Inspection device, blister packaging machine, and method for manufacturing blister packaging

Abstract

Provided are an inspection device, a blister packaging machine, and a method for manufacturing blister packaging that can detect molding defects in the side of a bag portion with higher accuracy. The bag portion inspection device (21) includes: an illumination device (50) that can irradiate a container film (3) on which a bag portion (2) has been formed with a predetermined electromagnetic wave; and a camera (51) for capturing the electromagnetic wave at the bottom of the bag portion (2) that has passed through the container film (3). Based on the image data thus obtained, shading pattern data corresponding to the shading pattern generated in the bottom of the bag portion (2) is extracted. Reconstructed shading pattern data is obtained, and the reconstructed shading pattern data is reconstructed by inputting the extracted shading pattern data into an AI model. The AI ​​model is generated by causing a neural network to learn only shading pattern data related to a bag portion (2) without molding defects as learning data. Next, the shading pattern data and the reconstructed shading pattern data are compared to perform a good / bad judgment related to the molding state of at least the side of the bag portion (2).

Description

Technical Field

[0001] The present invention relates to an inspection device for inspecting the forming state of a bag portion of a blister package, a blister packaging machine, and a method for manufacturing the blister package. Background Art

[0002] In the past, blister packs have been widely used as packaging containers for packaging pharmaceuticals, foods, electronic devices, etc. Among them, in the field of pharmaceuticals, PTP (blister pack) sheets used for packaging tablets, capsules, etc. are well known.

[0003] The PTP sheet is composed of a container film and a cover film. The container film is formed with a bag portion for receiving contents such as tablets. The cover film is installed on the container film in a manner that seals the opening side of the bag portion. The bag portion is pressed from the outside, and the cover film constituting the lid is pierced by the contents received therein, thereby allowing the contents to be taken out.

[0004] The PTP sheet is manufactured through a bag forming process of forming a bag portion relative to a strip-shaped container film, a filling process of filling the bag portion with contents, an installation process of installing a cover film on the container film in a manner to seal the opening side of the bag portion, and a cutting process of cutting the PTP sheet constituting the final product from the strip-shaped PTP film on which the two strip-shaped films are installed.

[0005] Here, the bag portion is generally formed by vacuum forming, pressure forming, plunger forming, plunger-assisted pressure forming, etc., and a portion (the portion to be formed) of a partially heated and softened strip of container film is stretched.

[0006] Therefore, the thicknesses of the bottom and side portions of the bag portion are correlated. If the bottom is thick, the side portions are thin, and if the bottom is thin, the side portions are thick.

[0007] When the thickness balance between the bottom and side portions is lost, a portion of the bag portion becomes excessively thinner, resulting in various problems such as reduced gas barrier properties. In particular, there is a concern about excessive thinning of the side portions, which are thinner than the bottom.

[0008] On the other hand, a technique has been proposed for detecting forming defects in the side portions of a bag portion based on image data obtained by imaging the bottom portion of the bag portion using the aforementioned correlation (for example, see Patent Document 1).

[0009] Prior art literature

[0010] Patent Literature

[0011] Patent Document 1: Japanese Patent No. 6368408 Summary of the Invention

[0012] Problems to be solved by the invention

[0013] However, in the prior art of patent document 1, the following structure is formed, wherein, based on the image data obtained by photographing the bottom of the bag portion, the wall thickness at each position on the bottom is calculated according to the relationship between the transmittance of light and the wall thickness of the bottom, and based on the average value thereof (the average wall thickness of the bottom), the poor forming of the side of the bag portion is detected.

[0014] In this way, the forming state of the side of the bag can be roughly estimated based on the wall thickness of the bottom. However, even if the average, maximum and minimum values ​​of the wall thickness of the bottom are the desired values ​​and the forming state of the bottom is judged to be appropriate, when there is a deviation in the wall thickness distribution of the bottom or when the shape of the bottom is complex, there is still a case where the side does not have the desired wall thickness or the wall thickness distribution of the side deviates.

[0015] Therefore, in the above-mentioned conventional technology, there is a problem that a forming defect (wall thickness defect) of the side portion of the bag portion cannot be detected with good accuracy.

[0016] The above-mentioned problems are not limited to PTP packaging, but are also inherent in other blister packaging fields.

[0017] The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an inspection device, a blister packaging machine, and a blister packaging manufacturing method that can detect forming defects on the side portions of a bag with higher accuracy.

[0018] Technical solutions to problems

[0019] Hereinafter, each technical solution suitable for solving the above-mentioned problems will be described in detail. In addition, the corresponding technical solution will be followed by a unique effect as needed.

[0020] Technical solution 1 relates to an inspection device for inspecting the forming state of a bag portion of a blister package, characterized in that the inspection device comprises:

[0021] an irradiation mechanism capable of irradiating a container film on which the bag portion is formed with predetermined electromagnetic waves;

[0022] a camera mechanism, which is disposed on the side opposite to the irradiation mechanism via the container film and is capable of capturing the electromagnetic waves that have passed through at least the bottom of the bag portion to obtain image data;

[0023] a shading pattern extraction mechanism that extracts shading pattern data corresponding to a shading pattern (shading distribution image) produced on the bottom of the bag portion by irradiation with the electromagnetic wave, based on the image data obtained by the imaging mechanism;

[0024] an identification unit (generation model) that generates a learning model by learning only the shading pattern data related to the pocket portion without defective forming for a neural network, the neural network comprising an encoding unit (encoder) that extracts a feature value from the input shading pattern data and a decoding unit (decoder) that reconstructs the shading pattern data based on the feature value;

[0025] a reconstructed data acquisition mechanism that can acquire reconstructed shading pattern data, the reconstructed shading pattern data being shading pattern data reconstructed by inputting the shading pattern data extracted by the shading pattern extraction mechanism into the recognition mechanism;

[0026] a comparing unit that compares the shading pattern data extracted by the shading pattern extracting unit with the reconstructed shading pattern data acquired by the reconstructed data acquiring unit; and

[0027] A quality judgment mechanism can judge the quality of at least the forming state of the side portion of the bag portion based on the comparison result of the comparison mechanism.

[0028] In the following methods, the same applies, but the "neural network" mentioned above includes, for example, a convolutional neural network having multiple convolutional layers. Furthermore, the "learning" mentioned above includes, for example, deep learning. The "recognition mechanism (generative model)" mentioned above includes, for example, an autoencoder (autoencoder) and a convolutional autoencoder (convolutional autoencoder).

[0029] In addition, the above-mentioned "blister packaging" includes, for example, PTP sheets for storing tablets, quantitative packaging for storing food, carrier tapes for storing electronic devices, etc., and the above-mentioned "electromagnetic waves" include, for example, visible light, ultraviolet light, X-rays, etc.

[0030] In addition, the above-mentioned "light and dark pattern (light and dark distribution image) produced at the bottom of the above-mentioned bag portion by irradiation of the above-mentioned electromagnetic waves" refers to the light and dark two-dimensional distribution image produced at the bottom of the bag portion based on the relationship between the difference in wall thickness (wall thickness distribution) at each position (two-dimensional coordinate position) on the bottom of the bag portion and the transmittance of the electromagnetic waves passing therethrough.

[0031] In other words, the "shade" here refers to the intensity (brightness) of the electromagnetic waves transmitted through various locations on the bottom of the bag. Therefore, the expression "shade pattern (shade distribution image) produced on the bottom of the bag by irradiation with the electromagnetic waves" can be replaced with expressions such as "intensity distribution image of the electromagnetic waves transmitted through the bottom of the bag," "two-dimensional distribution image of electromagnetic wave intensity (brightness) varying at each location due to differences in wall thickness at each location on the bottom of the bag," and "shade distribution image (electromagnetic wave intensity distribution image, brightness distribution image) corresponding to the wall thickness distribution of the bottom of the bag."

[0032] As described in the above "Background Art", the wall thicknesses of the bottom and side of the bag formed by partially stretching the container film are correlated. If the bottom is thick, the side is thin, and if the bottom is thin, the side is thick.

[0033] By utilizing such a correlation, in the above-mentioned technical solution 1, the following solution is formed, in which the light and dark pattern produced at the bottom of the bag (i.e., the wall thickness distribution state of the bottom) is extracted based on the image data obtained by irradiating the specified electromagnetic waves while photographing the bottom of the bag, and based on this, a good or bad judgment is made at least related to the forming state of the side of the bag.

[0034] According to this configuration, it is possible to detect with higher accuracy a forming defect (wall thickness defect) in the side portion of the bag portion, such as the presence or absence of a deviation in the wall thickness distribution of the side portion of the bag portion.

[0035] In particular, in the present invention, a recognition mechanism (generative model) such as an autoencoder constructed by learning a neural network is used to perform a structure for inspecting the forming state of the side portion of the bag.

[0036] Specifically, a structure is formed in which shading pattern data obtained by imaging the bottom of the bag portion constituting the inspection object is compared with reconstructed shading pattern data obtained by reconstructing the shading pattern data by a recognition mechanism to perform a quality judgment.

[0037] This makes it possible to inspect even tiny molding defects, bottom shapes, and pockets with complex wall thickness distributions, which were previously difficult to detect. As a result, inspection accuracy can be further improved.

[0038] In addition, in the present technical solution, in the two light and dark pattern data being compared, there is no influence of differences in the imaging conditions on the container film side (for example, the configuration position, configuration angle, deflection, etc. of the container film) as the inspection object and the imaging conditions on the inspection device side (for example, lighting conditions, camera viewing angle, etc.), so that more subtle forming defects can be detected more accurately.

[0039] Another option is to directly image the sides of the bag to inspect their forming state. In this case, considering factors such as gas barrier properties, it is necessary to understand the forming state of the entire circumference of the side. However, this approach requires a considerable amount of time and large-scale equipment to understand the forming state of the entire circumference of the side, thus raising concerns about reduced blister packaging productivity.

[0040] In this regard, according to the present technical solution, since a solution is formed that can grasp the forming state of the bottom of the bag by photographing it, and grasp the forming state of the entire side circumference in a short period of time and simply, it is possible to speed up the inspection and thus improve the productivity of the blister packaging.

[0041] Technical solution 2 relates to the inspection device of technical solution 1, characterized in that the above-mentioned container film is composed of a resin film material with light transmittance, and the above-mentioned irradiation mechanism is constructed in a manner that can irradiate ultraviolet light as the above-mentioned electromagnetic wave (for example, ultraviolet light having a peak wavelength in the range of 200nm to 280nm, etc.).

[0042] When the container film is made of a translucent resin film material and a structure is formed to irradiate visible light from an irradiation mechanism, it is difficult to create a difference in light transmittance between the thin-walled portion and the thick-walled portion of the bottom portion of the bag portion. In other words, the entire bottom portion is uniform, making it difficult to create a shading pattern. As a result, there is a concern that proper inspection may be difficult.

[0043] On the other hand, according to the above-mentioned technical solution 2, a structure is adopted in which the container film made of a light-transmitting resin film material is irradiated with ultraviolet light.

[0044] Since ultraviolet light has a lower transmittance than visible light and is less likely to pass through a translucent container film, the molding state of the bag portion can be inspected more appropriately.

[0045] In addition, here, the "resin film material with light transmittance" includes, for example, "transparent resin film material", "semi-transparent resin film material", etc., and the "transparent resin film material" is a film with the property of allowing light to pass (light transmittance), and the transmittance of electromagnetic waves (light) is extremely high, and the opposite side can be seen through the film; "semi-transparent resin film material" is "light transmittance, but due to the diffusion of the electromagnetic waves (light) transmitted or the low transmittance of the electromagnetic waves (light), the human eye cannot clearly identify or cannot identify at all the shape of the object on the opposite side through the film."

[0046] Furthermore, the terms "transparent" and "translucent" refer to the material of the film having light transmittance and have nothing to do with the presence or absence of color. Therefore, "transparent" or "translucent" films naturally include, for example, "colorless transparent" or "colorless translucent" films, as well as "colored transparent" or "colored translucent" films.

[0047] Technical solution 3 relates to the inspection device described in technical solution 1 or 2, characterized in that the above-mentioned electromagnetic waves include electromagnetic waves with a wavelength whose transmittance of the above-mentioned container film (such as resin film materials such as polypropylene, polyvinyl chloride, etc.) is in the range of 15% to 60%.

[0048] If the transmittance of electromagnetic waves through the container film is too high or too low, a difference in light transmittance between the thin-walled portion and the thick-walled portion of the bottom of the bag portion may occur, making proper inspection difficult.

[0049] In contrast, as in the aforementioned technical solution 3, inspection can be performed more appropriately by using electromagnetic waves having a wavelength in which the transmittance of the container film is within the range of 15% to 60%. More preferably, inspection can be performed by using electromagnetic waves having a wavelength in which the transmittance of the container film is within the range of 20% to 50% (e.g., 30%).

[0050] Technical solution 4 is the inspection device according to any one of technical solutions 1 to 3, characterized in that the bag portion is thermoformed on the flat container film.

[0051] Here, "thermoforming" refers to a forming method in which a portion of a flat container film (the portion to be formed) is partially heated and softened to perform stretching, and includes, for example, vacuum forming, pressure forming, plunger forming, and plunger-assisted pressure forming.

[0052] Therefore, based on the structure of the present technical solution 4, the effects of the above-mentioned technical solution 1 and the like are more effective.

[0053] Technical solution 5 relates to a blister packaging machine, characterized in that it includes the inspection device according to any one of technical solutions 1 to 4.

[0054] By installing the inspection device in a blister packaging machine (e.g., a PTP packaging machine), as in the fifth embodiment, it is possible to effectively eliminate defective products during the manufacturing process of blister packages (e.g., PTP sheets). Furthermore, the blister packaging machine may include a discharge mechanism that discharges blister packages determined to be defective by the inspection device.

[0055] As a more specific structure of the blister packaging machine, the following structures can be mentioned.

[0056] A blister packaging machine for producing blister packages, wherein a predetermined content is contained in a bag formed with a container film, and a cover film is attached to seal the bag, characterized in that the blister packaging machine comprises:

[0057] a bag forming mechanism for forming the bag on the container film conveyed in a belt shape;

[0058] a filling mechanism for filling the bag portion with the contents;

[0059] An installation mechanism for installing the strip-shaped cover film on the container film filled with the contents in the bag portion so as to close the bag portion;

[0060] a cutting mechanism (including a punching mechanism that punches out in units of sheets) that cuts the blister pack from a strip-shaped body (strip-shaped blister film) in which the cover film is attached to the container film; and

[0061] The inspection device according to any one of technical solutions 1 to 4.

[0062] In addition, in the case of assuming that a container film with an uncertain posture is to be inspected, it is of course necessary to process the position of a specific bag portion. In the case of a non-circular bag portion, the following processing must be performed, namely, the center position of the bag portion constituting the inspection object is calculated based on the image data, and on the basis of making the center of a pre-stored reference image for pattern matching consistent with the center position of the bag portion, the reference image is rotated by a specified angle each time, and each time it is determined whether the two are consistent. There is a concern that the number of processing steps involved in the inspection of the bag portion is very large and troublesome.

[0063] In contrast, by installing the inspection device in the blister packaging machine as in the aforementioned technical solution 5, the container film's stopping position and orientation (posture) relative to the imaging mechanism remain constant. Therefore, during inspection, there is no need for alignment or orientation adjustments of the inspection object, thereby enabling faster inspection. As a result, the number of bags processed per bag is significantly reduced, significantly increasing the inspection processing speed.

[0064] In addition, the following structure may be formed, wherein, based on the structure of the above-mentioned technical solution 5, the above-mentioned filling mechanism is configured at a position downstream of the above-mentioned inspection device;

[0065] The filling control mechanism includes a filling control mechanism that controls the operation of the filling mechanism based on the inspection result of the inspection device, and can switch whether the filling of the content into the bag portion can be carried out.

[0066] This solution also allows, for example, poorly formed pockets to be left unfilled. This prevents the undesirable situation of discarding the blister pack along with the contents when the pack is discarded due to poorly formed pockets. Furthermore, the cumbersome process of removing the temporarily filled pockets for reuse is eliminated. Consequently, a reduction in productivity can be minimized.

[0067] In addition, the following structure can also be formed, in which, based on the solution of the above-mentioned technical solution 5, "the above-mentioned bag forming mechanism includes a first mold, a second mold opposite to the first mold via the above-mentioned container film, and a stretching mechanism (stretching forming mechanism) for forming the above-mentioned bag relative to the above-mentioned container film clamped by the two molds."

[0068] In this structure, since the correlation between the wall thickness of the bottom and the side of the bag described in the above-mentioned "background technology" is generated, that is, if the wall thickness of the bottom is thicker, the wall thickness of the side is thinner, and if the wall thickness of the bottom is thinner, the wall thickness of the side is thicker, the effect of the above-mentioned technical solution 1, etc. for detecting poor forming of the side of the bag based on the image data obtained by photographing the bottom of the bag is more effective.

[0069] Technical solution 6 relates to a method for manufacturing a blister package, wherein a predetermined content is contained in a pocket formed in a container film, and a cover film is attached to seal the pocket, characterized in that the method comprises:

[0070] a bag forming step in which the bag is formed relative to the container film being conveyed in a belt shape;

[0071] a filling step of filling the bag portion with the contents;

[0072] An installation step, in which the strip-shaped cover film is installed on the container film filled with the contents in the bag portion so as to close the bag portion;

[0073] a cutting step (including a punching step of punching out in units of sheets) of cutting the blister pack from a strip-shaped body having the cover film attached to the container film; and

[0074] an inspection step of inspecting a formed state of the bag portion of the blister package;

[0075] The invention is characterized in that: the above inspection process includes:

[0076] an irradiation step of irradiating the container film having the bag portion formed thereon with predetermined electromagnetic waves;

[0077] an imaging step of imaging the electromagnetic wave that has passed through at least the bottom of the bag portion to obtain image data;

[0078] a shading pattern extraction step of extracting shading pattern data corresponding to a shading pattern (shading distribution image) produced at the bottom of the bag portion by irradiation with the electromagnetic wave, based on the image data obtained in the imaging step;

[0079] a reconstruction data acquisition step in which the shading pattern data extracted in the shading pattern extraction step is input to a recognition mechanism (generation model) to obtain reconstructed shading pattern data as reconstructed shading pattern data, wherein the recognition mechanism is generated by learning only the shading pattern data related to the pocket portion without forming defects as learning data for a neural network having an encoding unit (encoder) that extracts feature quantities from the input shading pattern data and a decoding unit (decoder) that reconstructs shading pattern data from the feature quantities;

[0080] a comparing step of comparing the shading pattern data extracted in the shading pattern extracting step with the reconstructed shading pattern data acquired in the reconstructed data acquiring step; and

[0081] A quality determination step is performed in which a quality determination is performed on a molding state of at least a side portion of the bag portion based on the comparison result in the comparison step.

[0082] According to the above technical solution 6, the same effect as the above technical solutions 1 and 5 is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a stereogram of the PTP sheet;

[0084] Figure 2 It is a partial enlarged cross-sectional view of the PTP sheet;

[0085] Figure 3 is a stereogram of the PTP film;

[0086] Figure 4 This is an overview diagram of the bag inspection device;

[0087] Figure 5 This is the general structure diagram of the PTP packaging machine;

[0088] Figure 6 is a block diagram showing the functional structure of the bag inspection device;

[0089] Figure 7A partially cutaway front view showing the general structure of the bag forming device and the heating device;

[0090] Figure 8 A schematic diagram illustrating the structure of a neural network;

[0091] Figure 9 is a flow chart showing the process of the bag forming process;

[0092] Figure 10 A flowchart showing the flow of a learning process of a neural network;

[0093] Figure 11 A flowchart showing the process of bag inspection processing;

[0094] Figure 12 This figure shows the dark and light patterns produced in a bag portion without forming defects.

[0095] Figure 13 To indicate along Figure 12 A curve diagram of the brightness values ​​of each pixel on the A-A line in FIG;

[0096] Figure 14 A diagram showing light and dark patterns etc. produced in a bag portion having a poorly formed shape;

[0097] Figure 15 To indicate along Figure 14 A curve diagram of the brightness values ​​of each pixel on the B-B line in FIG;

[0098] Figure 16 (a) is a diagram schematically showing reconstructed shading pattern data obtained by reconstructing shading pattern data. Figure 16 (b) is a diagram schematically showing a location where a defective pixel is generated during the comparison process;

[0099] Figure 17 FIG. 2 is a diagram for explaining a blister package according to another embodiment of the present invention. Figure 17 (a) is its stereogram, Figure 17 (b) is a top view thereof. DETAILED DESCRIPTION

[0100] One embodiment will be described below with reference to the drawings. First, a PTP sheet 1 as a blister package will be described.

[0101] picture Figure 1 、 Figure 2 As shown, the PTP sheet 1 includes a container film 3 and a cover film 4 . The container film 3 has a plurality of pockets 2 . The cover film 4 is mounted on the container film 3 to close the pockets 2 .

[0102] The container film 3 is formed of a colorless, transparent thermoplastic resin material such as PP (polypropylene) or PVC (polyvinyl chloride), and is light-transmissive. Meanwhile, the cover film 4 is formed of an opaque material (e.g., aluminum foil) with a sealant such as polypropylene resin provided on its surface.

[0103] The PTP sheet 1 is formed into a generally rectangular shape when viewed from above. In the PTP sheet 1, a row of five pockets 2 arranged along its longitudinal direction is formed in two rows along its transverse direction. In other words, a total of ten pockets 2 are formed. Each pocket 2 contains a single tablet 5 as its content.

[0104] The bag portion 2 is composed of a bottom 2a and a side portion 2b. The bottom 2a is arranged opposite to the cover film 4 and is roughly circular when viewed from above. The side portion 2b is connected to the periphery of the bottom 2a and connects the bottom 2a to the flat portion of the film (non-formed portion of the bag) 3b.

[0105] In this embodiment, the bottom portion 2a is formed into a generally arc-shaped cross section that is gently curved, but the present invention is not limited thereto and may be formed into a flat configuration. Alternatively, the bottom portion 2a may be formed into an arc-shaped cross section with a greater curvature so that the corner 2c where the bottom portion 2a intersects with the side portion 2b is not distinct.

[0106] PTP chip 1 (refer to Figure 1 ) by forming a strip-shaped PTP film 6 (refer to Figure 3 ) is punched into a rectangular sheet.

[0107] Next, refer to Figure 5 , a schematic structure of a PTP packaging machine 11 as a blister packaging machine for manufacturing the above-mentioned PTP sheet 1 will be described.

[0108] At the most upstream side of the PTP packaging machine 11, a strip of container film 3 is wound into a roll. The unwinding end of the rolled container film 3 is guided by a guide roller 13. Downstream of the guide roller 13, the container film 3 is hung on an intermittent conveying roller 14. The intermittent conveying roller 14 is connected to an intermittently rotating motor to intermittently convey the container film 3.

[0109] Between the guide roller 13 and the intermittent transport roller 14, along the transport path of the container film 3, a heating device 15 and a bag forming device 16 are sequentially arranged. The heating device 15 and the bag forming device 16 constitute the bag forming mechanism of this embodiment. The structures of the heating device 15 and the bag forming device 16 will be described in detail later.

[0110] Here, after the container film 3 is heated by the heating device 15 and becomes relatively soft, a plurality of bags 2 are formed at a time at predetermined positions on the container film 3 by the bag forming device 16 (bag forming step). The bags 2 are formed during the intervals between the transport operations of the container film 3 by the intermittent transport rollers 14.

[0111] Furthermore, a bag inspection device 21 is provided between the guide roller 13 and the intermittent conveying roller 14 , downstream of the bag forming device 16 .

[0112] The bag inspection device 21 is used to inspect (inspection process) the formed state of the bag 2 formed by the bag forming device 16. The structure of the bag inspection device 21 will be described in detail later.

[0113] The container film 3 fed from the intermittent feed roller 14 is hung on the tension roller 18 , the guide roller 19 , and the film support roller 20 in this order.

[0114] Since the film support roller 20 is connected to a motor that rotates at a constant speed, the container film 3 is continuously conveyed at a constant speed. The tension roller 18 is in a state of stretching the container film 3 toward the side where it is stretched by elastic force, thereby preventing the container film 3 from loosening due to the difference in conveying motion between the intermittent conveying roller 14 and the film support roller 20, and constantly keeping the container film 3 in a stretched state.

[0115] A tablet filling device 22 is provided between the guide roller 19 and the film supporting roller 20 along the conveyance path of the container film 3 .

[0116] The tablet filling device 22 functions as a filling mechanism that automatically fills the bag sections 2 with tablets 5. The tablet filling device 22 opens its shutter at predetermined intervals in sync with the transport of the container film 3 by the film support roller 20, allowing the tablets 5 to fall. As the shutter opens, the tablets 5 are filled into each bag section 2 (filling process). The operation of the tablet filling device 22 is controlled by the filling control device 82, described below.

[0117] Meanwhile, a strip of cover film 4 is wound up in a roll at the upstream end. The drawn-out end of the roll of cover film 4 is guided by a guide roller 24 toward a heating roller 25. The heating roller 25 is press-fitted against the film support roller 20, and the container film 3 and cover film 4 are fed between the two rollers 20 and 25.

[0118] Furthermore, the container film 3 and the cover film 4 are passed between two rollers 20 and 25 in a heated and pressurized state. This causes the cover film 4 to adhere to the container film 3, and the bag portions 2 are sealed by the cover film 4 (installation process). This produces the PTP film 6 as a strip, with tablets 5 received in each bag portion 2. The film support roller 20 and the heated roller 25 constitute the installation mechanism in this embodiment.

[0119] The PTP film 6 fed from the film support roll 20 is sequentially wound around the tension roll 27 and the intermittent feed roll 28 .

[0120] Since the intermittent transport roller 28 is connected to an intermittently rotating motor, it intermittently transports the PTP film 6. The tension roller 27 is in a state of stretching the PTP film 6 toward the side where it is stretched by elastic force, preventing the PTP film 6 from being loosened due to the difference in transporting motion between the film support roller 20 and the intermittent transport roller 28, and always keeping the PTP film 6 in a tensioned state.

[0121] The PTP film 6 fed from the intermittent feed roller 28 is sequentially wound around the tension roller 31 and the intermittent feed roller 32 .

[0122] Since the intermittent transport roller 32 is connected to an intermittently rotating motor, the PTP film 6 is intermittently transported. The tension roller 31 is in a state of stretching the PTP film 6 toward the side stretched by elastic force, preventing the PTP film 6 from being slackened between the intermittent transport rollers 28 and 32.

[0123] Between the intermittent transport roller 28 and the tension roller 31, a slit forming device 33 and a marking device 34 are sequentially arranged along the transport path of the PTP film 6. The slit forming device 33 has the function of forming a slit for slitting at a predetermined position on the PTP film 6. The marking device 34 has the function of adding a mark to a predetermined position (e.g., the label portion) on the PTP film 6.

[0124] The PTP film 6 fed out from the intermittent feed roller 32 is wound around a tension roller 35 and a continuous feed roller 36 in this order on the downstream side.

[0125] A sheet punching device 37 is provided between the intermittent feed roller 32 and the tension roller 35 along the feed path of the PTP film 6. The sheet punching device 37 functions as a sheet punching mechanism (slitting mechanism) for punching the outer edge of the PTP film 6 per PTP sheet.

[0126] The PTP sheets 1 punched out by the sheet punching device 37 are transported by the take-out conveyor 38 and temporarily stored in the finished product hopper 39 (cutting process). However, if a defective product signal is input from the filling control device 82 described later to the defective product discharge mechanism 40 that can selectively discharge the PTP sheets 1, the defective PTP sheets 1 are separately discharged by the defective product discharge mechanism 40 and transferred to a defective product hopper (not shown).

[0127] A cutting device 41 is provided downstream of the continuous feed roller 36. The strip-shaped waste 42 remaining after punching by the sheet punching device 37 is guided by the tension roller 35 and the continuous feed roller 36 before being directed to the cutting device 41. The continuous feed roller 36 is pressed against a driven roller, which holds the waste 42 while conveying it.

[0128] The cutting device 41 has a function of cutting the waste part 42 into a predetermined size. The cut waste part 42 is stored in the waste hopper 43 and then disposed of separately.

[0129] Furthermore, the rollers 14, 19, 20, 28, 31, 32, etc. are positioned so that their surfaces face the bag 2. However, since recesses are formed on the surfaces of the rollers 14, etc. to receive the bag 2, the bag 2 is substantially not crushed. Furthermore, since the bag 2 is conveyed while being received in the recesses of the rollers 14, etc., both intermittent and continuous conveyance can be reliably performed.

[0130] Next, refer to Figure 7 , the structures of the heating device 15 and the bag forming device 16 are explained.

[0131] The heating device 15 includes an upper heating plate 15a and a lower heating plate 15b. The two heating plates 15a and 15b are configured to be heated by a heater (not shown). The two heating plates 15a and 15b are positioned to sandwich the transport path of the container film 3 and are movable toward and away from the container film 3.

[0132] Furthermore, each of the heating plates 15 a and 15 b includes a plurality of protrusions 15 c and 15 d at a position corresponding to the planned forming portion 3 a of the bag portion 2 in the container film 3 .

[0133] During a temporary stop, the intermittently conveyed container film 3 is sandwiched between the protrusions 15c and 15d as the two heating plates 15a and 15b approach each other, thereby being locally (point-wise) heated, softening the heated portion. In this embodiment, the contact areas of the protrusions 15c and 15d with the container film 3 are slightly smaller than the planar shape of the bag portion 2.

[0134] The bag forming device 16 includes a lower mold 61 as a second mold and an upper mold 71 as a first mold. The lower mold 61 is fixed to a fixed support 63 via a cylindrical lower mold cavity 62. The lower mold 61 also includes a plurality of insertion holes 64 at positions corresponding to the positions of the bag 2.

[0135] A plurality of through holes are formed in the support base 63, and a rod-shaped slider 65 is inserted into the through hole via a bearing mechanism. The slider 65 is movable up and down by a cam mechanism (not shown).

[0136] A bag forming die 66 is secured to the upper portion of the slider 65. This die 66 comprises a plurality of vertically extending rod-shaped plungers 66a that are inserted through the insertion holes 64. The distal ends of the plungers 66a are shaped to correspond to the inner surface of the bag 2. The bag forming die 66 moves up and down in response to the vertical movement of the slider 65 driven by the cam mechanism. Furthermore, the lower die 61, the bag forming die 66, and the like can be replaced as appropriate depending on the type of PTP sheet 1 being produced.

[0137] Furthermore, a circulation path 67 for circulating cooling water (or warm water) is formed inside each of the slider 65 and the bag portion forming die 66. This can suppress variations in the surface temperature of each plunger 66a.

[0138] The plunger 66a is set in the order of the initial position, the intermediate stop position, and the protruding position during the forming of the bag portion 2, and finally returns to the initial position.

[0139] The initial position is the position where the plunger 66 a is set at the start of the forming process of the bag portion 2 . The plunger 66 a set at this position is located below the insertion hole 64 and outside the insertion hole 64 .

[0140] The intermediate stop position is a position where the plunger 66 a is set in the intermediate stage of the bag portion 2 forming process. The plunger 66 a set at this position is set inside the insertion hole 64 and is in a state where a predetermined gap is formed between the plunger 66 a and the container film 3 .

[0141] The protruding position is where the plunger 66 a is set in the final stage of the bag 2 forming process. The front end surface of the plunger 66 a set at this position protrudes from the lower mold 61 by an amount corresponding to the depth of the bag 2 .

[0142] Meanwhile, upper mold 71 is fixed to upper plate 73 via plate 72 and is movable toward or away from lower mold 61. Upper mold 71 includes gas supply holes 74 at positions facing insertion holes 64 of lower mold 61.

[0143] Furthermore, a gas supply path 75 connected to the gas supply hole 74 is formed inside the plate 72 and the upper plate 73, and a specified high-pressure gas (inert gas, air in this embodiment) is supplied to the gas supply path 75 from a gas supply device 76 composed of, for example, a compressor.

[0144] In this embodiment, a total of 20 bags 2 corresponding to two PTP sheets 1 are simultaneously formed in a single operation of the bag forming device 16. That is, five bags 2 are simultaneously formed in the film width direction (Y direction) relative to the container film 3 and four bags 2 are simultaneously formed in the film conveying direction (X direction).

[0145] Here, the forming control device 81 will be described. The forming control device 81 controls the forming of the bag portion 2 by the heating device 15 and the bag portion forming device 16, and is composed of a computer system having a CPU, RAM, and the like.

[0146] The forming control device 81 sets and stores information related to the initial position of the plunger 66a of the pocket forming device 16, information related to the intermediate stop position of the plunger 66a, and information related to the protruding position of the plunger 66a. Based on this information, the operation of the plunger 66a is controlled. Furthermore, the information related to the initial position, intermediate stop position, and protruding position of the plunger 66a is appropriately changed according to the depth of the pocket 2 of the PTP sheet 1 to be manufactured.

[0147] Next, the structure of the bag inspection device 21 will be described in detail. Figures 4 to 6 As shown, the bag inspection device 21 includes a lighting device 50 as an irradiation mechanism, a camera 51 as an imaging mechanism, and an inspection control device 52 for controlling these.

[0148] The lighting device 50 is provided from the protruding side of the bag portion 2 ( Figure 4 The lighting device 50 is configured to emit surface light, and includes an electromagnetic wave irradiation device 50a and a diffuser plate 50b covering the electromagnetic wave irradiation device 50a. The lighting device 50 of this embodiment irradiates the container film 3 with electromagnetic waves including ultraviolet light.

[0149] The camera 51 has sensitivity in the wavelength region of the electromagnetic wave irradiated from the lighting device 50. The camera 51 is provided on the opening side of the bag portion 2 of the container film 3 ( Figure 4 The lens is disposed so that the optical axis OL thereof is along the vertical direction (Z direction) perpendicular to the film flat portion 3 b of the container film 3 .

[0150] Furthermore, a band-pass filter 51a is provided corresponding to the lens of the camera 51. The band-pass filter 51a is provided so that only ultraviolet light enters the lens.

[0151] By providing the bandpass filter 51a, the camera 51 captures two-dimensional images of only the ultraviolet light that has passed through the container film 3, among the electromagnetic waves emitted from the lighting device 50. Furthermore, the transmission image data acquired by the camera 51 in this manner is luminance image data with varying luminance at each pixel (at each coordinate position) due to differences in the transmittance of ultraviolet light through the container film 3.

[0152] In particular, in this embodiment, the bandpass filter 51a is a filter that transmits only ultraviolet light with a wavelength of 253±20 nm, for example, at a transmittance of approximately 30±10% through the container film 3. This is because, regardless of whether the transmittance of electromagnetic waves through the container film 3 is too high or too low, there is little concern that a difference in light transmittance will occur between the thin-walled portion and the thick-walled portion of the bottom 2a of the bag portion 2.

[0153] In addition, the camera range of the camera 51 of this embodiment is set as follows: the following range is photographed at one time, and the range includes a total of 20 bags 2 corresponding to two PTP sheets 1 formed on the container film 3 by at least one action of the bag forming device 16, that is, 5 bags in the film width direction (Y direction) relative to the container film 3, and 4 bags 2 in the film conveying direction (X direction).

[0154] The inspection control device 52 is composed of a computer including a CPU (Central Processing Unit) that performs predetermined calculations, a ROM (Read Only Memory) that stores various programs, fixed-value data, etc., a RAM (Random Access Memory) that temporarily stores various data when various calculations are performed, and their peripheral circuits.

[0155] Moreover, the inspection control device 52 operates according to various programs through the CPU, and functions as various functional units such as the main control unit 171, lighting control unit 172, camera control unit 173, image acquisition unit 174, image processing unit 175, learning unit 176, and inspection execution unit 177 described later.

[0156] However, the various functional units mentioned above are realized through the cooperation of various hardware such as the CPU, ROM, RAM, etc., and there is no need to clearly distinguish between functions realized by hardware or software. Some or all of these functions can also be realized by hardware circuits such as IC.

[0157] In addition, the inspection and control device 52 is provided with an input unit 185 composed of a keyboard, a mouse, a touch panel, etc., a display unit 186 with a display screen such as a liquid crystal display, a storage unit 187 capable of storing various data, programs, calculation results, etc., and a communication unit 188 capable of sending and receiving various data to the outside.

[0158] Here, the various functional units constituting the inspection control device 52 will be described in detail. The main control unit 171 is responsible for overall control of the bag inspection device 21 and is configured to transmit and receive various signals with other functional units such as the lighting control unit 172 and the camera control unit 173.

[0159] The lighting control unit 172 is a functional unit that controls the driving of the lighting device 50 , and controls the lighting timing and the like based on a command signal from the main control unit 171 .

[0160] The camera control unit 173 is a functional unit that drives and controls the camera 51 and controls the timing of image capture based on command signals from the main control unit 171. The main control unit 171 controls the timing of illumination and image capture based on signals from an encoder (not shown) provided in the PTP packaging machine 11.

[0161] Thus, for each interval in which the conveyance of the formed container film 3 of the bag portion 2 is temporarily stopped, a process is performed in which the container film 3 is irradiated with electromagnetic waves from the lighting device 50 and the electromagnetic waves (ultraviolet light) transmitted through the container film 3 is captured by the camera 51.

[0162] Then, the transmission image data captured and generated by the camera 51 is converted into a digital signal (image signal) inside the camera 51 and then transmitted to the inspection control device 52 (image acquisition unit 174 ) in the form of a digital signal.

[0163] The image acquisition unit 174 is a functional unit for acquiring image data captured by the camera 51 .

[0164] The image processing unit 175 is a functional unit that performs predetermined image processing on the image data acquired by the image acquisition unit 174. For example, in the learning process described later, a deep neural network 190 (hereinafter referred to as "neural network 190" is generated. Figure 8 ) is used for learning, i.e., learning shading pattern data. Furthermore, inspection shading pattern data is generated for use in the inspection process described later. Furthermore, it is preferred that the imaging conditions for the learning shading pattern data used in the learning process be as consistent as possible with the imaging conditions for the inspection shading pattern data used in the inspection process.

[0165] The learning unit 176 is a functional unit that performs learning of the neural network 190 using learning data and the like, and constructs an AI (Artificial Intelligence) model 200 as a recognition mechanism.

[0166] In addition, as described later, the AI ​​model 200 of this embodiment is a generative model constructed by using only the light and dark pattern data related to the bag portion 2 of the container film 3 of a qualified product without forming defects as learning data (light and dark pattern data for learning) to cause the neural network 190 to perform deep learning, and has a so-called autoencoder (autoencoder) structure.

[0167] Here, refer to Figure 8 The structure of the neural network 190 will be described. Figure 8 FIG is a schematic diagram conceptually showing the structure of the neural network 190. Figure 8 As shown, the neural network 190 has a convolutional auto-encoder (CAE) structure, which has an encoding unit 191 as an encoding unit and a decoding unit 192 as a decoding unit. The encoding unit 191 extracts a feature value (latent variable) TA from the input image data (light and dark pattern data) GA, and the decoding unit 192 reconstructs the image data (light and dark pattern data) GB based on the feature value TA.

[0168] Although the structure of the convolutional autoencoder is known, a detailed description is omitted. However, the encoding unit 191 includes a plurality of convolution layers 193, and in each convolution layer 193, the result obtained by performing a convolution operation on the input data using a plurality of filters (kernels) 194 is output as input data for the next layer. Similarly, the decoding unit 192 includes a plurality of deconvolution layers 195, and each deconvolution layer 195 outputs the result obtained by performing a deconvolution operation on the input data using a plurality of filters (kernels) 196 as input data for the next layer. In addition, in the learning process described later, the weights (parameters) of each filter 194, 196 are updated.

[0169] The inspection execution unit 177 is a functional unit that inspects the formed state of the bag portion 2. Details of the bag portion inspection executed by the inspection execution unit 177 will be described later.

[0170] The storage unit 187 is composed of an HDD (Hard Disk Drive) or SSD (Solid State Drive), and is used to store various setting information used for inspection, inspection results, etc. In this embodiment, the various setting information includes the shapes and sizes of the PTP sheet 1, the bag 2, and the tablet 5; the shape and size of the inspection frame used to define the inspection range (the range corresponding to one PTP sheet 1) and its relative positional relationship with the camera 51; and the shape and size of the bag frame W used to define the area of ​​the bag 2 and its relative positional relationship with the camera 51 (or the inspection frame).

[0171] Furthermore, the storage unit 187 is provided with a predetermined storage area for storing the AI ​​model 200 (the neural network 190 and the learning information obtained through learning).

[0172] The communication unit 188 includes, for example, a wireless communication interface based on a communication standard such as a wired LAN (Local Area Network) or a wireless LAN, and is configured to be able to transmit and receive various data to and from the outside. For example, the communication unit 188 is configured to be able to transmit and receive signals to and from the filling control device 82 described later, and is configured to be able to output inspection results and the like to the filling control device 82.

[0173] Next, refer to Figure 9 , while explaining the bag forming process performed by the control of the forming control device 81.

[0174] In the bag forming step, the intermediate stop position arrangement step S101 is first performed. In the intermediate stop position arrangement step, the bag forming die 66 is moved upward by the movement of the slider 65, thereby moving the plunger 66a set in the initial position upward.

[0175] Then, when the plunger 66a reaches the set intermediate stop position, the movement of the slider 65 stops, and the plunger 66a is positioned at the intermediate stop position. At this time, the front end of the plunger 66a is separated from the container film 3 by a predetermined distance. This predetermined distance is generally less than the depth of the bag portion 2.

[0176] Next, in the clamping step of step S102, the upper mold 71 is moved downward, and the container film 3 is clamped by the fixed lower mold 61 and the upper mold 71. At this time, the container film 3 is located at the molding planned portion 3a (refer to FIG. Figure 7 ) is in a state of being clamped by two molds 61, 71. In addition, the intermediate stop position configuration process and the clamping process can be carried out simultaneously, or the clamping process can be carried out before the intermediate stop position configuration process.

[0177] In the subsequent expansion step S103, gas is supplied from the gas supply device 76 to the gas supply hole 74 via the gas supply passage 75, thereby expanding the portion to be formed 3a of the bag portion 2 in the container film 3 from its surface side ( Figure 7 By supplying the gas, the predetermined portion 3a is formed on the protruding side ( Figure 7 The opposite side (upper side) Figure 7 lower side) bulges, stretches and becomes thinner.

[0178] Then, the planned forming portion 3a bulges until it is supported by the front end surface of the plunger 66a. When the planned forming portion 3a bulges by supplying gas, the wall thickness of the planned forming portion 3a after bulging is substantially uniform throughout.

[0179] Furthermore, the amount of expansion of the container film 3 changes depending on the intermediate stop position of the plunger 66a, and the wall thickness of the intended forming portion 3a also changes. When the intermediate stop position of the plunger 66a is high, the amount of expansion of the container film 3 is small, so the intended forming portion 3a is thicker overall.

[0180] On the other hand, when the intermediate stop position of the plunger 66a is low, the container film 3 is stretched to a large extent, so that the planned forming portion 3a is in a thin state as a whole.

[0181] In the final forming step of the subsequent step S104, the plunger 66a moves upward and is positioned in the protruding position. As a result, the expansion direction of the intended forming portion 3a is reversed, forming the bag portion 2 having a predetermined depth. Therefore, in this embodiment, the plunger 66a, the gas supply device 76, and the like constitute an extending mechanism (extending forming mechanism) that extends a portion of the container film 3 (the intended forming portion 3a) to form the bag portion 2.

[0182] Furthermore, when the container film 3 is deformed by pressing, the portion of the intended forming portion 3a corresponding to the bottom 2a cools due to contact with the plunger 66a, so the portion corresponding to the bottom 2a hardly stretches. Therefore, if the intermediate stop position is set higher, making the intended forming portion 3a thicker overall, the portion corresponding to the bottom 2a remains thicker during the pressing of the plunger 66a, resulting in a thinner side portion 2b of the formed bag portion 2.

[0183] On the other hand, if the intermediate stop position is lowered and the planned forming portion 3a is in a thinner state as a whole, the portion corresponding to the bottom 2a is maintained in a thinner state when the plunger 66a is pressed, so that the side portion 2b of the formed bag portion 2 is thicker.

[0184] By adjusting the intermediate stop position of the plunger 66a in this manner and adjusting the wall thickness of the planned forming portion 3a, the balance of the wall thicknesses of the bottom portion 2a and the side portion 2b of the finally formed bag portion 2 can be adjusted.

[0185] After the final forming step, the plunger 66 a is set to the initial position, and the clamping of the container film 3 by the two molds 61 and 71 is released, thereby completing the bag portion forming step.

[0186] Next, the filling control device 82 will be described. The filling control device 82 controls the filling of the tablets 5 by the tablet filling device 22 and is composed of a computer system including a CPU, RAM, etc. The filling control device 82 constitutes the filling control mechanism in this embodiment.

[0187] In particular, the filling control device 82 of the present embodiment is configured to switch and control whether or not to fill the tablets 5 into a predetermined pocket 2 based on the inspection result of the pocket inspection device 21 .

[0188] Specifically, the filling control device 82 inputs the inspection result of the specified PTP sheet 1 (the molding state of 10 bag portions 2) from the bag portion inspection device 21. When the inspection result is a qualified product judgment result, the tablet filling device 22 is controlled in a manner such that tablets 5 are filled relative to all 10 bag portions 2 included in the PTP sheet 1.

[0189] On the other hand, if the inspection result for a given PTP sheet 1 is a defective determination, the tablet filling device 22 is controlled so that tablets 5 are not filled into all ten pockets 2 included in the PTP sheet 1. Simultaneously, a defective signal is output to the defective sheet discharge mechanism 40. As a result, the PTP sheet 1 (defective sheet) for which the defective signal is given is discharged by the defective sheet discharge mechanism 40.

[0190] Next, refer to Figure 10 The flowchart of will now be used to illustrate the learning process of the neural network 190 performed by the bag inspection device 21.

[0191] First, the operator prepares a container film 3 of acceptable quality (container film 3 corresponding to the plurality of PTP sheets 1 on which the plurality of pockets 2 of acceptable quality are formed). Preferably, the container film 3 prepared here has pockets 2 of the same shape as the container film 3 to be inspected. However, the thickness, material, size, and layout of the pockets 2 of the container film 3 do not necessarily need to be identical. For versatility, it is best to perform learning based on a variety of learning data.

[0192] Next, the operator first places the pre-prepared qualified container film 3 (container film 3 having qualified bag portions 2 formed thereon) at a predetermined inspection position of the bag portion inspection device 21 and then causes the main control unit 171 to execute a predetermined learning program.

[0193] When the learning process starts based on the execution of a predetermined learning program, the main control unit 171 first performs pre-processing for learning the neural network 190 in step S201 .

[0194] Specifically, based on a command from the main control unit 171, the lighting control unit 172 first illuminates the lighting device 50. Next, based on a command from the main control unit 171, the camera control unit 173 drives the camera 51 to capture a predetermined range of the container film 3. This acquires image data of the predetermined range of the container film 3. The image data acquired by the camera 51 is then acquired by the image acquisition unit 174, subjected to predetermined image processing (e.g., shading correction, tilt correction, etc.) by the image processing unit 175, and then stored in the storage unit 187.

[0195] The above series of processing is repeatedly performed while moving the imaging range on the container film 3 until a necessary number of image data (shading pattern data) of the bag portion 2 are obtained as learning data.

[0196] After obtaining the required amount of bag portion 2 image data (shading pattern data) in step S201, the learning unit 176 prepares an unlearned neural network 190 in response to an instruction from the main control unit 171 in the following step S202. For example, the learning unit 176 reads the neural network 190 pre-stored in the storage unit 187 or the like. Alternatively, the learning unit 176 constructs the neural network 190 based on the network structure information stored in the storage unit 187 or the like (e.g., the number of neural network layers or the number of nodes in each layer).

[0197] In step S203, learning shading pattern data is acquired as learning data. Specifically, based on the image data stored in the storage unit 187 in step S201, the image processing unit 175 extracts one bag 2 from the plurality of bags 2 included in the image data and obtains shading pattern data corresponding to the shading pattern (shading distribution image) generated on the bottom 2a of the bag 2 as a single piece of learning shading pattern data. This learning shading pattern data is then output to the learning unit 176. Specifically, only the shading pattern data of the bag 2 of the container film 3 that is a qualified product without forming defects is used as the learning data (learning shading pattern data).

[0198] In step S204, reconstructed shading pattern data is acquired. Specifically, based on a command from the main control unit 171, the learning unit 176 provides the learning shading pattern data acquired in step S203 as input data to the input layer of the neural network 190. This acquires the reconstructed data output from the output layer of the neural network 190, i.e., the shading pattern data, as the reconstructed shading pattern data.

[0199] In the next step S205 , the learning unit 176 compares the learning shading pattern data acquired in step S203 with the reconstructed shading pattern data output by the neural network 190 in step S204 , and determines whether the error is sufficiently small (below a predetermined threshold).

[0200] Here, when the error is sufficiently small, the neural network 190 and its learning information (updated parameters described later, etc.) are stored in the storage unit 187 as the AI ​​model 200, and the learning process is terminated.

[0201] On the other hand, if the error is not sufficiently small, after performing network update processing (learning of the neural network 190 ) in step S206 , the process returns to step S203 and the above series of processing is repeated.

[0202] Specifically, in the network update process of step S206, the weights (parameters) of the filters 194 and 196 in the neural network 190 are updated to more appropriate weights using a known learning algorithm such as backpropagation, so as to minimize the loss function representing the difference between the learning shading pattern data and the reconstructed shading pattern data. Alternatively, for example, a binary cross-entropy (BCE) can be used as the loss function.

[0203] By repeating these processes multiple times, the error between the learning shading pattern data and the reconstructed shading pattern data is minimized in the neural network 190, and more accurate reconstructed shading pattern data is output.

[0204] Next, refer to Figure 11 The flowchart of the bag inspection device 21 is used to illustrate the process of bag inspection.

[0205] in addition, Figure 11 The bag inspection process shown is a process that is performed for each inspection range corresponding to the range of a PTP sheet 1 punched into a rectangular sheet as a product. That is, at each interval when the transport of the container film 3 is temporarily stopped, the two inspection ranges are respectively inspected. Figure 11 The following is a detailed description of the bag inspection.

[0206] When a predetermined range of the container film 3 formed in the bag portion 2 by the bag portion forming device 16 temporarily stops in the bag portion inspection device 21 , the inspection control device 52 causes the main control unit 171 to execute a predetermined inspection program.

[0207] When the inspection process starts according to the execution of a predetermined inspection program, first, electromagnetic waves (ultraviolet light) are irradiated from the lighting device 50 to a predetermined area of ​​the container film 3 (irradiation step), and imaging is performed by the camera 51 (imaging step).

[0208] Specifically, based on a command from the main control unit 171, the lighting control unit 172 illuminates the lighting device 50, and the camera control unit 173 drives the camera 51. This captures a predetermined inspection range on the container film 3, obtaining transmission image data encompassing the plurality of bags 2. This transmission image data is then acquired by the image acquisition unit 174.

[0209] When the transmission image data of the container film 3 is input to the image acquisition unit 174 , the inspection control device 52 first performs an inspection image acquisition process (step S301 ).

[0210] Specifically, according to the instructions from the main control unit 171, the image processing unit 175 uses the above-mentioned inspection frame based on the transmission image data of the container film 3 obtained by the image acquisition unit 174 to obtain image data of the inspection range corresponding to one PTP sheet 1 (including the range of 10 bag parts 2) as an inspection image, and stores it in the storage unit 187.

[0211] Furthermore, in this embodiment, the position at which the area corresponding to each PTP sheet 1 on the container film 3 stops is always constant relative to the imaging range of the camera 51, and the set position of the inspection frame is predetermined based on its relative positional relationship with the camera 51. Therefore, in this embodiment, the set position of the inspection frame is not adjusted each time based on the image data. However, the present invention is not limited to this, and the set position of the inspection frame may be appropriately adjusted based on information obtained from the image data, taking into account the occurrence of positional deviations, etc.

[0212] In addition, various processing schemes can be formed for the inspection image. For example, since there are technical limitations in uniformly irradiating the entire imaging range with electromagnetic waves from the lighting device 50, a shading correction scheme can be used to correct for variations in electromagnetic wave intensity (brightness) caused by different positions.

[0213] When acquiring the inspection image, the inspection control device 52 performs a masking process in the next step S302 .

[0214] Specifically, based on the instruction from the main control unit 171, the image processing unit 175 sets the bag frames W (see FIG. 1 ) in correspondence with the positions of the ten bag portions 2 on the inspection image obtained in step S301. Figure 12 ), and a process of setting a mask M is performed on an area outside the bag area determined by the bag frame W, that is, an area corresponding to the film flat portion 3b.

[0215] Furthermore, in this embodiment, the set position of the pocket frame W is predetermined based on its relative positional relationship with the inspection frame. Therefore, in this embodiment, the set position of the pocket frame W is not always adjusted based on the inspection image. However, the present invention is not limited to this. Alternatively, the set position of the pocket frame W may be appropriately adjusted based on information obtained from the inspection image, taking into account the occurrence of positional deviations.

[0216] Next, in step S303 , the inspection control device 52 sets the values ​​of the bag-conforming flags of all the bag units 2 to “0”.

[0217] The "good bag flag" indicates the quality determination result of the corresponding bag portion 2 and is set in a predetermined area of ​​the storage unit 187. If a predetermined bag portion 2 is determined to be good, the corresponding good bag flag is set to "1".

[0218] In the next step S304 , the inspection control device 52 sets the value C of the bag number counter set in the storage unit 187 to “1” as an initial value.

[0219] The "bag number" is a consecutive number set corresponding to each of the ten bags 2 within an inspection range, and the position of the bag 2 can be specified by the value C of the bag number counter (hereinafter referred to as "bag number C").

[0220] Next, in step S305 , the inspection control device 52 determines whether the bag number C is equal to or less than the number of bags N (“10” in this embodiment) per inspection range (per PTP sheet 1 ).

[0221] If the determination is "yes," the process proceeds to step S306. In step S306, based on a command from the main control unit 171, the inspection execution unit 177 executes a shading pattern extraction process (shading pattern extraction process) to extract shading pattern data for the bag 2 of the current bag number C. The function executing this process primarily constitutes the shading pattern extraction mechanism of this embodiment.

[0222] Specifically, the shade pattern within the bag frame W of the bag portion 2 corresponding to the current bag portion number C (for example, C=1) in the inspection image (masked image data) subjected to the masking process in step S302 is used as the shade pattern K generated on the bottom 2a of the bag portion 2 (refer to Figure 12 The light and dark patterns K1 and Figure 14 The light and dark pattern data corresponding to the light and dark pattern K2, etc. are extracted.

[0223] in addition, Figure 12 2 is a diagram showing a light and dark pattern K1 produced on the bottom 2a of the bag portion 2 without forming failure. Figure 13 To indicate along Figure 12 The graph of the brightness value of each pixel on the line A-A of the shading pattern K1 is shown.

[0224] also, Figure 14 FIG. 2 is a diagram showing a light and dark pattern K2 produced on the bottom 2a of the bag portion 2 having a poorly formed shape. Figure 15 To indicate along Figure 14 The graph of the brightness value of each pixel on the line BB of the shading pattern K2 is shown.

[0225] That is, the light and dark pattern (light and dark pattern data) K is two-dimensional image information having brightness information (such as any value in 256 grayscales from 0 to 255) for each pixel, corresponding to the relationship between the difference in wall thickness (wall thickness distribution) at each position (coordinate position) of the bottom 2a of the bag portion 2 and the transmittance of the electromagnetic wave passing therethrough, and represents an image of the two-dimensional distribution of light and dark produced in the bottom 2a of the bag portion 2 and the like (an image of the intensity distribution of the transmitted electromagnetic wave).

[0226] In addition, in this embodiment, since the bag frame W is set corresponding to the opening peripheral portion of the bag portion 2 (the connecting portion between the side portion 2b and the membrane flat portion 3b), the light and dark pattern K obtained through the process S306 includes not only the bottom 2a of the bag portion 2, but also the side 2b of the bag portion 2, and the light and dark pattern of the corner 2c of the bag portion 2 where the bottom 2a and the side 2b intersect.

[0227] Furthermore, within the shading pattern K, the area corresponding to the bottom 2a exhibits a shading distribution (brightness distribution) that roughly corresponds to the wall thickness distribution of the bottom 2a. Meanwhile, the luminance information for the areas corresponding to the side portions 2b and corner portions 2c corresponds not to electromagnetic waves transmitted along the wall thickness directions (X and Y directions) of the side portions 2b and the like, but rather to electromagnetic waves transmitted along the direction of extension during molding (Z direction). Therefore, the luminance information is less correlated with the wall thickness of the side portions 2b and the like.

[0228] In the next step S307, the inspection execution unit 177 executes a reconstruction process (reconstruction data acquisition process). The function of executing this process mainly constitutes the reconstruction data acquisition means in this embodiment.

[0229] Specifically, the inspection execution unit 177 inputs the shading pattern data K related to the bottom portion 2a of the bag portion 2 with bag number C (e.g., C=1) extracted in step S306 into the input layer of the AI ​​model 200. Then, the inspection execution unit 177 obtains the reconstructed data, i.e., the shading pattern data, reconstructed by the AI ​​model 200 and output from the output layer as the reconstructed shading pattern data KS of the bottom portion 2a of the bag portion 2 with bag number C (e.g., C=1).

[0230] Here, AI model 200 not only inputs Figure 12 In the case of the light and dark pattern data K1 related to the bottom 2a of the bag portion 2 without the defective forming, and in the case of inputting Figure 14 In the case of the shading pattern data K2 related to the bottom 2a of the bag portion 2 having a poor formation, the reconstructed shading pattern data KS is outputted by learning as described above. Figure 16 (a) shows the shading pattern data related to the bottom 2a of the bag portion 2 without forming defects.

[0231] In the next step S308, the inspection execution unit 177 executes a comparison process (comparison process). The function of executing this process mainly constitutes the comparison means in this embodiment.

[0232] Specifically, the inspection execution unit 177 first compares the shading pattern data K extracted in step S306 with the reconstructed shading pattern data KS acquired in step S307, and calculates the difference in brightness between each pixel in the two data sets. Next, pixels for which this difference does not fall within a predetermined tolerance range are identified as defective pixels Px, and the total number of these defective pixels Px (defective area) is calculated. Figure 16 (b) is a diagram schematically showing a location where a defective pixel Px has occurred during the comparison process.

[0233] Then, in step S309, the inspection execution section 177 determines whether the number of defective pixels Px calculated in step S308 is equal to or less than a preset determination standard Po. Specifically, by determining whether the number of defective pixels Px is within an allowable range, a quality determination is made regarding the formed state of the bag portion 2. The quality determination process (quality determination step) executed in step S309 constitutes the quality determination mechanism of this embodiment.

[0234] Without limitation, a method for determining whether a product is good or bad can be formed by determining whether the connected component with the largest area among the connected components of the defective pixel Px is within an allowable range, or by determining the degree of deviation (distribution) of the connected components of the defective pixel Px. Of course, another method is also possible, that is, regardless of the size, as long as there is a single defective pixel Px, the product is determined to be defective.

[0235] If it is determined in step S309 that the number of defective pixels Px is equal to or less than the determination standard Po, the process proceeds to step S310. On the other hand, if the determination is "No," the bag portion 2 corresponding to the current bag number C is deemed defective, and the process proceeds directly to step S311.

[0236] In step S310 , the inspection execution unit 177 regards the bag portion 2 corresponding to the current bag number C as a qualified product, sets the value of the bag portion qualified product flag corresponding to the bag portion number C to “1”, and proceeds to step S311 .

[0237] Then, the inspection control device 52 adds "1" to the current bag number C in step S311 and then returns to step S305.

[0238] Here, if the newly set bag number C is still less than the number of bags N ("10" in this embodiment), the process goes to step S306 again and the above series of processes are repeatedly performed.

[0239] On the other hand, when it is determined that the newly set bag number C exceeds the number of bags N, the quality determination process for all the bag portions 2 is deemed to be completed, and the process proceeds to step S312.

[0240] In step S312, the inspection control device 52 determines whether the value of the bag conforming product flag of all the bags 2 within the inspection range is "1." This determines whether the PTP sheet 1 corresponding to the inspection range is a conforming product or a defective product.

[0241] Here, when the judgment is "yes", that is, all the bag parts 2 within the inspection range are "qualified products" and there is no bag part 2 judged as "unqualified products", in step S313, the PTP sheet 1 corresponding to the inspection range is judged as "qualified product" and the inspection process is ended.

[0242] On the other hand, if the judgment in step S312 is "No", that is, if there is even one bag part 2 judged as "unqualified" within the inspection range, in step S314, the PTP sheet 1 corresponding to the inspection range will be judged as "unqualified", and the inspection process will be terminated.

[0243] In addition, in the conforming product determination process of step S313 and the defective product determination process of step S314 , the inspection control device 52 stores the inspection results related to the PTP sheet 1 corresponding to the inspection range in the storage unit 187 and outputs them to the filling control device 82 .

[0244] As described in detail above, according to this embodiment, the following scheme is formed, wherein, at each interval when the transportation of the formed container film 3 of the bag portion 2 is temporarily stopped, electromagnetic waves are irradiated from the lighting device 50 to the container film 3, and the electromagnetic waves (ultraviolet light) passing through the container film 3 are photographed by the camera 51. Based on the acquired transmission image data, the light and dark pattern (light and dark pattern data) K generated at the bottom 2a of the bag portion 2 is extracted, and based on this, a judgment is made on the quality of the formed state of the bag portion 2.

[0245] Through this scheme, of course, it is possible to make a good or bad judgment related to the forming state (wall thickness distribution state) of the bottom 2a of the bag part 2, and it is also possible to make a good or bad judgment related to the forming state (wall thickness distribution state) of the side 2b and corner 2c of the bag part 2. It can better detect the presence or absence of deviation in the wall thickness distribution of the side 2b of the bag part 2, etc., and the poor forming (poor wall thickness) of the side 2b of the bag part 2.

[0246] In particular, in the present embodiment, the AI ​​model 200 constructed by learning the neural network 190 is used to inspect the forming state of the side portion 2 b of the bag 2 and the like.

[0247] Specifically, a configuration is adopted in which shading pattern data K obtained by imaging the bottom portion 2a of the bag portion 2 is compared with reconstructed shading pattern data KS obtained by reconstructing the shading pattern data K using the AI ​​model 200 to perform quality determination.

[0248] This makes it possible to inspect even small molding defects, bottom 2a shapes, and bag portions 2 with complex thickness distributions that were previously difficult to detect. As a result, inspection accuracy can be further improved.

[0249] In addition, in the present embodiment, since the light and dark pattern data K obtained by photographing the bag portion 2 is compared with the reconstructed light and dark pattern data KS obtained by reconstructing the light and dark pattern data K, the two light and dark pattern data being compared are not affected by the differences in the photographing conditions on the side of the container film 3 serving as the inspection object (such as the setting position, setting angle, deflection, etc. of the container film 3) and the photographing conditions on the side of the bag portion inspection device 21 (such as the lighting status of the lighting device 50, the viewing angle of the camera 51, etc.), and smaller forming defects can be detected more accurately.

[0250] Furthermore, in this embodiment, since the forming state of the entire circumference of the side portion 2b can be grasped by imaging the bottom portion 2a of the bag 2 at once, inspection can be accelerated, thereby improving the productivity of the blister package.

[0251] Furthermore, in this embodiment, ultraviolet light with a wavelength of 253±20 nm, which has a transmittance of approximately 30±10% through the container film 3, is used as electromagnetic wave for inspection of the translucent container film 3. Because ultraviolet light has a lower transmittance than visible light and is less likely to penetrate the translucent container film 3, inspection of the formed state of the bag 2 can be performed more effectively. Furthermore, the difference in light transmittance between the thin-walled and thick-walled portions of the bottom 2a of the bag 2 facilitates more effective inspection.

[0252] Furthermore, the present invention is not limited to the contents of the above-mentioned embodiment, and may be implemented as follows, for example. Of course, other application examples and modification examples not listed below are also possible.

[0253] (a) The structure of the blister package to be inspected is not limited to the above-described embodiment. For example, in the above-described embodiment, the blister package includes the PTP sheet 1 for accommodating contents such as tablets 5.

[0254] The inspection is not limited to this. For example, various blister packages can be inspected, such as peel-off blister packages (bags for storing food, etc.) that remove the contents by peeling off the cover film from the container film, blister packages (carrier tapes, etc.) that store and transport electronic devices, etc., and blister packages that do not install a cover film on the container film but assemble backing paper.

[0255] (b) The configuration of the pockets in the container film, such as their shape, size, depth, number, and arrangement, is not limited to the above-described embodiment and can be appropriately selected depending on the type, shape, and intended use of the contents. For example, the bottom 2a of the pocket 2 can be substantially triangular, elliptical, quadrilateral, or rhombus-shaped when viewed from above.

[0256] More specifically, for example, Figure 17 (a) Figure 17 The blister package 500 shown in (b) is the inspection object. The blister package 500 has a recessed portion 501. The pocket 501 consists of a rectangular bottom 501a in a plan view and a rectangular frame-shaped side portion 501b extending from the bottom 501a. The bottom 501a of the pocket 501 has a plurality of bulging ribs 501c formed thereon, bulging toward the inside of the pocket.

[0257] For such a bag portion 501, according to the inspection process of the above-mentioned embodiment, the light and dark images generated at the bottom 501a are extracted (process S306). After reconstruction processing (process S307) and comparison processing (process S308), low-brightness areas corresponding to the thick-walled areas (bulging ribs 501c) and high-brightness areas corresponding to the thin-walled areas (ordinary parts of the bottom 501a) are obtained, and a good or bad judgment related to the forming state of the bag portion 501 can be made.

[0258] (c) The material and layer structure of the container film and cover film are not limited to those in the above embodiment. For example, in the above embodiment, the container film 3 is formed of a colorless and transparent thermoplastic resin material such as PP and PVC and has light transmittance.

[0259] The container film 3 is not limited thereto. For example, the container film 3 may be made of a colorless, translucent resin material, a colored, transparent, or colored, translucent resin material, or may be formed of an opaque material (opaque resin material, metal material, etc.). Examples of the metal material include aluminum-based materials such as aluminum laminate films.

[0260] Furthermore, as will be described later, the container film 3 formed of an opaque material can be inspected by irradiating electromagnetic waves such as X-rays that can transmit the opaque material from the lighting device 50 .

[0261] (d) The method of forming the bag portion is not limited to the above-described embodiment. For example, in the above-described embodiment, the bag portion 2 is formed by a plunger-assisted pressure-air forming method.

[0262] Instead of this method, various known forming methods such as vacuum forming, pressure forming, and plunger forming may be used to partially heat and soften a portion of the flat container film 3 (the portion to be formed 3 a ) and then perform a stretching process.

[0263] However, when the container film is an aluminum laminate, there is a risk of delamination between the adhesive layers due to heating, leading to rupture during forming. Therefore, cold forming (cold forming) without prior heating is suitable. Even in such cases, the container film does not necessarily stretch uniformly during bag forming, as it is prone to stretching near the clamping portion, for example. This can lead to variations in wall thickness across the bag.

[0264] (e) The configuration of the illumination mechanism and the imaging mechanism is not limited to the above-described embodiment. For example, in the above-described embodiment, the illumination device 50 is disposed on the protruding side of the bag portion 2, and the imaging camera 51 is disposed on the opening side of the bag portion 2. However, the positions of the two may be reversed.

[0265] Furthermore, in the above-described embodiment, the lighting device 50 is configured to emit electromagnetic waves including ultraviolet light. However, the wavelength of the electromagnetic waves emitted from the lighting device 50 may be appropriately changed according to the material, color, etc. of the container film 3. Of course, a configuration may be adopted in which the bandpass filter 51a is omitted and the electromagnetic waves emitted from the lighting device 50 and transmitted through the container film 3 are directly incident on the camera 51.

[0266] For example, when the container film 3 is made of an opaque material such as aluminum, X-rays may be irradiated from the lighting device 50. Alternatively, when the container film 3 is made of a colored translucent material, visible light such as white light may be irradiated from the lighting device 50.

[0267] (f) In the above embodiment, ultraviolet light of a wavelength of 253±20 nm with a transmittance of approximately 30±10% of the container film 3 is used for inspection. However, electromagnetic waves of a different wavelength may be used for inspection.

[0268] However, regardless of whether the transmittance of electromagnetic waves passing through the container film 3 is too high or too low, there is a problem that the transmittance of light between the thin-walled portion and the thick-walled portion of the bottom 2a of the bag portion 2 is unlikely to differ. Therefore, it is preferred to use electromagnetic waves with a wavelength in which the transmittance of the container film 3 is in the range of 15% to 60%, and more preferably, electromagnetic waves with a wavelength in which the transmittance of the container film is in the range of 20% to 50%.

[0269] (g) The structure of the AI ​​model 200 (neural network 190 ) as the recognition mechanism and its learning method are not limited to the above-described embodiment.

[0270] (g-1) Although not specifically mentioned in the above embodiment, various data may be normalized as needed during the learning process of the neural network 190 or the reconstruction process in the bag inspection process.

[0271] (g-2) The structure of the neural network 190 is not limited to Figure 8 The structure shown may be, for example, a structure in which a pooling layer is provided after the convolution layer 193. Of course, different structures such as the number of layers of the neural network 190, the number of nodes in each layer, and the connection structure of each node may also be adopted.

[0272] (g-3) In the above embodiment, the AI ​​model 200 (neural network 190) is a generative model having a structure of a convolutional autoencoder (CAE), but is not limited to this. For example, it can also be a generative model having a structure of different types of autoencoders such as a variational autoencoder (VAE).

[0273] (g-4) In the above embodiment, the neural network 190 is learned by the error back propagation method, but the present invention is not limited thereto and may be learned using various other learning algorithms.

[0274] (g-5) Neural network 190 may also be formed by a dedicated AI processing circuit, such as a so-called AI chip. In this case, only learning information such as parameters may be stored in storage unit 187, and the dedicated AI processing circuit may read this information and set it in neural network 190, thereby forming AI model 200.

[0275] (g-6) In the above embodiment, the learning unit 176 is included, and the neural network 190 is learned within the inspection and control device 52. However, the present invention is not limited to this. At least the AI ​​model 200 (the learned neural network 190) may be stored in the storage unit 187, and the learning unit 176 may be omitted. Therefore, the neural network 190 may be learned outside the inspection and control device 52 and stored in the storage unit 187.

[0276] (h) In the above embodiment, a bag inspection device 21 is provided inside the PTP packaging machine (blister packaging machine) 11 that performs filling with contents such as tablets 5. This is not limiting. For example, in a production line that separately manufactures the container film 3 and packages the contents, the bag inspection device 21 may be included in the container film 3 manufacturing apparatus. Alternatively, a configuration may be provided that includes an inspection device that inspects the formed container film 3 of the bag 2 offline, independent of the container film 3 manufacturing apparatus.

[0277] Description of the label:

[0278] Label 1 indicates a PTP sheet; label 2 indicates a bag portion; label 2a indicates a bottom portion; label 2b indicates a side portion; label 2c indicates a corner portion; label 3 indicates a container film; label 4 indicates a cover film; label 5 indicates a tablet; label 11 indicates a PTP packaging machine; label 15 indicates a heating device; label 16 indicates a bag forming device; label 21 indicates a bag inspection device; label 50 indicates a lighting device; label 51 indicates a camera; label 52 indicates an inspection control device; label 176 indicates a learning portion; label 177 indicates an inspection execution portion; label 190 indicates a neural network; label 200 indicates an AI model; label K (K1, K2) indicates a light and dark pattern.

Claims

1. An inspection device for inspecting the forming state of a bag portion of a blister package, characterized in that: The inspection device includes: an irradiation mechanism capable of irradiating a container film on which the bag portion is formed with a predetermined electromagnetic wave; a camera mechanism, which is provided on the side opposite to the irradiation mechanism via the container film and is capable of capturing the electromagnetic waves that have passed through at least the bottom of the bag portion to obtain image data; a shading pattern extraction mechanism capable of extracting shading pattern data corresponding to the shading pattern produced on the bottom of the bag portion by irradiation with the electromagnetic wave, based on the image data obtained by the imaging mechanism; an identification unit that generates learning data for a neural network using only the shading pattern data associated with the pocket portion without defective forming as learning data, the neural network comprising an encoding unit that extracts a feature value from the input shading pattern data and a decoding unit that reconstructs the shading pattern data based on the feature value; a reconstructed data acquisition mechanism capable of acquiring reconstructed shading pattern data, the reconstructed shading pattern data being reconstructed by inputting the shading pattern data extracted by the shading pattern extraction mechanism into the recognition mechanism; a comparing mechanism capable of comparing the shading pattern data extracted by the shading pattern extracting mechanism with the reconstructed shading pattern data acquired by the reconstructed data acquiring mechanism; and A quality judgment mechanism is provided, which can judge the quality of at least the forming state of the side portion of the bag portion based on the comparison result of the comparison mechanism.

2. The inspection device according to claim 1, characterized in that The container film is made of a light-transmitting resin film material; The irradiation mechanism is configured to be capable of irradiating ultraviolet light as the electromagnetic wave.

3. The inspection device according to claim 1, characterized in that The electromagnetic waves include electromagnetic waves having a wavelength at which the transmittance of the container film is within a range of 15% to 60%.

4. The inspection device according to claim 2, characterized in that The electromagnetic waves include electromagnetic waves having a wavelength at which the transmittance of the container film is within a range of 15% to 60%.

5. The inspection device according to claim 1, wherein: The bag portion is thermoformed on the flat container film.

6. The inspection device according to claim 2, characterized in that The bag portion is thermoformed on the flat container film.

7. The inspection device according to claim 3, characterized in that The bag portion is thermoformed on the flat container film.

8. The inspection device according to claim 4, characterized in that The bag portion is thermoformed on the flat container film.

9. A blister packaging machine, characterized in that: This blister packaging machine includes the inspection device according to any one of claims 1 to 8.

10. A method for manufacturing a blister package, wherein a predetermined content is contained in a pocket formed on a container film, and a cover film is attached to close the pocket, wherein: The manufacturing method of the blister package comprises: a bag forming step in which the bag is formed relative to the container film conveyed in a belt shape; a filling step of filling the bag portion with the contents; An installation step, in which the strip-shaped cover film is installed on the container film filled with the contents in the bag portion so as to close the bag portion; a cutting step of cutting the blister package from a strip-shaped body having the cover film attached to the container film; and an inspection step of inspecting a formed state of the bag portion of the blister package; The above inspection process includes: an irradiation step of irradiating the container film having the bag portion formed thereon with predetermined electromagnetic waves; an imaging step of imaging the electromagnetic wave that has passed through at least the bottom of the bag portion to obtain image data; a shading pattern extraction step of extracting shading pattern data corresponding to the shading pattern produced on the bottom of the bag portion by irradiation with the electromagnetic wave based on the image data obtained in the imaging step; a reconstruction data acquisition step in which the shading pattern data extracted in the shading pattern extraction step is input to a recognition mechanism to obtain reconstructed shading pattern data as reconstructed shading pattern data, wherein the recognition mechanism learns and generates, as learning data, only the shading pattern data related to the pocket portion without forming defects, for a neural network having an encoding unit that extracts a feature value from the input shading pattern data and a decoding unit that reconstructs the shading pattern data from the feature value; a comparing step of comparing the shading pattern data extracted in the shading pattern extracting step with the reconstructed shading pattern data acquired in the reconstructed data acquiring step; and A quality determination step is performed in which a quality determination is performed on a molding state of at least a side portion of the bag portion based on the comparison result in the comparison step.

Images