Packaging defect prediction device, packaging defect prediction method and paper leaf processing system

Abstract

To provide a packaging defect prediction device, a packaging defect prediction method and a paper leaf processing system for predicting shrink-wrapping defects in advance.SOLUTION: A packaging defect prediction device predicts, based on prior information, the occurrence of shrink-wrap defects on packaged products just before the products are fed into a shrink-wrapping machine for shrink-wrapping.SELECTED DRAWING: Figure 8

Description

[Technical field]

[0001] An embodiment of the present invention relates to a packaging defect prediction device, a packaging defect prediction method, and a paper sheet processing system that predict packaging defects caused by a shrink packaging device. [Background technology]

[0002] 2. Description of the Related Art In paper sheet processing systems that process paper sheets such as securities, shrink devices are used to wrap the paper sheets that are the packaged goods in packaging material and then shrink the packaging material by applying heat or the like to complete the package. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-20811 [Patent Document 2] JP 2019-182576 A [Patent Document 3] JP 2004-256161 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, if the shrink device fails to shrink the packaging material, the packaging material must be discarded. Also, if the packaging material that has failed to shrink is stored in a warehouse, the quality of the packaged products is adversely affected.

[0005] The problem to be solved by the present invention is to provide a packaging defect prediction device, a packaging defect prediction method, and a paper sheet processing system that predict shrink packaging defects in advance. [Means for solving the problem]

[0006] A packaging defect prediction device according to an embodiment predicts the occurrence of shrink packaging defects for a packaged product based on prior information immediately before the packaged product enters a shrink packaging machine that shrink packages the packaged product. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is a configuration block diagram of a packaging defect prediction system according to an embodiment. [Diagram 2] FIG. 2 is a configuration block diagram of the prediction device according to the embodiment. [Diagram 3] FIG. 3 is a schematic diagram illustrating an example of the operation of the labeling unit according to the embodiment. [Figure 4] FIG. 4 is a schematic diagram showing the configuration and data flow when the learning unit of the prediction device according to the embodiment learns an AI model. [Diagram 5] FIG. 5 is an example of a data table stored in the learning unit of the prediction device according to the embodiment. [Figure 6] FIG. 6 is a flowchart illustrating an example of an operation performed by the learning unit of the prediction device according to the embodiment when learning an AI model. [Figure 7] FIG. 7 is a schematic diagram showing a data flow when the determination unit of the prediction device according to the embodiment determines whether the packaging condition is defective. [Figure 8] FIG. 8 is a flowchart illustrating an example of an operation performed by the determination unit of the prediction device according to the embodiment when determining whether a packaging condition is defective. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. (Embodiment) In this embodiment, a system for predicting packaging defects caused by a shrink packaging machine is shown. This embodiment can be applied to a paper sheet processing system that processes packaging of paper sheets such as securities.

[0009] FIG. 1 is a configuration block diagram of a packaging defect prediction system according to an embodiment.

[0010] The packaging defect prediction system includes a prediction device 1 and a packaging machine 2, and is a system in which the prediction device 1 predicts whether or not a packaged product output by the packaging machine 2 is defective, or the probability that the packaged product is defective.

[0011] The prediction device 1 is, for example, a computer server equipped with a CPU having computer functions such as a calculation function and a control function, a volatile memory or a non-volatile memory for storing data, a communication function with the outside, and has an artificial intelligence function (sometimes called AI), learns one or more pieces of data as learning data by machine learning or the like to obtain a model (parameters), and uses the obtained model (called a learning model) to make predictions about defects in packaged products. The functions of the prediction device 1 may include a processing function by a program executed by the CPU or the like.

[0012] The packaging machine 2 is a device that packages a packaged product 4 (e.g., paper sheets in a paper sheet processing system) that is the object to be packaged, in a packaging material 5 (e.g., a plastic bag) and outputs the packaged product. The packaged product packaged in the packaging material is called a packaged product. The packaging machine 2 is roughly equipped with the functions of a labeling section 21 and a shrink section 22. The labeling unit 21 has a function of covering the packaged product 4 with the packaging material 5. The details of the operation will be described later.

[0013] The shrink unit 22 shrinks the packaging material 5 by heat or the like for the packaged product 6 (hereinafter referred to as a temporary packaged product) output by the labeling unit 21, and generates and outputs a packaged product (which may be called a fully packaged product when it is particularly distinguished from the temporary packaged product) by matching the packaging material 5 of the temporary packaged product to the shape of the packaged product 4. In Fig. 1, for example, packaged product 6-3 shows an example image of a temporary packaged product, and packaged product 6-5 shows an example image of a fully packaged product.

[0014] The conveyor 3 is a conveyor along which packaged products 4 and packaging products 6 move, and in the figure, it moves from the left, where a packaged product generating device (not shown) is located, to the right, where a warehouse is located. Packaged goods 4 are products packaged in packaging material 5 such as paper sheets. Packaged goods 4-1, 4-2, 4-3, 4-4, and 4-5 shown in the figure indicate packaged goods 4 at each position, and may be seen as the same packaged goods 4 flowing on a conveyor, or as separate packaged goods 4 continuously flowing on a conveyor. When packaged goods 4-1, 4-2, 4-3, 4-4, and 4-5 are not particularly distinguished from one another, they are referred to as packaged goods 4.

[0015] The packaging material 5 is used to package the packaged product 4, and in this embodiment, it is, for example, a plastic bag. Like the packaged product 4, the packaging materials 5-1, 5-2, 5-3, 5-4, and 5-5 shown in the figure indicate the packaging material 5 in each position. When there is no particular distinction between the packaging materials 5-1, 5-2, 5-3, 5-4, and 5-5, they will be referred to as packaging material 5.

[0016] Packaged product 6 indicates the whole packaged product 4 packaged in packaging material 5. Like packaged product 4, packages 6-1, 6-2, 6-3, 6-4, and 6-5 shown in the figure indicate packaged product 6 in each position. When there is no particular distinction between packages 6-1, 6-2, 6-3, 6-4, and 6-5, they will be referred to as packaged product 6. However, packaged product 6 before shrink-wrapping may also be referred to as temporary packaged product 6, and packaged product 6 after shrink-wrapping may also be referred to as fully packaged product 6.

[0017] The speed measuring unit 23 measures the moving speed of the conveyor 3 and the packaged products 6 . The static electricity measuring unit 24 measures the amount of static electricity in the packaged product 4. In particular, when the packaged product 4 is a paper sheet, static electricity is generated in the packaged product 4 due to friction with the conveyor 3, etc., but it is difficult to remove the static electricity by earthing, etc., and it affects the function of the labeling unit 21 in the subsequent stage. The temperature measuring section 211 measures the temperature inside the shrink section 22. The temperature inside the shrink section 22 has a strong effect on the packaging material 5, and a strong effect on the quality of the packaging of the packaged product 6.

[0018] FIG. 2 is a configuration block diagram of the prediction device according to the embodiment.

[0019] The learning unit 11 is equipped with AI machine learning based on a neural network such as deep learning, acquires input data TD, executes learning by machine learning, and generates a learning model. The AI ​​of this embodiment has a K-dimensional input layer (K ​​is a natural number) and a one-dimensional output layer. The data input to the input layer is called model input data, and the data output by the output layer is called model output data.

[0020] The input data TD is an example of learning data for the learning unit 11, and is data linked to the packaged product 4, such as the amount of static electricity, heater temperature, packaged product intrusion speed, packaged product size, packaged product type, packaging material type, and the presence or absence of packaging defects. In this embodiment, of the input data TD, the pass / fail judgment information is treated as correct answer data (corresponding to model output data), and at least one of the remaining data is treated as model input data. Furthermore, as described above, when the packaged product 4 is a paper sheet, the effect of static electricity is large, and it is desirable to include at least the amount of static electricity in the model input data.

[0021] The packaging information recording unit 111 is, for example, a storage device including a non-volatile memory, and stores the input data TD acquired by the learning unit 11. The defective packaged product characteristic learning unit 112 executes learning by machine learning or the like based on the input data TD, and generates a learning model, etc.

[0022] The judgment unit 12 acquires necessary data from the packaging machine 2 while the packaging machine 2 is in operation, and determines the presence or absence of packaging defects (good / bad judgment information), the incidence rate of packaging defects, etc., using the learning model generated by the learning unit 11. The incidence rate of packaging defects indicates the probability that the packaging of perfectly packaged products 6 output from the packaging machine 2 has failed, and the probability that the packaging of perfectly packaged products 6 output from the packaging machine 2 is normal is particularly referred to as the normal packaging rate. The judgment unit 12 may output the obtained presence or absence of packaging defects, the incidence rate of packaging defects, etc., or may make a predictive judgment of packaging defects based on the incidence rate of packaging defects.

[0023] The packaging defect occurrence rate calculation unit 121 calculates and outputs the packaging defect occurrence rate using the learning model generated by the learning unit 11. For example, the packaging defect occurrence rate (=1-normal packaging rate) may be calculated using an output value of an output layer (e.g., a sigmoid function) of the learning model.

[0024] The packaging defect determination unit 122 predicts whether or not a packaging defect will occur in the packaged product 4 immediately before entering the shrink unit 22 based on the presence or absence of packaging defects, the rate of occurrence of packaging defects, and the like, and outputs the result.

[0025] The control unit 13 generates and outputs a control signal for controlling the packaging machine 2 based on the prediction result output by the packaging defect judgment unit 122. The control unit 13 may also control each function inside the prediction device 1 in response to an instruction from a keyboard, mouse, or the like (not shown) that is a peripheral device of the prediction device 1. For example, the control unit 13 may switch between a learning mode in which the learning unit 11 is activated and an operation mode in which the judgment unit 12 is activated in response to an instruction from a keyboard, mouse, or the like (not shown).

[0026] The presentation control unit 14 generates display data for causing the presentation unit 15 to display the prediction result output by the packaging defect determination unit 122 and audio data for causing the presentation unit 15 to output the prediction result as audio, and outputs the display data to the presentation unit 15. The presentation unit 15 may be a monitor or a speaker, and may display the display data input from the presentation control unit 14, or output audio data as audio from the speaker. The input / output unit 16 exchanges data with the outside of the prediction device 1.

[0027] 3A to 3D are schematic diagrams showing an example of the operation of the labeling unit according to the embodiment. The operation of the labeling unit is shown in chronological order from (a) to (d) of FIG. The conveyor 3 is assumed to move from left to right.

[0028] The packaging material bonding section 222 is provided inside the labeling section 21, and operates to place the packaging material 5 on the top of the conveyor 3 and close the opening of the packaging material 5 when the packaged product 4 is covered with the packaging material 5. Details of the operation will be described below with reference to the drawings.

[0029] 3(a) shows the state in which the packaged product 4 is in front of the packaging material 5, and the packaging material 5 is placed so as to be lowered vertically onto the conveyor 3 from the packaging material adhesive part 222. As the packaged product 4 moves together with the conveyor 3 and hits the packaging material 5 as shown in FIG. 3(b), the packaging material 5 stretches as if being pulled out from the packaging material adhesive part 222.

[0030] As the packaged product 4 moves further and is covered with packaging material 5 as shown in Fig. 3(c), the upper part of packaging material adhesive part 222 moves down as shown in Fig. 3(d) and the lower part of packaging material adhesive part 222 moves up from the conveyor 3, so that the upper and lower parts of packaging material adhesive part 222 sandwich the opening of packaging material 5. In this state, the opening of packaging material 5 is adhered by applying heat to the upper and lower parts of packaging material adhesive part 222, etc., to create packaged product 6 (temporarily packaged product).

[0031] The operation of the prediction device 1 will be described below in the order of AI learning and operation.

[0032] 4 is a schematic diagram showing the configuration and data flow when the learning unit of the prediction device according to the embodiment learns an AI model. Blocks with the same functional names and symbols as those in FIGS. 1 and 2 indicate the same functional blocks.

[0033] The packaged product 4 is input to the packaging machine 2 and output as a packaged product 6. The packaged product 6 output from the packaging machine 2 is input to an inspection device 7.

[0034] The inspection device 7 inspects the packaged product 6 output by the packaging machine 2. For example, the inspection device 7 may inspect the packaged product 6 by X-ray inspection or image inspection using AI. Specifically, the packaging defect determination unit 71 determines whether the packaged product 6 input to the inspection device 7 is good or bad, and outputs the good or bad data (OK / NG, etc.) to the prediction device 1. The prediction device 1 stores the input good or bad data in the packaging information recording unit 111. For example, if the packaging material 5 is torn or the binding is weak, the packaged product 6 is determined to be good or bad (NG), and otherwise the packaged product 6 is determined to be good (OK). Also, without using the inspection device 7, a human may inspect the packaged product 6 visually or the like, determine the inspection result, and store the good or bad data in the packaging information recording unit 111.

[0035] The packaging machine 2 outputs measurement data such as the amount of static electricity, heater temperature, and packaged product intrusion speed obtained immediately before inputting the temporarily packaged product 6 into the shrink section 22 to the prediction device 1, and stores the data in the packaging information recording section 111 of the learning section 11. Furthermore, packaged product information, which is attribute information of the packaged product 4 such as the height of the packaged product 4 and the horizontal plane size (length and width size), the type of the packaged product 4, and packaging material information, which is attribute information of the packaging material 5, are stored in the packaging information recording section 111 of the learning section 11. The packaged product information and packaging material information may be input from a keyboard (not shown), or the packaging machine 2 may acquire the packaged product information from the label of the packaged product 4 and transmit it to the prediction device 1 together with the measurement data. The measurement data may be referred to as environmental data as data other than the packaged product information and packaging material information. The measurement data, packaged product information, and packaging material information affect the shrink state of the packaging material 5 of the packaged product 6.

[0036] FIG. 5 is an example of a data table stored in the learning unit of the prediction device according to the embodiment, in which each column shows data for each item, and each row shows the packaged product ID, measurement data, packaged product information, and quality assurance information for the package linked to the packaged product 6 input into the inspection device.

[0037] In the data table DT, the data items for each column are shown as an example including static electricity amount C-1, heater temperature C-2, intrusion speed C-3 as measurement data, packaged product size C-4 and packaged product type C-5 as packaged product information, and quality judgment information C-6 as a quality judgment result, but any data items related to the quality judgment of the packaging may be added in M ​​or more items (natural number). For example, packaging material information such as the type of material of the packaging material may be added to the data items. Furthermore, it is desirable that the data for each package 6 shown for each row be a large amount of data of N or more items (natural number greater than 1), assuming that it will be input into the learning function of an AI.

[0038] For example, taking data E-2 as an example, the measurement data when the packaging machine 2 generated packaged item ID2, such as static electricity amount CD-12, heater temperature CD-22, and intrusion speed CD-32, as well as packaged item information of the packaged item 4 packaging packaged item ID2, such as packaged item size CD-42 and packaged item type CD-52, and pass / fail judgment information CD-62, which is the result of inspection of packaged item ID2 by inspection device 7, are stored in the packaging information recording unit 111.

[0039] In this embodiment, pass / fail judgment information is treated as correct answer data (corresponding to model output data), and at least one item of data among measurement data, packaged product information, and packaging material information is treated as model input data.

[0040] FIG. 6 is a flowchart illustrating an example of an operation performed by the learning unit of the prediction device according to the embodiment when learning an AI model.

[0041] The learning unit 11 acquires input data (corresponding to TD in FIG. 2) and stores it in the packaging information recording unit 111 (step S11). In step S11, the input data TD may be data acquired during testing or the like, or may be data acquired during operation. The learning unit 11 performs data preprocessing such as selecting and processing data (features) using the measurement data and packaged product information stored in the packaging information recording unit 111 as model input data, and performs learning using the quality assessment information of the packaging as correct answer data to create a learning model (steps S12 and 13).

[0042] Below, an example of the operation of the judgment unit to calculate the packaging defect occurrence rate and detect packaging defects using the created learning model will be shown.

[0043] FIG. 7 is a schematic diagram showing a data flow when the determination unit of the prediction device according to the embodiment determines whether the packaging condition is defective.

[0044] In normal operation of the packaging machine 2, when a packaged product 4 output from a packaged product generation device and carried on a conveyor 3 is input, processing such as attaching packaging material 5 to the packaged product 4 is performed to generate a completely packaged product 6, which is output to a warehouse or the like. In the packaging machine 2 of this embodiment, the labeling unit 21 wraps the packaged product 4 in packaging material 5 to generate a temporarily packaged product 6, and immediately before the temporarily packaged product 6 is placed in the shrink unit 22, the static electricity measurement unit 24, temperature measurement unit 211 and speed measurement unit 23 measure the static electricity amount CD-1, heater temperature CD-2 and penetration speed CD-3, and output them to the prediction device 1.

[0045] FIG. 8 is a flowchart illustrating an example of an operation performed by the determination unit of the prediction device according to the embodiment when determining whether a packaging condition is defective.

[0046] When the prediction device 1 detects that a temporary packaged product has been output from the labeling unit 21 by a detection unit not shown (step S100), it acquires measurement data from the packaging machine 2 (step S101). In step S100, when the product number of the detected temporary packaged product (which may be the packaged product ID in FIG. 5) is received from an external device or the like, the product number is stored in the packaging information recording unit 111, and the measurement data acquired in step S101 is linked to the product number and stored in the packaging information recording unit 111. The packaging defect occurrence rate calculation unit 121 of the determination unit 12 obtains the packaging defect occurrence rate from the acquired measurement data and a learning model generated in advance (step S102).

[0047] In the packaging defect occurrence rate calculation unit 121, it is assumed that the learning model generated by the learning unit 11 has been downloaded in advance to a memory (not shown) and that packaged product information and packaging material information have been separately input. The packaging defect occurrence rate calculation unit 121 inputs the input measurement data, packaging material information, and packaged product information to the learning model and outputs the packaging defect occurrence rate. The calculated packaging defect occurrence rate is stored in the packaging information recording unit 111 in association with the product number.

[0048] The predicted packaging defect occurrence rate is input to the packaging defect determination unit 122, which compares the packaging defect occurrence rate with a preset threshold value (step S103). If the packaging defect occurrence rate is equal to or greater than the threshold value, the packaging defect determination unit 122 determines that there is a high possibility of a packaging defect occurring (step S104). Based on the determination result of the packaging defect determination unit 122, the control unit 13 generates a control signal, such as stopping the conveyor 3 to prevent the temporarily packaged products 6 from entering the shrink unit 22, and outputs the control signal to the packaging machine 2 (step S105). The packaging machine 2 performs an operation to prevent the temporarily packaged products 6 from entering the shrink unit 22, such as temporarily stopping the conveyor 3 that is normally operating.

[0049] Also, in step S105, the presentation control unit 14 may generate image data or audio data for presenting the determination result of the packaging defect determination unit 122 on the presentation unit 15, and present the result to an operator by displaying the image data on the presentation unit 15 or outputting audio from a speaker or the like. Similarly, in step S104, the presentation unit 15 may display image data of the packaging defect occurrence rate, and the operator may control the packaging machine 2 based on the packaging defect occurrence rate.

[0050] On the other hand, if the packaging defect occurrence rate is lower than the threshold value, the packaging defect determination unit 122 determines that the possibility of packaging defects occurring is low. In the packaging machine 2, the conveyor 3 is not stopped as in normal operation, and the temporarily packaged products 6 are placed in the shrink section 22 and output as completely packaged products 6 (step S106). Also, in step S106, the presentation control unit 14 may generate image data or audio data for presenting the determination result of the packaging defect determination unit 122 on the presentation unit 15, and notify the operator by causing the presentation unit 15 to display the image data or output audio from a speaker or the like.

[0051] By the above procedure, it is possible to predict whether or not packaging defects such as opening or tearing will occur before shrinking the packaging material 5 of the packaged product 6. Conventionally, if shrinking of the packaging material 5 fails, the packaging material 5 must be discarded, but according to this embodiment, prediction can be made before shrinking the packaging material 5, thereby eliminating waste of the packaging material 5.

[0052] In the present embodiment, the learning unit 11 generates a learning model in which the presence or absence of packaging defects (pass / fail judgment information) is the correct answer data, but the learning unit 11 may execute machine learning using the packaging defect occurrence rate as the correct answer data and packaged product information, packaging material information, and environmental data as input data to create a learning model. In this case, for example, in FIG. 5, when the measurement data and packaged product information are almost constant during the period when the pass / fail judgment information CD-6 is acquired, the ratio of rejects (NG) in the entire data of the pass / fail judgment information CD-6 is set as the packaging defect occurrence rate, and is set as the correct answer data for the learning model of the learning unit 11. On the other hand, when the measurement data value and packaged product information fluctuate during the period when the pass / fail judgment information CD-6 is acquired, for example, the data table DT is divided for each measurement data value and packaged product information that can be considered to be constant, and the packaging defect occurrence rate is calculated for each data, and is set as the correct answer data. The packaging defect occurrence rate calculation unit 121 of the judgment unit 12 directly acquires the packaging defect occurrence rate from the acquired measurement data and the learning model (corresponding to step S102 in FIG. 8).

[0053] In this embodiment, the packaging defect judgment unit 122 predicts whether the product is good or bad just before the shrink section 22, but the prediction may also be made just before the packaging machine 2 (before the packaged product 4 enters the packaging machine 2).

[0054] In particular, this embodiment can be applied to a paper sheet processing system that processes paper sheets such as securities. In a paper sheet processing system in which the packaged material 5 is paper sheets, the processing time from when the packaging machine 2 outputs the packaged product 6 to the warehouse is short for security reasons, and if an inspection device 7 such as that shown in FIG. 4 cannot be installed, it is necessary to inspect the packaged products 6 that have accumulated in the warehouse for defects after the packaged products 6 are stored in the warehouse, which is very time-consuming, but this embodiment can solve this problem. In addition, in a paper sheet processing system, the packaging machine 2 can be controlled, but it may be difficult to stop the packaged products 6 output from the packaging machine 2 from being stored in the warehouse. According to this embodiment, the packaging machine 2 can be stopped before the packaging machine 2 outputs the packaged product 6, so that the possibility of a packaged product 6 with poor packaging being transported to the warehouse can be reduced.

[0055] According to at least one of the embodiments described above, it is possible to provide a packaging defect prediction device, a packaging defect prediction method, and a paper sheet processing system that predict shrink packaging defects in advance.

[0056] The present invention is not limited to the above-described embodiment, and the components can be modified and embodied in the implementation stage without departing from the gist of the invention. In addition, various inventions can be formed by appropriately combining the multiple components disclosed in the above-described embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, components from different embodiments may be appropriately combined.

[0057] In addition, the processing shown in the flowcharts, sequence charts, etc. may be realized by hardware such as a CPU, an IC chip, or a digital signal processor (DSP), or by software (e.g., a program) operated on a computer including a microcomputer, or by a combination of hardware and software. In addition, the present invention also includes cases where the order of steps in the flowcharts and sequence charts is changed without departing from the spirit of the invention.

[0058] The present invention is also applicable when the claims are expressed as a control logic, a program including instructions for causing a computer to execute the control logic, or a computer-readable recording medium having the instructions written thereon. The names and terms used are not intended to be limiting, and other expressions that have substantially the same content and intent are also included in the present invention. [Explanation of symbols]

[0059] 1... prediction device, 2... packaging machine, 3... conveyor, 4... packaged item, 5... packaging material, 6... packaged item, 7... detection device, 11... learning unit, 12... judgment unit, 13... control unit, 14... presentation control unit, 15... presentation unit, 16... input / output unit, 23... speed measurement unit, 24... static electricity measurement unit, 111... packaging information recording unit, 112... defective packaged item characteristic learning unit, 121... packaging defect occurrence rate calculation unit, 122... packaging defect judgment unit, 211... temperature measurement unit, 222... packaging material bonding unit.

Claims

1. A packaging defect prediction device that predicts the occurrence of defects in shrink packaging of a packaged product based on prior information immediately before the packaged product enters a shrink packaging machine that shrink packages the packaged product.

2. The packaging defect prediction device according to claim 1, wherein the advance information includes at least one data item selected from packaged product information, which is attribute information of the packaged product, packaging material information, which is attribute information of the packaging material, and the measurement data measured immediately before.

3. The packaging defect prediction device according to claim 2, further comprising a defect occurrence rate which is a probability of occurrence of a defect in the shrink packaging.

4. The packaging defect prediction device according to claim 3 , further comprising a determination unit that determines whether or not a packaging defect has occurred based on the defect occurrence rate.

5. The packaging defect prediction device according to claim 4 , wherein the determination unit determines that a packaging defect will occur when the defect occurrence rate is equal to or greater than a threshold value.

6. Equipped with an artificial intelligence learning section based on neural networks, The packaging defect prediction device according to claim 3 , wherein the learning unit generates a learning model using the packaged product information, the packaging material information, and the measurement data as input data and the presence or absence of packaging defects as correct answer data.

7. 7. The packaging defect prediction device according to claim 6, further comprising a determination unit that generates the defect occurrence rate based on the packaged product information, the packaging material information, the measurement data, and the learning model acquired before the shrink packaging machine shrink-processes the packaged products.

8. The packaging defect prediction device according to claim 5 , further comprising a step of controlling the shrink packaging machine based on the determination by the determination unit.

9. A packaging defect prediction method for predicting the occurrence of packaging defects due to shrink processing of a packaged product based on prior information immediately before the packaged product enters a shrink packaging machine that shrink-packages the packaged product.

10. A paper processing system comprising the packaging defect prediction device according to claim 1.

Images