JP2019201741A - Seam inspection device - Google Patents

Abstract

To provide a seam inspection device capable of efficiently inspecting seams.SOLUTION: A seam inspection device comprises a tension sensor for detecting the tension of a needle thread hooked on a sewing needle, and a processing unit for detecting the disorder in a seam formed on a sewing object by the sewing needle based on the detection value from the tension sensor.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a seam inspection device.

  In a garment factory, clothes are produced using sewing machines. An example of a seam inspection device for inspecting a seam formed by a sewing machine is disclosed in Patent Document 1.

JP-A-11-090077

  If the seam inspection is complicated, the productivity of the garment may be reduced. In order to suppress a decrease in the productivity of clothing, there is a demand for a technology that can efficiently perform the seam inspection.

  An object of an aspect of the present invention is to efficiently perform a seam inspection.

  According to the aspect of the present invention, the tension sensor that detects the tension of the upper thread that is applied to the sewing needle, and the abnormality of the seam formed on the sewing object by the sewing needle is detected based on the detection value of the tension sensor. A seam inspection device comprising: a processing device;

  According to the aspect of the present invention, the seam inspection can be performed efficiently.

FIG. 1 is a perspective view schematically showing a sewing machine according to an embodiment. FIG. 2 is a diagram illustrating a motion diagram of the needle bar and the balance according to the embodiment. FIG. 3 is a functional block diagram illustrating the seam inspection device according to the embodiment. FIG. 4 is a diagram schematically showing normal seams. FIG. 5 is a diagram schematically showing an abnormal seam. FIG. 6 is a diagram schematically showing an abnormal seam. FIG. 7 is a diagram illustrating detection values of the tension sensor. FIG. 8 is a diagram illustrating detection values of the tension sensor. FIG. 9 is a diagram showing a cumulative curve. FIG. 10 is a diagram showing a frequency spectrum of tension. FIG. 11 is a flowchart illustrating a sewing method according to the embodiment. FIG. 12 is a flowchart showing a sewing method according to the embodiment.

  Hereinafter, embodiments according to the present invention will be described with reference to the drawings, but the present invention is not limited thereto. The components of the embodiments described below can be combined as appropriate. Some components may not be used.

[First Embodiment]
<Sewing machine>
The sewing machine 1 according to the present embodiment will be described. In the present embodiment, the positional relationship between the respective units will be described based on the local coordinate system defined in the sewing machine 1. The local coordinate system is defined by an XYZ orthogonal coordinate system. A direction parallel to the X axis in the predetermined plane is defined as an X axis direction. A direction parallel to the Y axis in a predetermined plane orthogonal to the X axis is taken as a Y axis direction. A direction parallel to the Z axis perpendicular to the predetermined plane is taken as a Z axis direction. A rotation direction around the X axis is defined as a θX direction.

  FIG. 1 is a perspective view schematically showing a sewing machine 1 according to this embodiment. As shown in FIG. 1, the sewing machine 1 includes a sewing head 2, a needle bar 4, a balance 5, a thread tensioner 6, a needle plate 7, a presser member 8, a hook 9, a motor 10, and a seam. And an inspection device 20.

  The needle bar 4 holds the sewing needle 3 and reciprocates in the Z-axis direction. The needle bar 4 holds the sewing needle 3 so that the sewing needle 3 and the Z axis are parallel to each other. The needle bar 4 is supported by the sewing machine head 2. The needle bar 4 is disposed above the needle plate 7 and can face the surface of the sewing object S. The upper thread UT is hung on the sewing needle 3. The sewing needle 3 has a threading hole through which the upper thread UT passes. The sewing needle 3 holds the upper thread UT on the inner surface of the threading hole. As the needle bar 4 reciprocates in the Z-axis direction, the sewing needle 3 reciprocates in the Z-axis direction while holding the upper thread UT.

  The balance 5 supplies the upper thread UT to the sewing needle 3. The balance 5 is supported by the sewing machine head 2. The balance 5 has a balance hole through which the upper thread UT passes. The balance 5 holds the upper thread UT on the inner surface of the balance hole. The balance 5 reciprocates in the Z-axis direction while holding the upper thread UT. The balance 5 reciprocates in conjunction with the needle bar 4. The balance 5 reciprocates in the Z-axis direction to draw out or pull up the upper thread UT.

  The thread tension device 6 applies tension to the upper thread UT. The upper yarn UT is supplied from the yarn supply source to the yarn tensioner 6. In the path through which the upper thread UT passes, the balance 5 is disposed between the sewing needle 3 and the thread tensioner 6. The thread tensioner 6 adjusts the tension of the upper thread UT supplied to the sewing needle 3 via the balance 5.

  The needle plate 7 supports the sewing object S. The sewing needle 3 held by the needle bar 4 and the needle plate 7 face each other. The needle plate 7 has a needle hole through which the sewing needle 3 can pass. The sewing needle 3 penetrating the sewing object S supported by the needle plate 7 passes through the needle hole.

  The presser member 8 presses the sewing object S from above. The presser member 8 is supported by the sewing machine head 2. The presser member 8 is disposed above the needle plate 7 and holds the sewing object S with the needle plate 7.

  The hook 9 holds the bobbin housed in the bobbin case. The shuttle 9 is disposed below the needle plate 7. The hook 9 rotates in the θX direction. The hook 9 rotates in conjunction with the needle bar 4. The hook 9 supplies the lower thread LT. The shuttle 9 passes through the sewing object S supported by the needle plate 7 and scoops the upper thread UT from the sewing needle 3 that has passed through the needle hole of the needle plate 7.

  The motor 10 generates power. The motor 10 includes a stator that is supported by the sewing head 2 and a rotor that is rotatably supported by the stator. As the rotor rotates, the motor 10 generates power. The power generated by the motor 10 is transmitted to each of the needle bar 4, the balance 5 and the shuttle 9 via a power transmission mechanism (not shown). The needle bar 4, the balance 5, and the hook 9 are interlocked. When the power generated by the motor 10 is transmitted to the needle bar 4, the needle bar 4 and the sewing needle 3 held by the needle bar 4 reciprocate in the Z-axis direction. When the power generated by the motor 10 is transmitted to the balance 5, the balance 5 reciprocates in the Z-axis direction in conjunction with the needle bar 4. When the power generated by the motor 10 is transmitted to the hook 9, the hook 9 rotates in the θX direction in conjunction with the needle bar 4 and the balance 5. The sewing machine 1 sews the sewing object S in cooperation with the sewing machine needle 3 held by the needle bar 4 and the shuttle 9.

  In the following description, the rotation of the rotor is referred to as the rotation of the motor 10. The rotation angle of the rotor is referred to as the rotation angle of the motor 10. The sewing needle 3 reciprocates as the motor 10 rotates. The balance 5 reciprocates in conjunction with the sewing needle 3 by the rotation of the motor 10. The shuttle 9 rotates in conjunction with the sewing needle 3 and the balance 5 by the rotation of the motor 10.

  The upper yarn UT from the yarn supply source is hung on the thread tensioner 6 and then hung on the sewing needle 3 via the balance 5. When the motor 10 rotates and the needle bar 4 is lowered, the sewing needle 3 held by the needle bar 4 passes through the sewing object S and passes through the needle hole provided in the needle plate 7. When the sewing needle 3 passes through the needle hole of the needle plate 7, the lower thread LT supplied from the hook 9 is hung on the upper thread UT hung on the sewing needle 3. In a state where the lower thread LT is hung on the upper thread UT, the sewing needle 3 rises and retreats from the sewing object S. When the sewing needle 3 passes through the sewing object S, the sewing machine 1 stops the sewing object S. When the sewing needle 3 moves away from the sewing object S, the sewing machine 1 moves the sewing object S in the + Y direction. The sewing machine 1 forms a seam SE on the sewing object S by reciprocating the sewing needle 3 while repeating the movement and stop of the sewing object S in the + Y direction. The seam SE formed on the sewing object S extends in the Y-axis direction.

<Motion diagram>
FIG. 2 is a diagram showing a motion diagram of the needle bar 4 and the balance 5 according to this embodiment. In FIG. 2, the horizontal axis represents the rotation angle [°] of the motor 10 with respect to the needle bar top dead center. The vertical axis shows the needle bar stroke and the balance stroke. The motion diagram shown in FIG. 2 shows a needle bar motion curve and a balance yarn supply curve.

  The needle bar top dead center is the position of the needle bar 4 closest to the + Z side in the movable range of the needle bar 4 in the Z-axis direction. The needle bar bottom dead center is the position of the needle bar 4 closest to the −Z side in the movable range of the needle bar 4 in the Z-axis direction. The balance top dead center means the position of the balance 5 on the most + Z side in the movable range of the balance 5 in the Z-axis direction. The bottom dead center of the balance means the position of the balance 5 closest to the −Z side in the movable range of the balance 5 in the Z-axis direction.

  The needle bar stroke refers to the position of the needle bar 4 in the Z-axis direction. The balance stroke refers to the position of the balance 5 in the Z-axis direction. The balance thread supply amount refers to the amount of the upper thread UT that the balance 5 supplies to the sewing needle 3.

  Each of the needle bar 4 and the balance 5 reciprocates in the Z-axis direction in conjunction with the rotation of the motor 10. When the rotation angle of the motor 10 is 0 [°], the needle bar 4 is disposed at the top dead center of the needle bar. When the rotation angle of the motor 10 is about 70 [°], the balance 5 is arranged at the top dead center of the balance. When the rotation angle of the motor 10 is about 180 [°], the needle bar 4 is disposed at the needle bar bottom dead center. When the rotation angle of the motor 10 is about 320 [°], the balance 5 is disposed at the bottom dead center of the balance. Based on the difference between the position of the needle bar 4 and the position of the balance 5 in the Z-axis direction, the tension of the upper thread UT changes.

<Seam inspection device>
The seam inspection device 20 inspects the seam SE formed on the sewing object S. The seam inspection device 20 inspects the seam SE formed on the sewing object S in a state where the sewing object S is supported by the needle plate 7. The seam inspection device 20 inspects the seam SE in parallel with the formation of the seam SE by the sewing machine 1. The seam inspection device 20 detects whether there is an abnormality in the seam SE formed on the sewing object S. Further, the seam inspection device 20 detects an abnormal pattern of the seam SE.

  FIG. 3 is a functional block diagram showing the seam inspection device 20 according to the present embodiment. As shown in FIGS. 1 and 3, the seam inspection device 20 includes a tension sensor 21 that detects the tension of the upper thread UT that is applied to the sewing needle 3, an angle sensor 22 that detects the rotation angle of the motor 10, and a tension sensor. 21, a processing device 30 that detects an abnormality in the seam SE formed on the sewing object S by the sewing needle 3, an output device 23 that outputs a detection result of the processing device 30, and an operation for the operator. The input device 24 is provided.

  The processing device 30 includes a computer system. The processing device 30 includes an arithmetic device 30A including a processor such as a CPU (Central Processing Unit), a nonvolatile memory such as a ROM (Read Only Memory) or a storage, and a volatile memory such as a RAM (Random Access Memory). A storage device 30B including an input / output interface 30C including an input / output circuit capable of inputting and outputting signals and data;

  The tension sensor 21 detects the tension of the upper thread UT supplied to the sewing needle 3. The tension sensor 21 includes, for example, a strain gauge and a piezoelectric element. As shown in FIG. 1, the tension sensor 21 is disposed between the balance 5 and the sewing needle 3 in the path through which the upper thread UT passes. The tension sensor 21 detects the tension of the upper thread UT between the balance 5 and the sewing needle 3. The tension sensor 21 is connected to the input / output interface 30C. The tension sensor 21 outputs a detected value of the tension of the upper thread UT to the processing device 30.

  The angle sensor 22 detects the rotation angle of the motor 10. The angle sensor 22 includes an encoder, for example. The angle sensor 22 detects the rotation angle of the motor 10 by setting the rotation angle of the motor 10 when the needle bar 4 is disposed at the top dead center of the needle bar to 0 [°]. The angle sensor 22 is connected to the input / output interface 30C. The angle sensor 22 outputs a detected value of the rotation angle of the motor 10 to the processing device 30.

  The output device 23 outputs the detection result of the processing device 30. The output device 23 is connected to the input / output interface 30C. Examples of the output device 23 include at least one of a display device and a printing device. The display device includes a flat panel display such as a liquid crystal display (LCD) or an organic EL display (OELD).

  The input device 24 generates input data when operated by an operator. The input device 24 is connected to the input / output interface 30C. The input device 24 outputs the generated input data to the processing device 30. Examples of the input device 24 include at least one of an operation button, a touch panel, and a computer keyboard.

  The arithmetic device 30A includes a detection value acquisition unit 31, a detection feature amount analysis unit 32, a determination unit 33, an output unit 34, and a reference feature amount generation unit 35. The storage device 30B includes a reference feature amount storage unit 36.

  The detection value acquisition unit 31 acquires the detection value of the tension sensor 21. Further, the detection value acquisition unit 31 acquires the detection value of the angle sensor 22.

  The detected feature amount analysis unit 32 analyzes the detection value of the tension sensor 21 acquired by the detection value acquisition unit 31, and calculates a detection value feature amount indicating the feature amount of the detection value of the tension sensor 21. The detected feature amount indicates an actual feature amount of the tension of the upper thread UT.

  The determination unit 33 determines whether there is an abnormality in the seam SE based on the analysis result of the detected feature amount analysis unit 32. Further, the determination unit 33 determines an abnormal pattern of the seam SE based on the analysis result of the detected feature amount analysis unit 32.

  The output unit 34 outputs output data including at least one of the detection value acquired by the detection value acquisition unit 31, the analysis result of the detection feature amount analysis unit 32, and the determination result of the determination unit 33 to the output device 23. The output device 23 outputs the output data from the output unit 34.

  The reference feature value generation unit 35 indicates the feature value of the tension of the upper thread UT when the normal seam SE is formed based on the detection value of the tension sensor 21 when the normal seam SE is formed. Calculate the amount. That is, in the present embodiment, the reference feature amount includes a normal feature amount that indicates the feature amount of the tension of the upper thread UT when the normal seam SE is formed.

  The reference feature quantity storage unit 36 stores the reference feature quantity of tension generated by the reference feature quantity generation unit 35.

  The determination unit 33 compares the detected feature value calculated by the detected feature value analysis unit 32 with the reference feature value stored in the reference feature value storage unit 36 to determine an abnormality in the seam SE. The determination of the abnormality of the seam SE includes the determination of the presence or absence of the abnormality of the seam SE and the determination of the pattern of the abnormality of the seam SE.

<Abnormal seam>
FIG. 4 is a diagram schematically showing a normal seam SE. 5 and 6 are diagrams schematically showing an abnormal seam SE.

  In the present embodiment, the sewing machine 1 is a lock stitch sewing machine that performs lock stitching. As shown in FIG. 4, the normal seam SE in the main stitch is formed linearly in the Y-axis direction. One normal seam SE is formed with a normal length Pn. Each of the plurality of seams SE is formed with a normal length Pn.

  There are a plurality of abnormal patterns of the seam SE. As an abnormality pattern of the seam SE, an abnormality of the first pattern in which at least one length of the seam SE is increased, an abnormality of the second pattern in which at least a part of the seam SE is eliminated, and an upper thread UT and a lower thread in the sewing object S The abnormality of the third pattern in which at least one of the yarns LT is loosened is exemplified. The abnormality of the first pattern is called “jumping”. The abnormality of the second pattern is called “thread break”. The abnormality in the third pattern is called “lantern”.

  FIG. 5 is a diagram illustrating an example of “eye skipping” that is an abnormality of the first pattern. “Stitch skipping” refers to a phenomenon in which the seam SE is formed with an abnormal length Pu longer than the normal length Pn without the seam SE being formed with a constant normal length Pn. As a pattern of “stitch skipping”, “single stitch skip” in which a stitch SE having an abnormal length Pu1 as shown in FIG. 5 (A) occurs once, and an abnormal length Pu1 as shown in FIG. 5 (B). Examples include “continuous stitch skip” in which the stitch SE is continuously generated and “long stitch skip” in which the stitch SE having an abnormal length Pu2 longer than the abnormal length Pu1 as shown in FIG. 5C is generated. The

  FIG. 6 is a diagram illustrating an example of a “lantern” that is an abnormality of the third pattern. The “lantern” is a phenomenon in which the upper thread UT is loosened on the surface of the sewing object S or the lower thread LT is loosened on the back surface of the sewing object S, although the upper thread UT and the lower thread LT are hung. . FIG. 6 shows an example in which the lower thread LT is loose on the back surface of the sewing object S.

<First method of seam abnormality detection>
A first method for detecting an abnormality in the seam SE will be described. In the first method, the determination unit 33 determines the abnormality of the seam SE based on the detection value of the tension sensor 21 and the detection value of the angle sensor 22.

  The tension sensor 21 outputs a detected value of the tension of the upper thread UT at a specified cycle (for example, every 0.1 [ms]) in the rotation of the motor 10. That is, the tension sensor 21 outputs a detection value at each of the plurality of rotation angles when the motor 10 rotates at each of the plurality of rotation angles.

  The reference feature amount is output from the tension sensor 21 at each of a plurality of rotation angles when the normal stitch SE as shown in FIG. 4 is formed and the motor 10 rotates at each of the plurality of rotation angles. The detected tension value is included. The reference feature amount is derived based on a preliminary experiment (including a simulation experiment) and stored in the reference feature amount storage unit 36 in advance.

  The detected feature amount is the detection of the actual tension output from the tension sensor 21 at each of the plurality of rotation angles when the motor SE rotates at each of the plurality of rotation angles when the seam SE is actually formed. Contains a value.

  In the present embodiment, the determination unit 33 determines the abnormality of the seam SE by comparing the detected feature amount derived when the rotation angle of the motor 10 is a specific angle with the reference feature amount.

  When determining whether or not “eye skipping” has occurred, the determination unit 33 detects the tension sensor 21 acquired when the rotation angle of the motor 10 is a specific angle of 270 ° to 360 °. The detected feature quantity derived from the value is compared with the reference feature quantity stored in the reference feature quantity storage unit 36 to determine whether or not “skipping” has occurred.

  FIG. 7 is a diagram illustrating the detection value of the tension sensor 21. In FIG. 7, the horizontal axis indicates the rotation angle of the motor 10, and the vertical axis indicates the detection value of the tension sensor 21. In FIG. 7, a range Sa and a range Sn indicate a range in which the rotation angle of the motor 10 is a specific angle of 270 [°] to 360 [°]. The determination unit 33 collates the detected feature amount derived when the rotation angle of the motor 10 is not less than 270 [°] and not more than 360 [°] with the reference feature amount, and determines whether or not “skipping” has occurred. Determine.

  When the normal seam SE is formed, the tension sensor 21 outputs a detection value as shown in the range Sn. When “skipping” occurs, the tension sensor 21 outputs a detection value as shown in the range Sa. As shown in FIG. 7, the detected value of the tension sensor 21 in the range Sn is different from the detected value of the tension sensor 21 in the range Sa. When the rotation angle of the motor 10 is in the range of 270 [°] to 360 [°], the tension of the upper thread UT when the “jumping” occurs is that of the upper thread UT when the normal seam SE is formed. It is also reduced by tension.

  In the present embodiment, a detected value as shown in the range Sn is stored in the reference feature quantity storage unit 36 as a reference feature quantity. The detection feature amount analysis unit 32 extracts a detection value in a range where the rotation angle of the motor 10 is 270 [°] or more and 360 [°] or less from the detection value of the tension sensor 21 acquired by the detection value acquisition unit 31. A detection feature amount for determining “skipping” is derived. The determination unit 33 collates the detected feature amount calculated by the detected feature amount analysis unit 32 with the reference feature amount stored in the reference feature amount storage unit 36, and determines whether or not “skipping” has occurred. Determine. When the detected feature value includes a detection value as shown in the range Sa, the determination unit 33 can determine that “skipping” has occurred based on the result of matching between the detected feature value and the reference feature value. it can. When the detected feature value includes a detected value as shown in the range Sn, the determination unit 33 can determine that “no skipping” has occurred based on the result of matching the detected feature value with the reference feature value. it can.

  When determining whether or not “thread breakage” has occurred, the determination unit 33 detects the tension sensor 21 acquired when the rotation angle of the motor 10 is a specific angle of 0 [°] or more and 90 [°] or less. The detected feature amount derived from the value is compared with the reference feature amount stored in the reference feature amount storage unit 36 to determine whether or not “thread breakage” has occurred.

  FIG. 8 is a diagram showing the detection value of the tension sensor 21. In FIG. 8, the horizontal axis indicates the rotation angle of the motor 10, and the vertical axis indicates the detection value of the tension sensor 21. In FIG. 8, the range Sa and the range Sn indicate a range where the rotation angle of the motor 10 is a specific angle of 0 [°] or more and 90 [°] or less. The determination unit 33 collates the detected feature amount derived when the rotation angle of the motor 10 is 0 [°] or more and 90 [°] or less with the reference feature amount, and determines whether or not “thread break” has occurred. Determine.

  When the normal seam SE is formed, the tension sensor 21 outputs a detection value as shown in the range Sn. When “thread break” occurs, the tension sensor 21 outputs a detection value as shown in the range Sa. As shown in FIG. 8, the detected value of the tension sensor 21 in the range Sn is different from the detected value of the tension sensor 21 in the range Sa. When the rotation angle of the motor 10 is in the range of 0 [°] to 90 [°], the tension of the upper thread UT when “thread breakage” occurs is that of the upper thread UT when the normal seam SE is formed. It becomes smaller than the tension.

  In the present embodiment, a detected value as shown in the range Sn is stored in the reference feature quantity storage unit 36 as a reference feature quantity. The detection feature amount analysis unit 32 extracts a detection value in a range where the rotation angle of the motor 10 is 0 [°] or more and 90 [°] or less from the detection value of the tension sensor 21 acquired by the detection value acquisition unit 31. A detection feature amount for determining “thread break” is derived. The determination unit 33 collates the detected feature amount calculated by the detected feature amount analysis unit 32 with the reference feature amount stored in the reference feature amount storage unit 36 to determine whether “thread break” has occurred. Determine. When the detected feature value includes a detection value as shown in the range Sa, the determination unit 33 can determine that “thread break” has occurred based on a result of matching between the detected feature value and the reference feature value. it can. When the detected feature value includes a detected value as shown in the range Sn, the determination unit 33 can determine that “thread break” has not occurred based on the result of matching between the detected feature value and the reference feature value. it can.

<Second method of seam abnormality detection>
A second method for detecting an abnormality in the seam SE will be described. Also in the second method, the determination unit 33 determines the abnormality of the seam SE based on the detection value of the tension sensor 21 and the detection value of the angle sensor 22.

  In the second method, the detected feature amount includes a detection cumulative curve indicating the cumulative value of the detection value of the tension sensor 21 during the rotation of the motor 10. The reference feature amount includes a reference cumulative curve indicating a cumulative value of tension when the seam SE is normal.

  The determination unit 93 includes a detection cumulative value curve indicating the cumulative value of the detection value of the tension sensor 21 in the rotation of the motor 10 and a reference cumulative curve indicating the cumulative value of the tension of the upper thread UT when the seam SE is normal. Collation is performed to determine whether the seam SE is abnormal. In the present embodiment, the determination unit 93 collates the detected cumulative curve with the reference cumulative curve to determine whether or not “lantern” has occurred.

  As described above, the tension sensor 21 outputs the detected value of the tension of the upper thread UT at a specified period during the rotation of the motor 10. That is, the tension sensor 21 outputs a detection value at each of the plurality of rotation angles when the motor 10 rotates at each of the plurality of rotation angles.

  The cumulative value refers to the cumulative value of the detected tension value detected by the tension sensor 21 at each of a plurality of rotation angles of the motor 10. The cumulative value increases as the detection time of the tension sensor 21 increases. The cumulative curve refers to a curve indicating a relationship between a plurality of rotation angles of the motor 10 and a cumulative value of tension at each of the plurality of rotation angles.

  The reference cumulative curve is a cumulative curve indicating a cumulative value of tension when a normal seam SE is formed, and is stored in advance in the reference feature amount storage unit 36.

  The detected cumulative curve is a cumulative curve that indicates the cumulative value of the actual detected values, and is calculated by the detected feature amount analysis unit 32.

  The determination unit 93 compares the detected cumulative curve with the reference cumulative curve to determine an abnormality in the seam SE.

  FIG. 9 is a diagram showing a cumulative curve. In FIG. 9, the horizontal axis indicates the rotation angle of the motor 10, and the vertical axis indicates the cumulative value of tension. FIG. 9 shows a cumulative curve when the motor 10 is, for example, 1000 revolutions / minute. For example, a detection value of the tension when the rotation angle of the motor 10 is 180 [°] is acquired, and then a detection value of the tension when the rotation angle is 181 [°] is acquired and added to the detection value of 180 [°]. . Similarly, a cumulative value is calculated by sequentially adding 182 [°], 183 [°],..., 0 [°],.

  When the normal seam SE is formed, a cumulative curve as shown by the curve Ln is generated. When the “lantern” occurs, a cumulative curve as shown by the curve La is generated. As shown in FIG. 9, the slope of the curve Ln is different from the slope of the curve La. The slope of the cumulative curve (curve La) when the “lantern” occurs is larger than the slope of the cumulative curve (curve Ln) when the normal seam SE is formed.

  In the present embodiment, a cumulative curve as indicated by the curve Ln is stored in the reference feature quantity storage unit 36 as a reference cumulative curve. Based on the detection value of the tension sensor 21 acquired by the detection value acquisition unit 31, the detection feature amount analysis unit 32 calculates a detection cumulative curve for determining “lantern”. The determination unit 33 collates the detected cumulative curve calculated by the detected feature amount analysis unit 32 with the reference cumulative curve stored in the reference feature amount storage unit 36 to determine whether or not “lantern” has occurred. judge. When the detected cumulative curve is a cumulative curve as indicated by the curve La, the determination unit 33 can determine that “lantern” has occurred based on the result of collation between the detected cumulative curve and the reference cumulative curve. . When the detected cumulative curve is a cumulative curve as indicated by the curve Ln, the determination unit 33 can determine that the “lantern” has not occurred based on the matching result between the detected cumulative curve and the reference cumulative curve. .

<Third method of seam abnormality detection>
A third method for detecting an abnormality in the seam SE will be described. In the third method, the detected feature amount includes a detection frequency spectrum calculated by Fourier transforming the detection value of the tension sensor 21 in the rotation of the motor 10. The reference feature amount includes a reference frequency spectrum calculated by Fourier transforming the tension when the seam SE is normal.

  The determination unit 93 calculates the detected frequency spectrum calculated by Fourier transforming the detected value of the tension sensor 21 in the rotation of the motor 10 and Fourier transforming the tension of the upper thread UT when the seam SE is normal. The reference frequency spectrum is checked to determine whether the seam SE is abnormal. The determination unit 93 can collate the detected frequency spectrum with the reference frequency spectrum to determine whether, for example, a “lantern” has occurred.

  FIG. 10 is a diagram showing a frequency spectrum of tension. In FIG. 10, the horizontal axis indicates the frequency of the detected tension value, and the vertical axis indicates the amplitude of the detected tension value. FIG. 10 shows the result of fast Fourier transform (FFT) of the detected tension value detected when the motor 10 is rotated at 1000 rpm, for example.

  When a normal seam SE is formed, a frequency spectrum as shown in FIG. 10A is generated. When the “lantern” occurs, a frequency spectrum as shown in FIG. 10B is generated. As shown in FIG. 10, when a “lantern” occurs, a peak PK is formed at a specific frequency. When the “lantern” occurs, the tension of the upper thread UT slightly vibrates, so that a peak PK is formed.

  In the present embodiment, a frequency spectrum as shown in FIG. 10A is stored in the reference feature quantity storage unit 36 as a reference frequency spectrum. Based on the detection value of the tension sensor 21 acquired by the detection value acquisition unit 31, the detection feature amount analysis unit 32 calculates a detection frequency spectrum for determining “lantern”. The determination unit 33 collates the detection frequency spectrum calculated by the detection feature amount analysis unit 32 with the reference frequency spectrum stored in the reference feature amount storage unit 36 to determine whether or not a “lantern” has occurred. judge. When the detected frequency spectrum is a frequency spectrum as shown in FIG. 10B, the determination unit 33 determines that “lantern” has occurred based on the result of matching between the detected frequency spectrum and the reference frequency spectrum. be able to. When the detected frequency spectrum is a frequency spectrum as shown in FIG. 10A, the determination unit 33 determines that “lantern” has not occurred based on the result of matching between the detected frequency spectrum and the reference frequency spectrum. be able to.

<Sewing method>
Next, the sewing method according to this embodiment will be described. FIG. 11 is a flowchart showing a sewing method according to this embodiment. As shown in FIG. 11, the sewing method according to the present embodiment includes a reference feature amount calculation process and a seam inspection process.

  The reference feature amount calculation process will be described. The sewing object S is installed on the throat plate 7 of the sewing machine 1. The sewing machine 1 starts sewing the sewing object S. The tension sensor 21 detects the tension of the upper thread UT in parallel with the sewing of the sewing object S. The tension sensor 21 outputs a detected value of the tension of the upper thread UT to the processing device 30. The reference feature value generation unit 35 acquires the detection value of the tension sensor 21 via the detection value acquisition unit 31 (step SA1).

  The seam SE formed on the sewing object S is confirmed. When the normal seam SE is formed, the input device 24 is operated by the operator so that input data indicating that the normal seam SE has been formed is generated. The reference feature value generation unit 35 acquires input data indicating that a normal seam SE has been formed (step SA2).

  Based on the detection value of the tension sensor 21 and the input data, the reference feature value generation unit 35 calculates a reference feature value indicating the feature value of the tension of the upper thread UT when the normal seam SE is formed (step) SA3).

  The reference feature value generation unit 35 stores the calculated reference feature value in the reference feature value storage unit 36 (step SA4).

  Next, the seam inspection process will be described. The sewing object S is installed on the throat plate 7 of the sewing machine 1. The sewing machine 1 starts sewing the sewing object S. The tension sensor 21 detects the tension of the upper thread UT in parallel with the sewing of the sewing object S. The tension sensor 21 outputs a detected value of the tension of the upper thread UT to the processing device 30. The detected feature amount analysis unit 32 acquires the detection value of the tension sensor 21 via the detection value acquisition unit 31 (step SB1).

  The detection feature amount analysis unit 32 analyzes the detection value of the tension sensor 21 and calculates the detection feature amount of the detection value (step SB2).

  The determination unit 33 collates the detected feature amount calculated by the detected feature amount analysis unit 32 with the reference feature amount stored in the reference feature amount storage unit 36 to determine an abnormality of the seam SE (step SB3). ).

  The determination unit 33 determines the abnormality of the seam SE based on at least one of the first method, the second method, and the third method of detecting the abnormality of the seam SE. The output unit 34 causes the output device 23 to output the determination result of the determination unit 33 (step SB4).

<Effect>
As described above, according to the present embodiment, the tension of the upper thread UT is detected by the tension sensor 21. Based on the detected value of the tension of the upper thread UT, it is possible to easily detect an abnormality in the seam SE. Therefore, the inspection of the seam SE can be performed efficiently.

  In the present embodiment, the reference feature amount is derived in advance based on the detected tension value when the normal seam SE is formed. Thereby, it is possible to efficiently determine the abnormality of the seam SE simply by comparing the reference feature amount with the detected feature amount.

  As in the first method and the second method of abnormality detection of the seam SE, not only the detection value of the tension sensor 21 but also the detection value of the angle sensor 22 is acquired, so that the detection value and angle of the tension sensor 21 are obtained. Based on the detection value of the sensor 22, the abnormality of the seam SE can be determined with high accuracy.

  Further, even if noise is included in the detected value of the tension sensor 21 by generating an accumulated curve based on the accumulated value of the tension as in the second method of detecting an abnormality of the seam SE, the influence of the noise is affected. Is reduced. Therefore, the abnormality of the seam SE can be determined with high accuracy.

  Further, the frequency spectrum of the tension detection value is generated as in the third method of detecting the abnormality of the seam SE, so that the abnormality of the seam SE can be detected with high accuracy without using the detection value of the angle sensor 22. Can be determined.

<Modification>
In the above-described embodiment, the abnormality of the seam SE is detected by comparing the detected feature quantity with the reference feature quantity. A threshold value related to the detection value of the tension may be provided, and an abnormality of the seam SE may be detected based on a comparison result between the detection value and the threshold value.

[Second Embodiment]
A second embodiment will be described. In the following description, the same or equivalent components as those of the above-described embodiment are denoted by the same reference numerals, and the description thereof is simplified or omitted.

  FIG. 12 is a flowchart showing a sewing method according to this embodiment. As shown in FIG. 12, the sewing method according to the present embodiment includes a learning process and an identification process.

  The learning process will be described. The sewing object S is installed on the throat plate 7 of the sewing machine 1. The sewing machine 1 starts sewing the sewing object S. The tension sensor 21 detects the tension of the upper thread UT in parallel with the sewing of the sewing object S. The tension sensor 21 outputs a detected value of the tension of the upper thread UT to the processing device 30. The reference feature value generation unit 35 acquires the detection value of the tension sensor 21 via the detection value acquisition unit 31 (step SC1).

  The reference feature value generation unit 35 gives a label to the detection value of the tension sensor 21 (step SC2). The label is a correct label indicating that a normal seam SE has been formed, a first abnormal label indicating that a “juncture” that is an abnormality of the first pattern has occurred, and a “thread break” that is an abnormality of the second pattern. And a third abnormality label indicating that a “lantern” that is an abnormality of the third pattern has occurred. The reference feature value generating unit 35 gives a correct answer label to the detection value of the tension sensor 21 acquired when the normal seam SE is formed. The reference feature value generating unit 35 gives a first abnormality label to the detection value of the tension sensor 21 acquired when the abnormality of the first pattern occurs. The reference feature value generation unit 35 gives a second abnormality label to the detection value of the tension sensor 21 acquired when the abnormality of the second pattern occurs. The reference feature value generation unit 35 gives a third abnormality label to the detection value of the tension sensor 21 acquired when the abnormality of the third pattern occurs.

  The reference feature value generation unit 35 performs machine learning on the detected value to which the label is attached using a specified machine learning algorithm (step SC3). Examples of the machine learning algorithm include at least one of a decision tree, a random forest, and a neural network.

  A learning model is generated by machine learning of the detection value to which the label is attached (step SC4). The generated learning model is stored in the reference feature amount storage unit 36.

  Next, the identification process will be described. The sewing object S is installed on the throat plate 7 of the sewing machine 1. The sewing machine 1 starts sewing the sewing object S. The tension sensor 21 detects the tension of the upper thread UT in parallel with the sewing of the sewing object S. The tension sensor 21 outputs a detected value of the tension of the upper thread UT to the processing device 30. The detected feature quantity analysis unit 32 acquires the detection value of the tension sensor 21 via the detection value acquisition unit 31 (step SD1).

  The detection feature amount analysis unit 32 analyzes the detection value of the tension sensor 21 and calculates the detection feature amount of the detection value. The determination unit 33 identifies an abnormal pattern of the seam SE based on the detected feature amount calculated by the detected feature amount analysis unit 32 and the learning model stored in the reference feature amount storage unit 36 (step SD2 ).

  The output unit 34 causes the output device 23 to output the identification result of the determination unit 33 (step SD3). For example, when the abnormal pattern of the seam SE is identified as the first pattern, the output unit 34 outputs output data indicating that the “juncture”, which is an abnormality of the first pattern, has occurred to the output device 23. Let When the abnormal pattern of the seam SE is identified as the second pattern, the output unit 34 causes the output device 23 to output output data indicating that a “thread break” that is an abnormality of the second pattern has occurred. When the abnormal pattern of the seam SE is identified as the third pattern, the output unit 34 causes the output device 23 to output output data indicating that a “lantern” that is an abnormality of the third pattern has occurred. When the normal seam SE is identified, the output unit 34 causes the output device 23 to output output data indicating that the seam SE is normal.

  As described above, machine learning (supervised learning) may be performed based on the detection value of the tension sensor 21, and the abnormality of the seam SE may be detected based on the generated model.

  DESCRIPTION OF SYMBOLS 1 ... Sewing machine, 2 ... Sewing head, 3 ... Sewing needle, 4 ... Needle bar, 5 ... Balance, 6 ... Thread tensioner, 7 ... Needle plate, 8 ... Presser member, 9 ... Hook, 10 ... Motor, 20 ... Seam Inspection device, 21 ... tension sensor, 22 ... angle sensor, 23 ... output device, 24 ... input device, 30 ... processing device, 30A ... arithmetic device, 30B ... storage device, 30C ... input / output interface, 31 ... detection value acquisition unit , 32 ... detected feature amount analysis unit, 33 ... determination unit, 34 ... output unit, 35 ... reference feature amount generation unit, 36 ... reference feature amount storage unit, LT ... lower thread, S ... sewing object, SE ... seam, UT ... Upper thread.

Claims (7)

A tension sensor that detects the tension of the upper thread that is hung on the sewing needle;
A processing device for detecting an abnormality of a seam formed on a sewing object by the sewing needle based on a detection value of the tension sensor;
A seam inspection device comprising: The processor is
A detection feature amount analysis unit that analyzes a detection value of the tension sensor and calculates a detection feature amount indicating a feature amount of the detection value;
A reference feature quantity storage unit for storing a reference feature quantity of the tension;
A determination unit that compares the detected feature quantity with the reference feature quantity to determine the abnormality;
Having
The seam inspection device according to claim 1. The sewing needle is reciprocated by the rotation of a motor,
The tension sensor outputs the detection value at a specified period in the rotation of the motor,
An angle sensor for detecting a rotation angle of the motor;
The determination unit determines the abnormality based on a detection value of the tension sensor and a detection value of the angle sensor.
The seam inspection device according to claim 2. The determination unit determines the abnormality by comparing the detected feature amount derived when the rotation angle of the motor is a specific angle with the reference feature amount.
The seam inspection device according to claim 3. The detected feature amount includes a detection cumulative curve indicating a cumulative value of the detection value in rotation of the motor,
The reference feature amount includes a reference cumulative curve indicating a cumulative value of the tension when the seam is normal.
The seam inspection device according to claim 3. The sewing needle is reciprocated by the rotation of a motor,
The tension sensor outputs the detection value at a specified period in the rotation of the motor,
The detection feature amount includes a detection frequency spectrum calculated by Fourier transforming the detection value in the rotation of the motor,
The reference feature amount includes a reference frequency spectrum calculated by Fourier transforming the tension when the seam is normal.
The seam inspection device according to claim 2. The processor is
A reference feature generating unit that generates a learning model by machine learning of the detection value of the tension sensor to which a label is attached;
A determination unit for identifying an abnormality pattern of the seam based on a detection feature amount of a detection value of the tension sensor and the learning model;
Having
The seam inspection device according to claim 1.

Images