Method for evaluating pleat shape and method for manufacturing pleat body

Abstract

A method for evaluating the pleat shape of a pleated body and a method for manufacturing a pleated body are provided. The method for evaluating the pleat shape of a pleated body (20) obtained by processing a sheet material (30) into a pleated shape includes: an extraction step of extracting feature quantities related to the pleat shape from image data (56) of the pleated body (20); and an evaluation step of evaluating the pleat shape based on the feature quantities extracted in the extraction step.

Description

Technical Field

[0001] This disclosure relates to a method for evaluating the shape of pleats and a method for manufacturing pleats. Background Technology

[0002] Previously, such as the filter media described in Patent Document 1 (Japanese Patent Application Publication No. 2016-026870), a pleated body is known for use by processing a sheet into a pleated shape.

[0003] Such pleats are intended to be processed into the desired pleat shape by a processing machine to achieve their intended function. Summary of the Invention

[0004] The technical problem that the invention aims to solve

[0005] However, in order to process sheet material into pleats of the desired shape, it is sometimes necessary to adjust the manufacturing conditions according to the material or properties of the sheet material, the settings of the processing machine, or the usage environment. Currently, the manufacturing conditions rely on the ability of skilled personnel who can visually adjust the pleat shape based on their senses.

[0006] Therefore, it is desirable to be able to objectively evaluate the pleat shape of the pleat body.

[0007] Technical solutions adopted to solve technical problems

[0008] The first evaluation method is an evaluation method for the pleat shape of a pleated body obtained by processing sheet material into a pleated shape. It includes an extraction step and an evaluation step. In the extraction step, features related to the pleat shape are extracted from the image data of the pleated body. In the evaluation step, the pleat shape is evaluated based on the features extracted in the extraction step.

[0009] This evaluation method allows for an objective assessment of the pleat shape of a pleated body.

[0010] The evaluation method for the second viewpoint is based on the evaluation method for the first viewpoint, with the pleats being the filter material.

[0011] This evaluation method can objectively evaluate the pleat shape of the filter media, and therefore can also objectively evaluate the performance of the filter media.

[0012] The evaluation method for the third viewpoint is based on the evaluation method for the first or second viewpoint, and includes a process in the evaluation process that compares the feature quantities extracted in the extraction process with a specified benchmark.

[0013] This evaluation method can determine whether the pleat shape of the pleat body meets the specified benchmarks.

[0014] The evaluation method for the fourth viewpoint is based on the evaluation methods for any of the first to third viewpoints. The characteristic quantity is any one of the following characteristic quantities and combinations thereof: pleat spacing; pleat height; the length of the connecting portion linking adjacent peaks and valleys in the second direction when viewed along the first direction; the curvature of the connecting portion when viewed along the first direction; the symmetry of adjacent connecting portions when viewed along the first direction; and the length and width of the spacing retaining portion when the pleat body is provided with a spacing retaining portion to maintain the pleat spacing. Here, the first direction is the direction in which the broken line extends. The second direction is the direction in which the pleat shape is continuous. The third direction is the direction perpendicular to both the first and second directions.

[0015] This evaluation method can improve the objectivity of pleat shape evaluation.

[0016] The evaluation method for the fifth viewpoint builds upon the evaluation methods for any of the first through fourth viewpoints, and further includes an evaluation model learning process. In this process, multiple sets of image data of pleats and corresponding sets of completed evaluation data are established. Machine learning is then applied to the relationship between the pleat image data and the completed evaluation data to create an evaluation model. Finally, the evaluation model created in the evaluation model learning process is used to evaluate the features extracted in the extraction process.

[0017] According to this evaluation method, the objectivity of the evaluation results is improved.

[0018] The evaluation method for the sixth viewpoint, based on the evaluation methods for any of the first through fifth viewpoints, further includes an input condition learning process. In this process, multiple sets of input conditions and evaluation result data are used to learn the relationship between the input conditions and the evaluation result data. Input conditions include processing condition data input to the machine that processes sheet material into pleats, as well as physical property data related to the sheet material. Evaluation result data is obtained by evaluating the pleats obtained corresponding to the input conditions in the evaluation process.

[0019] According to this evaluation method, the objectivity of the evaluation results is further improved.

[0020] The evaluation method of the seventh point is based on the evaluation method of the sixth point. The processing condition data includes any one of the following data and combinations thereof: the gap between the folding blade and the folding plate of the processing machine; the transport speed of the processing machine when transporting materials; and the amount of resin applied to the material to form the spacer.

[0021] According to this evaluation method, the objectivity of the evaluation results is further improved.

[0022] The evaluation method for the eighth viewpoint is based on the evaluation method for the sixth or seventh viewpoint. It outputs the input conditions to the output device based on the model obtained in the input condition learning process.

[0023] Based on this evaluation method, we can grasp the input conditions based on objective evaluation.

[0024] The ninth viewpoint's manufacturing method uses the input conditions output in the eighth viewpoint's evaluation method to manufacture pleats.

[0025] This manufacturing method makes it easy to achieve the desired pleat shape in the manufactured pleats. Attached Figure Description

[0026] Figure 1 This is a schematic perspective view of an example of a pleated structure.

[0027] Figure 2 This is a schematic perspective view of an example of a pleated body with a spacer retaining portion.

[0028] Figure 3 It is a diagram showing the shape of the pleats when viewed along the first direction.

[0029] Figure 4 This is a cross-sectional view of an example of sheet material.

[0030] Figure 5 This is a 3D view of the air filter unit.

[0031] Figure 6 This is a schematic diagram of a device for extending sheet-like materials.

[0032] Figure 7 The diagram schematically illustrates a device for extending sheet material in the width direction (left half) and a device for laminating a breathable support onto the sheet material (right half).

[0033] Figure 8 This is a schematic diagram of a reciprocating folding machine.

[0034] Figure 9 This is a schematic diagram showing the apparatus used for coating the spacer and for pleating reprocessing.

[0035] Figure 10 This diagram shows the retraction of the folding knife during the pleating reprocessing step.

[0036] Figure 11 This diagram shows the state of the folding knife moving to the fold-in position during the pleating reprocessing process.

[0037] Figure 12 This is a system structure diagram of an example of a camera and evaluation information processing device.

[0038] Figure 13 This is a flowchart of the evaluation process performed by the evaluation information processing device.

[0039] Figure 14 This is another example of a system architecture diagram for a camera and an evaluation information processing device.

[0040] Figure 15 This is a flowchart of the evaluation process performed by the evaluation information processing device.

[0041] Figure 16 This is a system structure diagram of an example of a processing machine, a camera, and an evaluation information processing device.

[0042] Figure 17 This is a flowchart of the manufacturing process for pleats.

[0043] Figure 18 It is a side view photograph of the pleats obtained by changing the gap between the folding knife and the folding plate. Detailed Implementation

[0044] (1) Evaluation method of pleat shape

[0045] In the evaluation method for pleat shape, the pleat shape of a pleated body obtained by processing a sheet of material into a pleated shape is evaluated. This evaluation method includes an extraction step and an evaluation step. In the extraction step, features related to the pleat shape are extracted from the image data of the pleated body. In the evaluation step, the pleat shape is evaluated based on the features extracted in the extraction step. Therefore, the pleat shape of the pleated body can be objectively evaluated without relying on visual evaluation by a skilled person.

[0046] The pleat shape of the pleat body, for example, is shown in a schematic perspective view of an example of pleat body 20. Figure 1 As shown, the shape can be formed by repeatedly folding the sheet material 30 through peak and valley folds, with multiple peak fold lines X and multiple valley fold lines Y arranged in parallel. Here, the straight lines orthogonal to the multiple peak fold lines X and the straight lines orthogonal to the multiple valley fold lines Y can be parallel to each other.

[0047] Furthermore, there are no particular limitations on the method for acquiring image data of the pleats; for example, it can be obtained by taking pictures from the direction in which the peak-to-valley lines extend. Also, there are no particular limitations on the image data of the pleats; for example, the viewfinder can be used to capture 3 to 30 pairs of peaks and valleys using the camera. Moreover, when acquiring images of the pleats from a specified position using the camera, it is preferable to take pictures at predetermined time intervals, where the peak-to-valley pairs move a predetermined number of times as the pleats are moved in the transport direction.

[0048] (2) Extraction process

[0049] In the extraction process, features related to the shape of the pleats are extracted from the image data of the pleats. This extraction process is preferably performed using an information processing device, such as a computer with a processor (CPU or GPU) and memory (RAM or ROM).

[0050] (2-1) Characteristic quantities of pleat shape

[0051] As a feature quantity related to the pleat shape, it can be obtained by numerically representing the shape of the edge portion of the curve or straight line visible when viewing the pleat shape from the direction extending from the peak and valley pleats. It can consist of only one feature quantity or can be composed of multiple feature quantities. This can improve the objectivity of the evaluation of the pleat shape.

[0052] More specifically, the characteristic quantities are, for example, any of the following characteristic quantities and combinations thereof: pleat spacing; pleat height; the length of the connecting portion that connects adjacent peaks and valleys when viewed along the first direction in the second direction; the degree of curvature of the connecting portion when viewed along the first direction; the symmetry of adjacent connecting portions when viewed along the first direction; and the length and width of the spacer holding portion when the pleat body is provided with a spacer holding portion to maintain the pleat spacing. Here, the first direction is the direction in which the multiple peak fold lines X and the multiple valley fold lines Y extend. The second direction is the direction in which the pleat shape is continuous, or the direction in which the multiple peak fold lines X are arranged, or the direction in which the multiple valley fold lines Y are arranged. The third direction is the direction perpendicular to both the first and second directions. In addition, for pleated filter media, the third direction can also be the direction through which the fluid being treated passes.

[0053] Here, as Figure 2 As shown in the schematic perspective view, the pleated body may be provided with either a peak spacing holding portion 26 for maintaining the spacing between adjacent peak portions and a valley spacing holding portion 27 for maintaining the spacing between adjacent valley portions, or both. The peak spacing holding portion 26, valley spacing holding portion 27, and other spacing holding portions are arbitrary, and their position, shape, thickness, etc., on the pleated body are also arbitrary. Furthermore, in Figure 2 The view shows the state after the folded gap of the pleated body 20 with the spacer retaining portion has been slightly expanded, but as seen in the cross-sectional view along the fold line direction... Figure 3 As shown, the peak spacing holding portion 26 has the function of maintaining a gap between adjacent peaks to prevent opposing surfaces from contacting each other, and the valley spacing holding portion 27 has the function of maintaining a gap between adjacent valleys to prevent opposing surfaces from contacting each other. Additionally, in Figure 3Among them, the depth direction of the paper surface is the first direction, the left - right direction of the paper surface is the second direction, and the up - down direction of the paper surface is the third direction. In addition, in Figure 3 it is exemplified that the peak interval holding part 26 is provided outside the peak fold line X in the third direction and the valley interval holding part 27 is provided outside the valley fold line Y in the third direction for the interval holding part.

[0054] Hereinafter, referring to Figure 3 , the parameters of the characteristic quantities will be described.

[0055] The pleat pitch a refers to the length in the second direction between adjacent peak fold lines X, and can also be said to be the length in the second direction between multiple valley fold lines Y. Additionally, here, the peak fold line X refers to the end in the third direction of the peak part. Moreover, the valley fold line Y refers to the end in the third direction of the valley part. In addition, as an alternative to the pleat pitch a, a value obtained by subtracting the thickness of the sheet from the pleat pitch a can also be used as a parameter.

[0056] The pleat height b refers to the length in the third direction between the intersection point Z of the line connecting adjacent peak fold lines X and the line extending from the valley fold line Y between the adjacent peak fold lines X in the third direction and the valley fold line Y. Moreover, the pleat height b refers to the length in the third direction between the intersection point Z of the line connecting adjacent valley fold lines Y and the line extending from the peak fold line X between the adjacent valley fold lines Y in the third direction and the peak fold line X.

[0057] When observing in the first direction, the length c in the second direction of the connecting part 35 connecting adjacent peaks and valleys in Figure 3 refers to the length (c1) in the second direction between the right - most undulating part on the valley side and the left - most undulating part on the peak side in the connecting part 35a connecting the valley and the peak adjacent to its right. In addition, when observing in the first direction, the length c in the second direction of the connecting part 35 connecting adjacent peaks and valleys in Figure 3 refers to the length (c2) in the second direction between the right - most undulating part on the peak side and the left - most undulating part on the valley side in the connecting part 35b connecting the peak and the valley adjacent to its right. Here, in Figure 3 , the right side can be called one side in the second direction, and the left side can be called the other side in the second direction.

[0058] The degree of curvature of the connecting portion when viewed along the first direction is not particularly limited. For example, it can be the ratio (c / b) of the length c of the connecting portion in the second direction to the length d of the connecting portion in the third direction. The larger the value of this ratio, the greater the degree of curvature can be assessed. Furthermore, for example, in the case where the pleats are filter media, it can be assessed that: the larger the value of the ratio (c / b), the greater the degree of curvature, which narrows the flow path of the fluid, reduces the effective surface area, and thus leads to an increase in pressure loss.

[0059] When viewed along the first direction, the symmetry of adjacent connecting parts is not particularly limited; for example, it could be the difference in the degree of curvature of adjacent connecting parts. More specifically, it could be, for example, the difference in the ratio (c / b) of adjacent connecting parts.

[0060] When a pleat body is provided with a spacing retaining portion to maintain the pleat spacing, the length of the spacing retaining portion can be the length of the portion extending in a direction intersecting the peak fold line X or the valley fold line Y. Alternatively, as an alternative to this length, for example, the ratio of the length of the spacing retaining portion in a third direction to the pleat height b can be used.

[0061] When the pleat body is provided with a spacing holding part to maintain the pleat spacing, the width of the spacing holding part can also be the length in the first direction parallel to the peak fold line X or the valley fold line Y.

[0062] Furthermore, when the interval holding portion for maintaining the pleat interval is provided on the pleat body at a predetermined interval in the first direction, the interval in the first direction can also be further used as a characteristic quantity.

[0063] (3) Evaluation process

[0064] In the evaluation process, the pleat shape is evaluated based on the feature quantities extracted in the extraction process. The evaluation process here preferably uses an evaluation information processing device for extraction, such as a computer equipped with a processor (CPU or GPU) for various information processing functions and a memory (RAM or ROM).

[0065] The evaluation process preferably includes, for example, a step of comparing the feature quantities extracted in the extraction process with a predetermined benchmark. This allows it to determine whether the pleat shape of the pleat body meets the predetermined benchmark.

[0066] Furthermore, the prescribed benchmark can be set for each of the above-mentioned characteristic quantities, or it can be determined by using at least any two of the above-mentioned characteristic quantities in mathematical formulas.

[0067] Furthermore, there are no particular limitations on the method for determining the specified benchmark. For example, it can be a benchmark used to ensure that the pleat shape of the obtained pleat body meets the desired conditions.

[0068] Furthermore, for example, when the pleats are filter media, a specified standard can be determined so that the pressure loss of the fluid being treated as it passes through the obtained pleats meets specified conditions.

[0069] (4) Uses of pleats

[0070] The material 30 processed into a pleated sheet shape is not particularly limited; examples include filter media and adsorbents. Such filter media or adsorbents can be configured to have multiple layers. As a filter media, for example... Figure 4 As shown in the cross-sectional view, it may include a main trapping layer 31, a first support layer 32 stacked on one side thereon, and a second support layer 33 stacked on the other side. In addition, as filter media, the support layers may be stacked only on one side or the other side relative to the main trapping layer 31 in the direction through which the fluid being treated passes.

[0071] In addition, the filter media can be air filter media where the fluid being processed is a gas, filter media where the fluid being processed is a liquid, filter media where the fluid being processed is a powder, or a mixture of at least two of the fluids being processed: gas, liquid, and powder.

[0072] Examples of filter media include glass fiber filter media, organic fiber filter media, and filter media containing nanofibers. Examples of organic fiber filter media include fluoropolymer porous membranes such as porous membranes made of polytetrafluoroethylene (hereinafter, sometimes referred to as PTFE). Compared to glass fiber filter media, such fluoropolymer porous membranes have higher dust collection efficiency under the same pressure loss, and are therefore particularly suitable for use in HEPA filters (High Efficiency Particulate Air Filters) or ULPA filters (Ultra-Low Permeability Air Filters).

[0073] Here, when a pleated filter media is used for the treated fluid to pass through in a direction intersecting the plane containing multiple peak fold lines X and multiple valley fold lines Y, the pleated shape deforms due to resistance. The opposing surfaces on the downstream side of the filter media come into contact with each other, resulting in increased resistance, reduced effective filtration area, and potentially decreased performance. This deformation of the pleated shape is more likely to occur when the uniformity of the pleated shape is low; therefore, a high degree of uniformity in the pleated shape of the obtained pleated body is required. In this evaluation method, since the pleated shape of the filter media can be objectively evaluated, the performance of the filter media can also be objectively evaluated. Furthermore, a pleated body whose pleated shape receives a good evaluation according to this evaluation method can be presumed to have good filter media performance.

[0074] (5) Units including pleats

[0075] Next, refer to Figure 5 The unit including pleats will be described.

[0076] The unit 100 including the pleats includes the pleats 20 described above and the frame 25 that houses the pleats 20.

[0077] The frame 25 is made of a combination of sheets such as resin or metal, and the pleats 20 and the frame 25 are preferably sealed with a sealant. The sealant is used to prevent leakage between the pleats 20 and the frame 25, and sealants made of resins such as epoxy, acrylic, or polyurethane can be used.

[0078] (6) Manufacturing method of filter media

[0079] Figure 6 and Figure 7 A general example of a method for manufacturing a filter material, such as a sheet-like material, is shown. Furthermore, a PTFE porous membrane, such as a fluoropolymer porous membrane, will be used as an example for explanation.

[0080] exist Figure 6 In the diagram, 1 is the take-up roller, 2 is the winding roller, 3-5 are rollers, 6 and 7 are heating rollers, and 8-12 are rollers. Furthermore, in... Figure 7 In the diagram, 14 is the roll-out roller, 15 is the preheating zone, 16 is the extension zone, 17 is the heat-fixing zone, 19 is the laminating roller, and 21 is the winding roller.

[0081] In the porous membrane fabrication process, through Figure 6 The process shown and Figure 7 The process shown in the left half involves biaxially stretching an unburned PTFE membrane to create a porous PTFE membrane. This fluoropolymer porous membrane is thin and flexible; therefore, it is supported by a laminated support structure in the following processes.

[0082] In the hot lamination process, through Figure 7 The process shown in the right half involves heat-laminating a breathable support member 18 made of nonwoven fabric on both sides of a PTFE porous membrane to obtain a sheet-like material 30 as a filter media.

[0083] In the winding process, the sheet material 30, which serves as the filter media, is wound onto the winding roller 21.

[0084] Furthermore, the filter media is preferably suitable for use as a HEPA filter, ULPA filter, or medium-efficiency filter. These filter media utilizing fluoropolymer porous membranes have very fine fiber diameters, exhibiting a fine fibrous structure and a small membrane thickness. Additionally, the membrane thickness of the fluoropolymer porous membrane can be, for example, 1 μm to 100 μm, and from the viewpoint of reducing pressure loss, is preferably 1 μm to 50 μm, and more preferably 1 μm to 30 μm. Moreover, the average fiber diameter of the fluoropolymer porous membrane can be, for example, 0.01 μm to 0.25 μm, preferably 0.05 μm to 0.2 μm. This average fiber diameter can be calculated as the arithmetic mean fiber diameter by randomly selecting 50 fibers from a scanning electron microscope image.

[0085] (7) Manufacturing method of pleats

[0086] Figure 8 and Figure 10 An outline of an example of a method for manufacturing the pleats 20 is shown.

[0087] The manufacturing method of the pleated body includes a roll-out process, a pleating process, an unfolding process, a coating process for the spacer holding part, and a pleating reprocessing process.

[0088] In the unwinding process, the sheet material 30 that was wound onto the winding roller 21 in the aforementioned winding process is unwound. In addition, the material 30 wound onto the winding roller 21 sometimes develops curl marks, and the curvature of the winding is greater at positions where the radius of the winding roller 21 is shorter, thus there is a tendency for the curl marks to become larger.

[0089] In the pleating process, such as Figure 8 As shown, a pair of upper folding blades 41 and lower folding blades 42 of a processing machine are used to alternately fold the sheet material 30 back and forth, processing it into a wavy shape (pleating), thereby forming a wavy fold line. Furthermore, while the processing machine feeds the sheet material 30 out at the folding processing section between the upper folding blades 41 and lower folding blades 42, it simultaneously folds the material 30 inward using the upper folding blades 41 and lower folding blades 42. Therefore, the material 30 is folded inward by the upper folding blades 41 and lower folding blades 42 under force in the transport direction. Here, the transport speed of the material 30 can be adjusted by receiving setting inputs, etc. Afterward, the material 30, processed into a wavy shape, is heated by the heater 22 to impart creases.

[0090] In the unfolding process, such as Figure 9 As shown, the pleated material 30 is unfolded into a sheet.

[0091] In the coating process of the spacer holding part, such as Figure 9As shown, the processing machine coats the unfolded material 30 with a resin or other material used to form peak spacing holding portions 26 and 28, valley spacing holding portions 27 and 29, etc. The coating of the spacing holding portions is performed by applying molten resin or the like through nozzles arranged at predetermined intervals in a direction orthogonal to both the transport direction and the thickness direction of the material 30. Furthermore, the processing machine can adjust the coating amount of the spacing holding portions by receiving setting inputs. Specifically, the coating amount is adjusted by regulating the discharge speed of the molten resin relative to the transport speed.

[0092] In the pleating reprocessing step, the material 30 coated with the spacer retainer is pleated again using a pair of upper folding blades 43 and lower folding blades 44 of a processing machine to obtain a pleated body 20. Furthermore, the upper folding blades 43 and lower folding blades 44 are formed in a comb-like shape to avoid interference with the spacer retainer. Similarly to the pleating process described above, the processing machine feeds the sheet-like material 30 out at the folding processing section between the upper folding blades 43 and lower folding blades 44 while simultaneously folding the material 30 inward using the upper folding blades 43 and lower folding blades 44. Therefore, the material 30 is folded inward by the upper folding blades 43 and lower folding blades 44 while under force in the transport direction. Here, the transport speed of the material 30 can be adjusted by receiving setting inputs, etc.

[0093] In addition, in the pleat reprocessing process, such as Figure 10 As shown, the material 30 is transported along the transport direction starting from the state where the upper folding blade 43 moves upward and backward and the lower folding blade 44 moves downward. Figure 11 As shown, the upper folding blade 43 descends to the lower folding position, and the lower folding blade 44 rises to the upper folding position, thus processing the material 30 into a pleated shape. Furthermore, the pleating process and the pleating reprocessing process are sometimes collectively referred to as pleating processing below. Here, as... Figure 11 As shown, the pleating machine can adjust the vertical distance u between the upper pleating plate 45 and the lower pleating plate 46 by receiving setting inputs, etc., to correspond to the pleat height b or a length greater than a predetermined amount relative to the pleat height b. Additionally, the pleating machine can adjust the vertical gap s when the upper surface of the lower pleating plate 46 is closest to the lower end of the upper pleating blade 43 by receiving setting inputs, etc. Similarly, the pleating machine can adjust the vertical gap t when the lower surface of the upper pleating plate 45 is closest to the upper end of the lower pleating blade 44 by receiving setting inputs, etc.

[0094] As described above, in a pleating machine, the shape of the obtained pleated body can be changed by adjusting the material 30's transport speed, gap s, gap t, and the amount of resin or other material applied to form the spacer holding portion. For example, if the material 30's transport speed changes, the pressure applied to the material 30 by the folding blades such as the upper folding blade 43 and lower folding blade 44 changes, sometimes resulting in a change in the length c of the connecting portion 35 connecting adjacent peaks and valleys in the second direction or a change in the pleat spacing a, thereby changing the shape of the obtained pleated body. Furthermore, if the amount of resin or other material applied to form the spacer holding portion changes, sometimes a change in the pleat spacing a occurs, thereby changing the shape of the obtained pleated body. Moreover, if the gap s or gap t changes, the pressure applied to the material 30 by the folding blades such as the upper folding blade 43 and lower folding blade 44 changes, the bending position of the material 30 by the folding blades changes, sometimes resulting in a change in the length c of the connecting portion 35 connecting adjacent peaks and valleys in the second direction or a change in the pleat spacing a, thereby changing the shape of the obtained pleated body. Furthermore, when the material 30 has high stiffness or high thickness, it is difficult to form creases during pleating, thus the shape of the pleated body will change.

[0095] Image data of the pleats can be obtained, for example, by taking a picture of the pleats obtained after the pleating reprocessing process using a camera 48.

[0096] (8) An example of the evaluation of pleat shape

[0097] The following explanation will focus on the pleats used as air filter media, using the average length c of the connecting portion that connects adjacent peaks and valleys in the second direction in an image as an indicator, and evaluating the pleat shape using the evaluation information processing device 50 and the camera 48.

[0098] Here, as Figure 9 As shown, camera 48 is configured to capture images of the pleated body after the pleating reprocessing has been completed and a pleated shape has been formed, from the direction of the crease extension. Furthermore, as... Figure 12 As shown, the camera 48 stores the image data 49 of the captured pleats and can send it to the evaluation information processing device 50 via a communication line.

[0099] The evaluation information processing device 50 is an information processing device such as a computer, which has a processor 51 (CPU or GPU, graphics processing unit) for performing various information processing, a memory 53 configured with RAM or ROM, and a display screen 52 capable of displaying and outputting data. In the memory 53, image data 49 sent from the camera 48 is directly stored as image data 56. Furthermore, the memory 53 also stores an indexing program 54 for indexing the image data 56 and reference value data 55.

[0100] In the structure shown above, Figure 13 The evaluation process performed by the evaluation information processing device 50 is shown in the figure.

[0101] In step S11, the processor 51 directly stores the image data 49 sent from the camera 48 as image data 56 in the memory 53.

[0102] In step S12, the processor 51 executes the indexing program 54 stored in the memory 53 on the image data 56 obtained in step S11, thereby calculating the feature quantity, i.e., the index value, corresponding to the image data 56. Specifically, it calculates the multiple connection parts 35 (refer to...) contained in an image data 56. Figure 3 The average value of the length c in the second direction is used as the index value. Specifically, the calculation... Figure 3 The sum of c1 and c2 shown in an image, divided by the number of connected portions 35 in that image, is used as the index value.

[0103] In step S13, the processor 51 determines whether the index value calculated in step S12 is less than the reference value shown in the reference value data stored in the memory 53. If the value is less than the reference value, the process proceeds to step S14; if the value is greater than the reference value, the process proceeds to step S16.

[0104] In step S14, the processor 51 determines that: since the average length c of the connecting portion 35 of the pleat body 20 in the second direction is less than the reference value, the undulation is small and the pleat shape is good.

[0105] In step S15, the processor 51 outputs the information of the good product to the display screen 52 and then moves on to the next image data acquisition process.

[0106] In step S16, the processor 51 determines that: because the average length c of the connecting portion 35 of the pleat body 20 in the second direction is greater than the reference value, the undulation is large and the pleat shape is poor.

[0107] In step S17, the processor 51 outputs the information of the defective product to the display screen 52 and then moves on to the next image data acquisition process.

[0108] Furthermore, in the above processing, the reference value data 55 can be determined based on the index values ​​of pleats in the obtained pleat body 20 that are considered to have a good pleat shape. Additionally, the reference value data 55 can be determined based on the index values ​​of pleats in the obtained pleat body 20 where the measured value of the specified pressure loss is confirmed to be lower than a specified value.

[0109] (9) An example of machine learning for evaluating methods

[0110] The following describes the pleats used as air filter media, using the average length c of the connecting portion that links adjacent peaks and valleys in the second direction in an image as an indicator, and employing methods such as... Figure 14 The following description uses an evaluation information processing device 50a and a camera 48 configured as shown, and illustrates the case where machine learning is used to evaluate the shape of the pleats. The difference between the evaluation information processing device 50a and the evaluation information processing device 50 is that the evaluation information processing device 50a stores a machine learning program 58 for evaluation in the memory 53, and stores evaluation correspondence data 57, which establishes corresponding image data and evaluation results, in the memory 53. The evaluation model 59 obtained through machine learning is also stored in the memory 53.

[0111] In the structure shown above, Figure 15 The evaluation processing flow performed by the evaluation information processing device 50a is shown. Furthermore, steps S11 to S15 and S17 are the same as the evaluation processing flow performed by the evaluation information processing device 50 described above, and are therefore omitted. After completing steps S15 or S17, the process proceeds to step S18.

[0112] In step S18, the processor 51 establishes a correspondence between the evaluation result from step S15 and its corresponding image data, and stores this correspondence as evaluation correspondence data 57 in the memory 53. Similarly, the processor 51 establishes a correspondence between the evaluation result from step S17 and its corresponding image data, and stores this correspondence as evaluation correspondence data 57 in the memory 53.

[0113] In step S19, the processor 51 determines whether the accumulated amount of evaluation correspondence data 57 exceeds a predetermined amount. Here, if the accumulated amount of evaluation correspondence data 57 is less than the predetermined amount, the process returns to step S11 and continues accumulating evaluation correspondence data 57. Furthermore, if the accumulated amount of evaluation correspondence data 57 exceeds the predetermined amount, the process proceeds to step S20.

[0114] In step S20, the processor 51 uses the image data and multiple sets of evaluation results contained in the evaluation correspondence data 57 to perform machine learning on the neural network using methods such as deep learning, thereby obtaining the evaluation model 59. Furthermore, the machine learning algorithm is not limited to deep learning; well-known algorithms such as support vector machines, Gaussian processes, decision trees, and random forests can also be used.

[0115] In step S21, the processor 51 stores the evaluation model 59 obtained in step S20 in the memory 53. Therefore, in the indexing step S12, the evaluation model 59 stored in the memory 53 is used for indexing, and the evaluation continues.

[0116] As described above, by using the evaluation model 59 obtained through machine learning to index and evaluate the pleat shape of the pleat body, an objective evaluation of the pleat shape of the pleat body can be achieved.

[0117] Furthermore, there are no particular limitations in the aforementioned machine learning. For example, common features can be extracted from the image data of pleats judged as good products using machine learning to obtain an evaluation model 59 capable of new indexing. The baseline value can also be updated to correspond to the value indexed by the evaluation model 59. In this case, the evaluation of the pleat shape can be made more objective. Additionally, any feature quantity or combination thereof described in "(2-1) Feature Quantities of Pleat Shape" can be used as the feature quantity extracted from the image data here.

[0118] (10) Information processing during the manufacturing of pleats

[0119] The following example uses a processing machine 60, an evaluation information processing device 50, and a camera 48, and is based on... Figure 16 block diagram and Figure 17 The flowchart illustrates the information processing during pleat fabrication. Furthermore, the machining machine 60, the evaluation information processing device 50, and the camera 48 are connected in a manner that allows them to communicate with each other.

[0120] The processing machine 60 includes processors such as CPU (central processing unit) and GPU (graphics processing unit) for various information processing, memory such as RAM and ROM 70, display screen 62, temperature sensor 63, humidity sensor 64, receiving unit 65, conveying unit 66, folding knife unit 67, coating unit 68, and roller unit 69.

[0121] Temperature sensor 63 is a sensor used to determine the air temperature of the space where pleating is performed.

[0122] Humidity sensor 64 is a sensor used to monitor the humidity of the space where pleating is performed.

[0123] The receiving unit 65 accepts various types of information input. Here, it accepts physical property data of the sheet-like material 30 to be processed as filter media, setting data for the vertical distance u between the upper folding plate 45 and the lower folding plate 46 corresponding to the pleat height b of the pleated body, setting data for the vertical gap s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, setting data for the vertical gap t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44, setting data for the coating amount of the spacer holding part, and setting data for the rotation speed of each roller, etc. For example, the physical property data of the sheet-like material 30 may include information indicating the stiffness of the material 30, information indicating the thickness of the material 30, information indicating the material composition of the material 30, information indicating the specified pressure loss of the material 30, information indicating the specified collection efficiency of the material 30, and information indicating the average fiber diameter of the main collection layer such as the fluoropolymer porous membrane in the material 30, etc.

[0124] The rotational speed of each roller in the conveying unit 66 is adjusted during the pleating process.

[0125] The folding blade 67 adjusts the vertical distance u between the upper folding plate 45 and the lower folding plate 46 during pleating processing, the vertical gap s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, and the vertical gap t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44, etc., based on the receiving information from the receiving part 65.

[0126] The coating unit 68 adjusts the coating amount of the interval holding part during pleating processing based on the receiving information received by the receiving unit 65.

[0127] The roller section 69 stores information indicating the degree of curl marks on various portions of the material 30 fed from the winding roller 21 during the pleating process. For example, the roller section 69 uses information about the radius of the winding roller 21 where the material 30 was originally located, determining that portions with smaller radii have larger curl marks. This determination of the degree of curl marks is performed simultaneously with the material's feed from the winding roller 21. Alternatively, the radius information of the winding roller 21 can be omitted, and the curl marks can be determined based on the relationship between the total length and thickness of the material 30.

[0128] The display screen 62 displays and outputs information during various control operations.

[0129] The processor 61 controls the driving of each part of the processing machine 60 to realize the setting data of the vertical distance u between the upper folding plate 45 and the lower folding plate 46 corresponding to the pleat height b of the pleat body, the setting data of the vertical distance s between the upper surface of the lower folding plate 46 and the lower end of the upper folding blade 43 when they are closest, the setting data of the vertical distance t between the lower surface of the upper folding plate 45 and the upper end of the lower folding blade 44 when they are closest, the setting data of the coating amount of the spacing holding part, and the setting data of the rotation speed of each roller, thereby performing pleating processing.

[0130] During this pleating process, the processor monitors in real time the physical properties of the sheet material 30 being processed, the degree of curling marks monitored by the rollers 69, the temperature information monitored by the temperature sensor 63, the humidity information monitored by the humidity sensor 64, the set data for the vertical distance u between the upper folding plate 45 and the lower folding plate 46, the set data for the vertical gap s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, the set data for the vertical gap t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44, the set data for the amount of coating on the spacing holding part, and the set data for the rotation speed of each roller, etc., and stores these manufacturing conditions in the memory 70, establishing a correspondence between these manufacturing conditions and the pleated body obtained under these manufacturing conditions. Furthermore, the determination of the correspondence is not particularly limited and can be appropriately calculated by the processor 61 based on factors such as transport speed, transport distance, and processing time.

[0131] Then, similarly as described above, the resulting pleated body is indexed and evaluated in the evaluation information processing device 50 while the image data 49 acquired and sent by the camera 48 is stored as image data 56 in the memory 53. This evaluation result is sent from the evaluation information processing device 50 to the processing machine 60, where it is correlated with the corresponding manufacturing conditions and stored in the memory 70 as manufacturing condition correspondence data 71, representing the relationship between the manufacturing conditions and the evaluation result.

[0132] Furthermore, once the manufacturing time condition correspondence data 71 has accumulated to a predetermined amount, the manufacturing time condition correspondence data, which represents the relationship between manufacturing time conditions and evaluation results, is used to perform machine learning on a neural network through methods such as deep learning, thereby obtaining a manufacturing time condition model 72 and storing it in memory 70. In addition, the machine learning algorithm is not limited to deep learning; well-known algorithms such as support vector machines, Gaussian processes, decision trees, and random forests can also be used.

[0133] After obtaining the manufacturing condition model 72, the processor 61 controls the drive of each part of the processing machine 60 to perform pleating processing by inputting the setting data of the vertical distance u between the upper folding plate 45 and the lower folding plate 46 corresponding to the pleat height b of the pleated body received by the receiving part 65, the physical property data of the sheet material 30 being processed, the crease degree information held by the roller part 69, the temperature information held by the temperature sensor 63, and the humidity information held by the humidity sensor 64 into the manufacturing condition model 72. Here, as processing machine control quantities, examples include the vertical distance s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, the vertical distance t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44, the coating amount of the spacing holding part, and the rotation speed of each roller.

[0134] according to Figure 15 The flowchart above will be used to explain the above processing again.

[0135] In step S31, the processor 61 of the processing machine 60 receives, via the receiving unit 65, physical property data of the sheet material 30, setting data for the vertical distance u between the upper folding plate 45 and the lower folding plate 46 corresponding to the pleat height b of the pleat body, setting data for the vertical gap s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, setting data for the vertical gap t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44, setting data for the coating amount of the spacing holding part, setting data for the rotation speed of each roller, and physical property data of the sheet material 30 being processed. Furthermore, the processor 61 receives information on the degree of creasing controlled by the roller unit 69, temperature information controlled by the temperature sensor 63, and humidity information controlled by the humidity sensor 64. Thus, the processor 61 determines the manufacturing conditions corresponding to the setting data. Additionally, the setting data is not particularly limited and can be conditions determined empirically to obtain the desired pleat shape.

[0136] In step S32, the processor 61 performs pleating processing according to the manufacturing conditions known in step S32, and obtains a pleated body.

[0137] In step S33, the processor 51 of the evaluation information processing device 50 evaluates the pleat shape of the pleated body manufactured under the known manufacturing conditions in step S32.

[0138] In step S34, the processor 61 of the processing machine 60 establishes a correspondence between the evaluation result from step S33 sent from the evaluation information processing device 50 and the corresponding manufacturing conditions, and stores it in the memory 70 as manufacturing conditions correspondence data 71.

[0139] In step S35, the processor 61 determines whether the amount of data stored as manufacturing condition correspondence data 71 has accumulated to a predetermined amount. If the amount of data has been sufficiently accumulated, the process proceeds to step S38. If the amount of data is insufficient, the process returns to step S31 to continue acquiring manufacturing and evaluation relationship data.

[0140] In step S36, the processor 61 uses multiple sets of data containing manufacturing time conditions and evaluation results in the manufacturing time condition correspondence data 71 to perform machine learning on the neural network through deep learning and other methods, thereby obtaining the manufacturing time condition model 72.

[0141] In step S37, the processor 61 stores the manufacturing time condition model 72 obtained in step S36 into the memory 70.

[0142] In step S38, the processor 61 acquires manufacturing conditions other than the machine control quantities. Specifically, it acquires the vertical distance u between the upper folding plate 45 and the lower folding plate 46, the physical property data of the sheet material 30 being processed, the information on the degree of creasing acquired by the roller 69, the temperature information acquired by the temperature sensor 63, and the humidity information acquired by the humidity sensor 64.

[0143] In step S39, the processor 61 inputs the manufacturing time conditions (excluding the machine control quantity) obtained in step S38 into the manufacturing time condition model 72 obtained in step S36, thereby mastering the manufacturing time conditions. Then, pleating is performed according to the obtained manufacturing time conditions to obtain a pleated body. In addition, the manufacturing time conditions determined here are output to the display screen 62 for display.

[0144] In step S40, the processor 51 of the evaluation information processing device 50 evaluates the pleat shape of the pleat body manufactured in step S39.

[0145] In step S41, the processor 61 of the processing machine 60 establishes a correspondence between the evaluation result from step 40 sent from the evaluation information processing device 50 and the corresponding manufacturing conditions, and stores it in the memory 70 as manufacturing conditions correspondence data 71.

[0146] In step S42, the processor 61 determines whether the amount of data stored as manufacturing time condition correspondence data 71 has been further accumulated to a predetermined amount. If the amount of data has been sufficiently accumulated, the process proceeds to step S36, and the manufacturing time condition model 72 is updated by performing machine learning. If the amount of data is insufficient, the process returns to step S38 to continue acquiring manufacturing time condition correspondence data 71 for manufacturing and evaluation.

[0147] As described above, by using manufacturing conditions determined by the manufacturing condition model to manufacture pleats, it is possible to stably manufacture pleats with a shape that can be objectively judged as good while suppressing deviations.

[0148] Furthermore, by obtaining the manufacturing condition model, even if the manufacturing conditions other than the machine control quantities change, such as the vertical distance u between the upper folding plate 45 and the lower folding plate 46, the physical property data of the sheet material 30 being processed, the information on the degree of creasing held by the roller 69, the temperature information held by the temperature sensor 63, and the humidity information held by the humidity sensor 64, deviations can be suppressed while the pleated body with a pleated shape that can be objectively judged as good can be stably manufactured.

[0149] Furthermore, taking the set data of the vertical gap s when the upper surface of the lower folding plate 46 is closest to the lower end of the upper folding blade 43, and the set data of the vertical gap t when the lower surface of the upper folding plate 45 is closest to the upper end of the lower folding blade 44 as examples, the change in the pleat shape of the pleated body obtained by changing these set data was confirmed.

[0150] Here, the gaps s and t in the manufacturing conditions of a pleated body are set to the same value, and the manufacturing conditions are changed in a way that makes the gaps s and t satisfy the relationship (a)<(b)<(c)<(d)<(e), thus obtaining Figure 18 The pleats shown are (a), (b), (c), (d), and (e).

[0151] from Figure 18 It can be clearly confirmed that the gaps s and t are correlated with the pleat shape of the pleat body.

[0152] The embodiments of this disclosure have been described above. However, it should be understood that various changes in form and detail can be made without departing from the spirit and scope of this disclosure as set forth in the claims.

[0153] Symbol Explanation

[0154] 20 pleated body;

[0155] 26, 27, 28, 29; (This likely refers to a specific type of section or unit, possibly related to spacing or maintenance.)

[0156] 30 sheet-like materials;

[0157] 35. Connection part;

[0158] 56 image data;

[0159] 59 Evaluation Model;

[0160] 62 Display screens (output devices);

[0161] a. Pleat spacing;

[0162] b. Pleat height;

[0163] c represents the length of the connecting portion in the second direction.

[0164] Existing technical documents

[0165] Patent documents

[0166] Patent Document 1: Japanese Patent Application Publication No. 2016-026870

Claims

1. An evaluation method, wherein the evaluation method is a method for evaluating the pleat shape of a pleated body (20) obtained by processing a sheet material (30) into a pleated shape, characterized in that, include: The extraction process involves extracting feature quantities related to the shape of the pleats from the image data (56) of the pleats (20); as well as The evaluation process assesses the pleat shape based on the feature quantities extracted in the extraction process. The pleats are filter material. The pleats are formed by repeatedly folding the sheet material through peak and valley folds, resulting in a pleated shape with multiple peak and valley fold lines arranged in parallel. The image data includes image data obtained by taking pictures of the pleat shape from the directions extending from the peak fold line and the valley fold line.

2. The evaluation method according to claim 1, characterized in that, The evaluation process includes a step of comparing the feature quantity extracted in the extraction process with a predetermined benchmark.

3. The evaluation method according to claim 1 or 2, characterized in that, The characteristic quantity is any one of the following characteristic quantities and combinations thereof: Pleat spacing (a); Pleat height (b); When the direction in which the fold line extends is set as the first direction, the direction in which the pleat shape continues is set as the second direction, and the direction perpendicular to both the first and second directions is set as the third direction, the length (c) of the connecting portion (35) that connects adjacent peaks and valleys when viewed along the first direction is in the second direction. The degree of bending of the connecting portion when viewed along the first direction; The symmetry of the adjacent connecting portions when viewed along the first direction; as well as The length and width of the interval holding portion when the pleat body is provided with an interval holding portion (26, 27, 28, 29) for maintaining the interval of the pleats.

4. The evaluation method according to claim 1, characterized in that, Also includes: In the evaluation model learning process, the image data of multiple pleats and the corresponding evaluation completion data set established with the image data of the pleats are used to perform machine learning on the relationship between the image data of the pleats and the evaluation completion data, thereby creating an evaluation model (59). In the evaluation process, the feature quantities extracted in the extraction process are evaluated based on the evaluation model created in the evaluation model learning process.

5. The evaluation method according to claim 1, characterized in that, Also includes: The input condition learning process involves using multiple sets of input conditions and evaluation result data to learn the relationship between the input conditions and the evaluation result data. The input conditions include processing condition data input to a processing machine that processes the sheet material into a pleated shape, as well as physical property data related to the sheet material. The evaluation result data is obtained by evaluating the pleated body corresponding to the input conditions in the evaluation process.

6. The evaluation method according to claim 5, characterized in that, The processing condition data includes any one of the following data and combinations thereof: the gap between the folding blade and the folding plate of the processing machine; the transport speed of the processing machine when transporting the material; and the amount of resin applied to the material to form the spacer.

7. The evaluation method according to claim 5, characterized in that, The input conditions are output to the output device (62) based on the model obtained in the input condition learning process.

8. A manufacturing method, characterized in that, The pleats are manufactured using the input conditions output from the evaluation method of claim 7.

Images