the example correlation table LS1 shown in Figure 9, there are no combinations of explanatory variable EVs whose absolute value of the correlation coefficient is greater than or equal to 0.9.

[0099] In step S116, the control device 101 calculates the VIF (Variance Inflation Factor) based on the correlation table LS1. The VIF is an indicator that shows how well the explanatory variable EV is correlated with other explanatory variables EV, and it indicates the degree of multicollinearity.

[0100] More specifically, first, the control device 101 refers to the correlation table LS1 described above and extracts a correlation table for the explanatory variables EV that remained after the selection in step S114. Figure 10 shows the correlation table LS1A after extraction.

[0101] Next, the control device 101 calculates the inverse matrix for the correlation coefficients between the explanatory variables EV specified in the correlation table LS1A. Figure 11 shows the inverse matrix table LS2A, which represents the result of this calculation. The diagonal elements of the inverse matrix table LS2A, from the top left to the bottom right, represent the respective VIFs of the explanatory variables EV1C, EV1D, EV1E, EV2C, and EV2D.

[0102] In step S120, the control device 101 determines whether all of the VIFs shown in the inverse matrix table LS2A are less than a predetermined threshold. The predetermined threshold is, for example, 10. If all of the VIFs are less than 10, the control device 101 determines that there is no multicollinearity among the explanatory variables EV1C, EV1D, EV1E, EV2C, and EV2D.

[0103] If the control device 101 determines that all VIFs shown in the inverse matrix table LS2A are less than 10 (YES in step S120), it switches the control to step S132. Otherwise (NO in step S120), the control device 101 switches the control to step S122.

[0104] In the example shown in Figure 11, since the VIF related to the explanatory variables EV1C and EV2C is 10 or greater, the control is switched from step S120 to step S122.

[0105] In step S122, the control device 101 further removes explanatory variables EV that are causing multicollinearity from the remaining explanatory variables EV. At this time, the control device 101 identifies other explanatory variables EV that have a strong correlation with the explanatory variable EV with the highest VIF as candidates for removal. Then, among these explanatory variables EV, the control device 101 removes explanatory variables EV with a lower absolute value of the correlation coefficient with the dependent variable DV.

[0106] In the example in Figure 11, the VIF of explanatory variable EV1C (i.e., "DR thermal stress") is the highest. In this case, the control device 101 refers to the correlation table LS1A and identifies explanatory variables EV1D and EV2C (i.e., "DR strong" and "T1") whose absolute value of the correlation coefficient with explanatory variable EV1C exceeds a predetermined value (for example, 0.8) as candidates for exclusion. Next, the control device 101 keeps explanatory variable EV1C, EV1D, and EV2C, which has the highest absolute value of the correlation coefficient with the objective variable DV, and excludes the other explanatory variables EV1D and EV2C.

[0107] Subsequently, the control device 101 returns control to step S116. Next, the control device 101 extracts a correlation table for the explanatory variables EV that remained after the selection in step S122. Figure 12 shows the extracted correlation table LS1B. In the example in Figure 12, the explanatory variables EV1C, EV1E, and EV2D remain.

[0108] The control device 101 recalculates the VIFs based on the correlation table LS1B. More specifically, the control device 101 calculates the inverse matrix for the correlation coefficients between the explanatory variables EV specified in the correlation table LS1B. Figure 13 shows the inverse matrix table LS2B, which represents the calculation result of the inverse matrix. The diagonal elements from the top left to the bottom right of the inverse matrix table LS2B represent the VIFs of the explanatory variables EV1C, EV1E, and EV2D. In the example in Figure 13, all VIFs are less than 10, so the judgment result in step S120 is "YES", and the control is switched to step S132.

[0109] In step S132, the control device 101 learns the correlation between the explanatory variables EV selected in steps S114 and S122 and the dependent variable DV by multiple regression analysis. In the example in Figure 13, since explanatory variables EV1C, EV1E, and EV2D remain, the control device 101 performs multiple regression analysis using the explanatory variables EV1C, EV1E, and EV2D and the dependent variable DV. As a result, the relationship between the explanatory variables EV1C, EV1E, and EV2D and the dependent variable DV is defined by the regression equation. When there are three explanatory variables EV, the regression equation is shown by equation (1) below.

[0110] y=a1 x1+a2 x2+a3 x3+b (1) In equation (1), "y" represents the dependent variable DV. "x1" represents the independent variable EV1C. "a1" represents the regression coefficient multiplied by the independent variable EV1C. "x2" represents the independent variable EV1E. "a2" represents the regression coefficient multiplied by the independent variable EV1E. "x3" represents the independent variable EV2D. "a3" represents the regression coefficient multiplied by the independent variable EV2D. "b" represents the intercept of the regression equation.

[0111] In step S132, the multiple regression analysis determines the values ​​of the regression coefficients "a1" to "a3" and the intercept "b". For example, the analysis tool in Microsoft Excel® is used to calculate the regression equation. Figure 14 shows the output RS1 of the analysis tool.

[0112] The output RS1 includes not only the "regression coefficient" and "intercept," but also the "standard error," "t-value," "p-value," "lower limit (95%)," and "upper limit (95%)."

[0113] Standard error is an indicator used to assess the reliability of estimated regression coefficients. It shows how much variation there is in the influence each explanatory variable has on the dependent variable.

[0114] The "t-value" is an indicator that shows whether the explanatory variable EV has a significant influence on the dependent variable DV. In other words, the "t-value" shows how large the estimated regression coefficient is relative to its standard error. That is, the "t-value" is an indicator that shows how far the regression coefficient is from zero. The larger the absolute value of the "t-value," the more statistically significant the influence of the explanatory variable EV on the dependent variable DV is. On the other hand, the smaller the absolute value of the "t-value," the less statistically significant the influence of the explanatory variable EV on the dependent variable DV is.

[0115] The "P-value" is an indicator that shows whether the explanatory variable EV has a significant influence on the dependent variable DV. The "P-value" represents the probability of obtaining the observed t-value based on the null hypothesis that the regression coefficient is zero. A smaller "P-value" indicates that the influence of the explanatory variable EV on the dependent variable DV is statistically significant. Conversely, a larger "P-value" indicates that the influence of the explanatory variable EV on the dependent variable DV is not statistically significant.

[0116] The "lower limit of 95%" indicates the lower end of the confidence interval for the regression coefficient. The "upper limit of 95%" indicates the upper end of the confidence interval for the regression coefficient. An explanatory variable EV whose "t-value" does not fall within the confidence interval is considered unreliable.

[0117] In step S136, the control device 101 refers to the output result RS1 and excludes the explanatory variable EV whose absolute value of the "t-value" is the smallest. In the example in Figure 14, the explanatory variable EV2D (i.e., "T2") whose absolute value of the "t-value" is the smallest is excluded.

[0118] In step S138, the control device 101 refers to the training data 122 and performs a multiple regression analysis again using the remaining explanatory variable EV and the dependent variable DV. As a result, when step S138 is performed for the first time, a regression equation is calculated that defines the relationship between the explanatory variables EV1C and EV1E and the dependent variable DV.

[0119] In step S140, the control device 101 determines whether there is one remaining explanatory variable EV. If the control device 101 determines that there is one remaining explanatory variable EV (YES in step S140), it switches control to step S142. Otherwise (NO in step S140), the control device 101 returns control to step S136.

[0120] As steps S136, S138, and S140 are repeated, the control device 101 calculates the regression equation while removing explanatory variables EV with small absolute values ​​one by one. Figure 15 shows the output results RS1 to RS3 that are output during the calculation of the regression equation.

[0121] In step S142, the control device 101 selects the regression equation that defines the optimal combination of explanatory variables EV from among the regression equations calculated in steps S132 and S138 as the estimation model 124. As an example, the control device 101 determines the regression equation to be adopted as the estimation model 124 based on the correction coefficient R2.

[0122] The correction coefficient R² is an index used to evaluate the estimation accuracy of the regression equation. A larger value of the correction coefficient R² indicates better estimation accuracy. Conversely, a smaller value of the correction coefficient R² indicates poorer estimation accuracy. Therefore, the control device 101 adopts the regression equation with the maximum correction coefficient R² as the estimation model 124.

[0123] Figure 16 shows the correction coefficient R2 for each regression equation. In the example in Figure 16, the correction coefficient R2 for the regression equation consisting of the three explanatory variables EV1C, EV1E, and EV2D is maximized, so this regression equation is adopted as estimation model 124. The correction coefficient "R2" is calculated based on the following equation (2).

[0124] R² = 1 - (1 - R 2 )·(n-1) / (nk-1)···(2) The "R" shown in equation (2) 2 " indicates the coefficient of determination. "n" indicates the number of training data 122. In the example of training data 122 shown in Figure 6, "n" is "12". "k" indicates the number of explanatory variables EV. "R 2 This is represented by equation (3) below.

[0125] R 2 =1-RSS / TSS···(3) The "RSS" shown in equation (3) is obtained by calculating the sum of squares of the differences between the estimated values ​​from the regression equation and the actual values ​​(correct values). The "TSS" is obtained by calculating the sum of squares of the differences between each estimated value from the regression equation and the mean value of the estimated values.

[0126] (F3. Verification process) Next, with reference to Figures 17 and 18, the validation process of the estimated model 124 generated by the learning process shown in Figure 8 will be described. Figure 17 is a flowchart showing the validation process. Figure 18 is a diagram illustrating the process of the validation process.

[0127] The control unit 101 of the information processing device 100 executes the various processes shown in Figure 17 by executing the estimation program 128 (see Figure 19), which will be described later. In other aspects, some or all of the processes shown in Figure 17 may be executed by circuit elements or other hardware.

[0128] In step S210, the control device 101 acquires validation data 123 to verify the estimation accuracy of the estimated model 124. The validation data 123 is different from the training data 122 (see Figure 6) used during training. In the example in Figure 18, 12 validation data 123 are shown. Each of the validation data 123 associates a data number NM with an explanatory variable EV and an actual value VA related to the "stained L* value".

[0129] Furthermore, the explanatory variables EV shown in the verification data 123 are adjusted to match the types of explanatory variables selected by the learning process shown in Figure 8 above. In the example in Figure 18, the explanatory variables EV are shown as explanatory variable EV1C related to "DR thermal stress", explanatory variable EV1E related to "BWS", and explanatory variable EV2D related to "T2".

[0130] In step S212, the control device 101 inputs the value of the explanatory variable EV specified in the verification data 123 into the estimation model 124. As a result, the estimation model 124 outputs an estimated value VB related to the "staining L* value".

[0131] In step S214, the control device 101 compares the measured value VA specified in the verification data 123 with the estimated value VB as the correct value and calculates the estimation accuracy of the estimation model 124 as the evaluation result RS4.

[0132] The "correlation coefficient" shown in the evaluation result RS4 is calculated based on the following formula (4) using an Excel function, for example. The "target range R1" shown in formula (4) is specified as the measured value VA. The "target range R2" shown in formula (4) is specified as the estimated value VB. The "degrees of freedom" shown in the evaluation result RS4 is calculated by "number of data points in the validation data 123 - 2". The "t-value" shown in the evaluation result RS4 is calculated based on the following formula (5) using an Excel function. The "boundary value" shown in the evaluation result RS4 is calculated based on the following formula (6) using an Excel function. The "p-value" shown in the evaluation result RS4 is calculated based on the following formula (7) using an Excel function.

[0133] Correlation coefficient = CORREL(range R1, range R2) ... (4) t-value = (correlation coefficient * SQRT(degrees of freedom)) / (SQRT(1 - correlation coefficient^2)) ... (5) Boundary value = T.INV.2T(5%, degrees of freedom)···(6) P-value = T.DIST.2T(ABS(t-value), degrees of freedom) ... (7) The control device 101 determines the superiority of the generated estimated model 124 based on the "P-value" shown in the evaluation result RS4, for example. Here, the smaller the "P-value", the higher the superiority of the estimated model 124, and the larger the "P-value", the lower the superiority of the estimated model 124. As an example, in step S220, the control device 101 determines whether the P-value is less than 0.05. If the control device 101 determines that the P-value is less than 0.05 (YES in step S220), it switches the control to step S222. Otherwise (NO in step S220), the control device 101 switches the control to step S232.

[0134] In step S222, the control device 101 determines that the explanatory variable EV selected by the learning process is superior and adopts the generated estimation model 124.

[0135] In step S232, the control device 101 determines that the explanatory variable EV selected by the learning process is not superior and rejects the generated estimation model 124. In this case, the control device 101 changes the learning conditions, such as using different training data 122, and executes the learning process shown in Figure 8 again.

[0136] (F4. Summary) Based on the above, it was confirmed that the combination of explanatory variable EV1C related to the thermal stress of the raw yarn Y0, explanatory variable EV1E related to the boiling water shrinkage rate of the raw yarn Y0, and explanatory variable EV2D related to the tension of the false-twisted yarn Y1 measured during the false-twisting process is particularly advantageous for estimating the dyeing quality of the false-twisted yarn Y1.

[0137] Furthermore, the combination of explanatory variables EV to be learned is not limited to explanatory variables EV1C, EV1E, and EV2D, and explanatory variable EV2D relating to the physical properties of the false-twisted yarn Y1 does not necessarily have to be adopted. As an example, any combination selected from the explanatory variables EV1A to EV1G mentioned above can be adopted as the learning target. Preferably, the combination of explanatory variable EV1C relating to the thermal stress of the raw yarn Y0 and explanatory variable EV1E relating to the boiling water shrinkage rate of the raw yarn Y0 is adopted as the learning target.

[0138] As another example, explanatory variable EV2 relating to the physical properties of the false-twisted yarn Y1 may be added to the explanatory variable EV being studied. In this case, at least one of the explanatory variables EV2A to EV2D mentioned above will be added to the study. As an example, explanatory variable EV2D relating to the tension of the false-twisted yarn Y1 measured during the false-twisting process will be added to the study.

[0139] Also, in the above description, an example using multiple regression analysis as the learning algorithm for the learning data 122 was explained, but the learning algorithm to be adopted is not limited to multiple regression analysis. As an example, learning algorithms such as deep learning and support vector machines may be adopted.

[0140] Preferably, the information processing apparatus 100 makes the manufacturing conditions when acquiring the explanatory variables for learning the same as the manufacturing conditions when acquiring the explanatory variables for estimation. Examples of the manufacturing conditions to be aligned include, for example, the spinning speed.

[0141] <G. Hardware Configuration of Information Processing Apparatus 100> Next, referring to FIG. 19, the hardware configuration of the information processing apparatus 100 will be described. FIG. 19 is a diagram showing an example of the hardware configuration of the information processing apparatus 100.

[0142] The information processing apparatus 100 includes the above-described control apparatus 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a communication interface 104, a display interface 105, an input interface 107, and an auxiliary storage device 120. These components are connected to a bus 115.

[0143] The control apparatus 101 is constituted by, for example, at least one integrated circuit. The integrated circuit may be constituted by, for example, at least one CPU (Central Processing Unit), at least one GPU (Graphics Processing Unit), at least one ASIC (Application Specific Integrated Circuit), at least one FPGA (Field Programmable Gate Array), or a combination thereof.

[0144] The control device 101 controls the operation of the information processing device 100 by executing various programs such as the learning program 126 and the estimation program 128. Based on the receipt of execution instructions for various programs, the control device 101 reads the program to be executed from the auxiliary storage device 120 or ROM 102 into the RAM 103. The RAM 103 functions as working memory and temporarily stores various data necessary for program execution.

[0145] The communication interface 104 is an interface for the information processing device 100 to communicate with external devices. The information processing device 100 exchanges data with external devices via the communication interface 104. Examples of such external devices include the spinning and drawing device 1 described above, the false twisting machine 30 described above, measuring instruments such as the thermal stress measuring device 400 described above, and servers.

[0146] A display 106 is connected to the display interface 105. The display interface 105 sends image signals to the display 106 for displaying images, according to commands from the control device 101 or the like. The display 106 is, for example, a liquid crystal display, an organic EL (Electro-Luminescence) display, or another type of display. The display 106 may be configured integrally with the information processing device 100, or it may be configured separately from the information processing device 100.

[0147] An input device 108 is connected to the input interface 107. The input device 108 is, for example, a mouse, keyboard, touch panel, or other device capable of receiving user input. The input device 108 may be configured integrally with the information processing device 100, or it may be configured separately from the information processing device 100.

[0148] The auxiliary storage device 120 is, for example, a storage medium such as a hard disk, flash memory, and SSD (Solid State Drive). The auxiliary storage device 120 stores, for example, the learning data 122, the verification data 123, the estimation model 124, the learning program 126, and the estimation program 128. The storage location of these is not limited to the auxiliary storage device 120, but may also be the storage area of ​​the control device 101 (for example, cache memory), ROM 102, RAM 103, external devices (for example, a server), etc.

[0149] The learning program 126 is a program for generating an estimated model 124 using training data 122. The learning program 126 may be provided not as a standalone program, but incorporated as part of any other program. In this case, the learning process by the learning program 126 is realized in cooperation with the other program. Even if a program does not include such a partial module, it does not deviate from the intent of the learning program 126 according to this embodiment. Furthermore, some or all of the functions provided by the learning program 126 may be realized by dedicated hardware. In addition, the information processing device 100 may be configured in a form such as a so-called cloud service in which at least one server executes part of the processing of the learning program 126.

[0150] The estimation program 128 is a program for estimating the dyeing quality of the false-twist yarn Y1 manufactured by false-twisting the raw yarn Y0, using the physical properties related to the raw yarn Y0 and the estimation model 124. The estimation program 128 may be provided not as a single program but incorporated into a part of an arbitrary program. In this case, the estimation process by the estimation program 128 is realized in cooperation with an arbitrary program. Even a program that does not include such a part of the module does not deviate from the gist of the estimation program 128 according to the present embodiment. Furthermore, part or all of the functions provided by the estimation program 128 may be realized by dedicated hardware. Furthermore, the information processing apparatus 100 may be configured in the form of a so-called cloud service in which at least one server executes part of the processing of the estimation program 128.

[0151] <H. Functional Configuration of Information Processing Apparatus 100> Next, referring to FIG. 20, the functional configuration of the information processing apparatus 100 will be described. FIG. 20 is a diagram showing an example of the functional configuration of the information processing apparatus 100.

[0152] As shown in FIG. 20, the information processing apparatus 100 includes a learning unit 150 and an estimation unit 152 as functional configurations.

[0153] The learning unit 150 performs machine learning on the correlation between the explanatory variable EV and the target variable DV defined in each of the learning data 122, and generates the estimation model 124. Since the learning process by the learning unit 150 has been described as in FIG. 8 and the like, the description thereof will not be repeated.

[0154] The estimation unit 152 at least acquires a plurality of physical properties regarding the raw yarn Y0 to be estimated, and inputs the plurality of physical properties into the estimation model 124. Thereby, the estimation unit 152 acquires the dyeing quality of the false-twist yarn Y1 from the estimation model 124, and outputs the dyeing quality as an estimation result.

[0155] The physical properties of the yarn Y0 obtained by the estimation unit 152 include, for example, the thermal stress of the yarn Y0 and the boiling water shrinkage rate of the yarn Y0. The thermal stress of the yarn Y0 may be obtained, for example, from the thermal stress measuring device 400 described above, or it may be input by the user. The boiling water shrinkage rate of the yarn Y0 may be obtained, for example, from a thermal shrinkage measuring device, or it may be input by the user.

[0156] Preferably, the estimation unit 152 acquires not only the physical properties of the raw yarn Y0 to be estimated, but also the physical properties of the false-twisted yarn Y1 manufactured from the raw yarn Y0. In this case, the estimation model 124 is trained to accept input of both the physical properties of the raw yarn Y0 and the physical properties of the false-twisted yarn Y1.

[0157] In a certain phase, the physical properties of the false-twisted yarn Y1 acquired by the estimation unit 152 include the tension measured for the false-twisted yarn Y1 during transport after the twisting process. This tension is obtained, for example, from the tension sensor 37 described above.

[0158] In other aspects, the physical properties of the false-twisted yarn Y1 acquired by the estimation unit 152 include the tension measured on the false-twisted yarn Y1 while it is being transported, prior to the twisting process. This tension is obtained, for example, from the tension sensor 35 described above.

[0159] The estimation unit 152 inputs multiple physical properties obtained for the raw yarn Y0 to be estimated, along with the physical properties of the false-twisted yarn Y1 manufactured from the raw yarn Y0, into the estimation model 124. As a result, the estimation unit 152 obtains the dyeing quality of the false-twisted yarn Y1 from the estimation model 124 and outputs the dyeing quality as the estimation result.

[0160] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of the present invention is indicated by the claims rather than by the foregoing description, and all modifications within the meaning and scope equivalent to the claims are intended to be included. [Explanation of Symbols]

[0161] 100 Information Processing Devices 101 Control device 122 Training Data 124 Estimated Models 128 Estimation Program DV (Dependent Variable) EV explanatory variables Y0 raw yarn Y1 False Twist Yarn

Claims

1. An information processing device capable of estimating the dyeing quality of false-twisted yarn, which is produced by false-twisting raw yarn, Equipped with a control device, The control device executes a process to obtain an estimation model generated by machine learning the correlation between the explanatory variables and the target variable defined in each of the multiple training data sets. The explanatory variables include several physical properties measured with respect to the yarn used for learning, The aforementioned objective variable includes the dye quality measured with respect to the false-twisted yarn for learning, which is produced by false-twisting the learning yarn. A process for obtaining multiple physical properties of the target yarn, An information processing device that performs the process of inputting the plurality of physical properties obtained with respect to the yarn to be estimated into the estimation model and outputting the dyeing quality obtained from the estimation model.

2. The aforementioned multiple physical properties are, The thermal stress relating to the aforementioned learning yarn, This includes the boiling water shrinkage rate of the yarn used for learning, The information processing apparatus according to claim 1, wherein the plurality of physical properties obtained with respect to the yarn subject to estimation include physical properties of the same type as the explanatory variables.

3. The information processing apparatus according to claim 1 or 2, wherein the explanatory variables further include physical properties measured for the learning false-twist yarn during the false-twist process.

4. The aforementioned false twisting process is The process of heating the raw yarn being transported, The process of stretching the raw yarn being transported, This includes a process of twisting the raw yarn being transported. The information processing apparatus according to claim 3, wherein the physical properties of the false-twisted yarn for learning include the tension measured for the false-twisted yarn during transport after the twisting process.

5. The information processing apparatus according to claim 1 or 2, wherein the machine learning includes a process of selecting a physical property that correlates with the objective variable from among a plurality of candidate physical properties relating to the yarn for learning as the explanatory variable.

6. The information processing apparatus according to claim 5, further comprising a process of learning the correlation between the explanatory variables selected by the selection process and the target variable by multiple regression analysis.

7. A method for estimating the dyeing quality of false-twisted yarn, which is produced by false-twisting raw yarn, The process includes a step to obtain an estimation model generated by machine learning the correlation between the explanatory variables and the target variable defined in each of the multiple training datasets. The explanatory variables include several physical properties measured with respect to the yarn used for learning, The aforementioned objective variable includes the dye quality measured with respect to the false-twisted yarn for learning, which is produced by false-twisting the learning yarn. The aforementioned estimation method further, A step of obtaining multiple physical properties of the yarn to be estimated, An estimation method comprising the step of inputting the plurality of physical properties obtained with respect to the yarn to be estimated into the estimation model and outputting the dyeing quality obtained from the estimation model.

8. An estimation program for estimating the dyeing quality of false-twisted yarn, which is produced by false-twisting raw yarn, The estimation program causes the information processing device to execute a process to obtain an estimation model that is generated by machine learning the correlation between the explanatory variables and the target variable defined in each of the multiple training data sets. The explanatory variables include several physical properties measured with respect to the yarn used for learning, The aforementioned objective variable includes the dye quality measured with respect to the false-twisted yarn for learning, which is produced by false-twisting the learning yarn. The estimation program further provides the information processing device with: A process for obtaining multiple physical properties of the target yarn, An estimation program that performs the process of inputting the multiple physical properties obtained with respect to the yarn to be estimated into the estimation model and outputting the dyeing quality obtained from the estimation model.

Images