JP2023031034A - Quality management device of take-up roll, quality management method of take-up roll, and quality management program of take-up roll - Google Patents

Abstract

To provide a quality management device of a take-up roll, a quality management method of a take-up roll, and a quality management program of a take-up roll in which a calculated value of an internal state of a take-up roll is outputted in real time regarding a take-up roll being manufactured, in a manufacturing process of the take-up roll.SOLUTION: A quality management device 102 of a take-up roll receives, during manufacturing of a take-up roll, input of advance data 111 that is physical property parameters at the time of manufacturing a take-up roll, and during-manufacturing data 112 that is operation parameters of a take-up device necessary for manufacturing a take-up roll, from an acquisition part such as a sensor provided in the take-up device 101 for taking up a take-up roll, and calculates a first roll internal state calculated value that is a calculated value of an internal state of a take-up roll during manufacturing based on a first algorithm 113.SELECTED DRAWING: Figure 1

Description

The present invention relates to a winding roll quality control device, a winding roll quality control method, and a winding roll quality control program.

In recent years, with the development of flexible electronics using flexible materials such as films and fibers, and printable electronics applying coating and printing technology to part or all of the manufacturing process, plastic film The web, which is a thin and flexible material such as , has come to be widely adopted as a high-performance core material. In particular, plastic films are indispensable as basic materials for flat panel displays, wireless tags, organic memories, pressure sensor sheets, biopotential sensors, organic image sensor sheets, flexible substrates and flexible batteries.

Here, the outline of the secondary processing process of the plastic film by the roll-to-roll method will be described. After the film is formed, the roll is transported while being unwound by an unwinding device. After being processed, it is wound into a roll again by a winding device.

Such plastic films undergo the various secondary processes described above according to the intended use of the finished product, are wound up into rolls, are stored and transported to the next processing step, and are repeatedly processed to produce film products. It is common to be finished. Here, when the winding roll is stored and transported, the internal stress state of the roll changes according to the environmental temperature and the passage of time. In some cases, defective winding occurs, and the flatness and flatness of the film are lost at the location where the defective winding occurs, resulting in a decrease in commercial value. The internal stress state of the roll means the radial stress, the circumferential stress, and the thickness of the air layer in the take-up roll.

The environmental temperature of the winding roll during transportation and storage in a storage location such as a warehouse varies depending on the season, the country or region where the storage location is located, the presence or absence of heat treatment, and the heat treatment conditions. are often different. When the roll temperature changes due to these environmental conditions, the film undergoes thermal strain and the internal stress of the take-up roll changes. In addition, the longer the storage period, the more the film creeps and the internal stress changes.

Thermal strain is caused by expansion or contraction of the plastic film due to temperature change, and this deformation changes the internal stress. In addition, the film has viscoelastic properties because it is made from a polymeric material. Since the film in the take-up roll is in a stressed state, the film creeps and the internal stress changes. It is known that stress and temperature affect the creep behavior of polymeric materials (Non-Patent Document 1). Therefore, it is believed that the internal stress in the take-up roll and the roll temperature affect film creep.

A winding roll internal stress analysis program capable of performing a simulation close to the actual behavior of the winding roll is known (Patent Document 1).

Further, a program for analyzing the internal stress of a winding roll that has wound a web, calculating the thickness of the initial air layer formed between the winding roll and the newly wound layer, and A method of calculating the thickness of the air layer in the roll and calculating the radial stress in the roll is known (Patent Document 2).
Furthermore, a thermoelastic model considering thermal strain and a viscoelastic model considering viscoelastic characteristics are known. Since the effects of thermal strain and viscoelastic properties act simultaneously on the internal stress during storage, transportation, and heat treatment of the winding roll, the predicted values of the internal stress obtained by the winding models of each previous study are sufficient. It is difficult to say that the prediction accuracy is ensured. To solve this problem, a thermoviscoelastic model regarding the internal stress of the take-up roll in an unsteady state, which combines the viscoelastic properties of the film and the effects of thermal strain, is known (Non-Patent Document 1).

JP 2013-64650 A JP 2012-017159 A

Toshimitsu Kanda, et al., ``Thermal Stress Analysis of Winding Roll Considering Viscoelastic Properties of Plastic Film (Second Report, Thermal Viscoelastic Model and Its Experimental Verification)'', Transactions of the Japan Society of Mechanical Engineers (C) , Vol. 77, No. 783, November 2011, p.4242 (p.282)

However, in the technical field of web handling, it is difficult to output in real time a first roll internal state calculation value, which is a calculated value of the internal state of the winding roll, regarding the winding roll being manufactured in the manufacturing process of the winding roll. could not.

The present invention provides a winding roll capable of outputting in real time a first roll internal state calculation value, which is a calculation value of the internal state of the winding roll, for the winding roll being manufactured in the manufacturing process of the winding roll. It is an object of the present invention to provide a take-up roll quality control device, a take-up roll quality control method, and a take-up roll quality control program.

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. a first algorithm calculation unit that inputs certain production data and calculates a first roll internal state calculation value, which is a calculation value of the internal state of the winding roll, based on a first algorithm; Characterized by

The winding roll quality control device further includes:
A first trained AI model that has learned a correlation between the first roll internal state calculated value and a first pass/fail judgment result that is a pass/fail judgment result regarding manufacturing of the winding roll. is characterized by having a learning unit of

The winding roll quality control device further includes:
A first inference unit that inputs the first roll internal state calculation value and outputs a first pass/fail judgment result regarding manufacturing of the winding roll using the first learned AI model. Characterized by

In the winding roll quality control device,
The first inference unit
When the first pass/fail judgment result regarding the manufacture of the winding roll is inferred to be acceptable, the data during manufacture corresponding to the input calculated value of the internal state of the roll is transferred to the next operation of the winding device. It is characterized in that it is reflected in manufacturing data, which is a parameter.

In the winding roll quality control device,
The first inference unit
An error is output when the first pass/fail judgment result regarding the manufacture of the winding roll is inferred to be rejected.

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Enter some manufacturing data and
A first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, is calculated based on a first algorithm.

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Input certain manufacturing data,
A first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, is calculated based on a first algorithm.

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. a first algorithm calculation unit for inputting certain manufacturing data and calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
A first learning unit that generates a first trained AI model that has learned a correlation between the first roll internal state calculated value and first suggestion data that is data that suggests a manufacturing parameter for the next process. characterized by having

The winding roll quality control device further includes:
It is characterized by comprising a first inference unit that inputs the first roll internal state calculated value and outputs the first suggestion data using the first trained AI model.

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Enter some manufacturing data and
calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
generating a first learned AI model that has learned a correlation between the first roll internal state calculated value and first suggestive data that is data suggestive of manufacturing parameters for the next process; .

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Input certain manufacturing data,
calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
generating a first learned AI model that has learned a correlation between the first roll internal state calculated value and first suggestive data that is data that suggests a manufacturing parameter for the next process; .

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
A first calculated roll internal state value of the winding roll in which the defect has occurred is input, and the first calculated roll internal state calculation value and operating parameters of the winding device necessary for manufacturing the winding roll are obtained. A first learned AI that has learned the correlation between first cause analysis data that is data that caused the defect, including data during manufacturing and prior data that is the physical property parameter of the winding roll It is characterized by having a first inference unit that outputs first cause analysis data using a model.

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
inputting the first roll internal state calculation value of the winding roll in which the defect occurred,
Defects including the first roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and prior data that are physical property parameters of the winding roll The first causal analysis data is output using the first trained AI model that has learned the correlation with the first causal analysis data that is the data that caused the.

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
Input the first roll internal state calculation value of the winding roll in which the problem occurred,
Defects including the first roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and prior data that are physical property parameters of the winding roll The first causal analysis data is output using the first learned AI model that has learned the correlation with the first causal analysis data that is the data that caused the.

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. a second algorithm calculation unit for
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; It is characterized by having a second learning unit.

The winding roll quality control device further includes:
a second inference unit that inputs the second roll internal state calculated value and outputs a second pass/fail judgment result regarding transportation or storage of the winding roll using the second learned AI model; It is characterized by

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. death,
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; It is characterized by

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. let
Generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll. It is characterized by

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. a second algorithm calculation unit for
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; It is characterized by having a second learning unit.

The winding roll quality control device further includes:
a second inference unit that inputs the second roll internal state calculated value and outputs a second pass/fail judgment result regarding transportation or storage of the winding roll using the second learned AI model; It is characterized by

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. death,
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; It is characterized by

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. let
Generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll. It is characterized by

The winding roll quality control device according to the present invention includes:
In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to the transportation or storage of the winding roll, are input, and the calculated value of the second roll internal state and the operating parameters of the winding device necessary for manufacturing the winding roll are used. A second trained AI model that learns the correlation between first cause analysis data, which is data that caused the defect, including certain in-manufacturing data and prior data, which is the physical property parameter of the winding roll. is used to output second cause analysis data.

A winding roll quality control method according to the present invention includes:
In the winding roll quality control method for quality control of the winding roll,
the computer
inputting transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll;
Defects including the second roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and preliminary data that are physical property parameters of the winding roll The second causal analysis data is output using the second trained AI model that has learned the correlation with the first causal analysis data, which is the data that caused the

A winding roll quality control program according to the present invention includes:
In the winding roll quality control program for quality control of the winding roll,
to the computer,
input transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll;
Defects including the second roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and preliminary data that are physical property parameters of the winding roll It is characterized by outputting second cause analysis data using a second learned AI model that has learned the correlation between the first cause analysis data that is the data that caused the above.

According to the present invention, a winding roll quality control apparatus according to the present invention is a winding roll quality control apparatus that performs quality control of a winding roll, and includes advance data, which are physical property parameters of the winding roll, and Calculating the internal state of the take-up roll by inputting in-manufacturing data, which are operation parameters of the take-up device necessary for manufacturing the take-up roll, from an acquisition unit provided in the take-up roll to take up the take-up roll By calculating the first roll internal state calculation value, which is a value based on the first algorithm, in the winding roll manufacturing process, the calculated value of the internal state of the winding roll for the winding roll being manufactured can be output in real time.

Block diagram of a winding roll quality control device Schematic diagram of a winding roll machine with central drive with nip rollers Flowchart for calculating the roll internal state calculation value at the time of manufacturing using the first algorithm Flowchart when using the first algorithm to learn the roll internal state calculation value at the time of manufacture to the first AI model Flowchart for inferring the roll internal state calculation value at the time of manufacturing using the first algorithm to the first AI model Flowchart for inferring suggested data for the next process during manufacturing using the first algorithm to the first AI model Flowchart for inferring manufacturing data that causes defects from market feedback data using the first AI model Flowchart for making the second AI model learn the roll internal state calculation value during transportation and storage using the second algorithm Flowchart for inferring the roll internal state calculation value during transportation and storage using the second algorithm to the second AI model Flowchart for inferring suggested data for the next process during transportation and storage using the second algorithm to the second AI model Flowchart when market feedback data is used to infer transportation storage data that causes defects using the second AI model

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a winding roll quality control system centering on a winding roll quality control apparatus according to the present invention.
The winding device 101 is a winding device that winds up the web in the final step of the winding roll manufacturing process. Here, the winding roll quality control device 102 is generally a computer or the like having a central processing unit and a main storage device (memory) connected to the central processing unit (CPU). is an information processing device.

In the first embodiment of the present invention, the winding roll quality control device 102 includes prior data 111, which are physical property parameters at the time of manufacturing the winding roll, and In-manufacturing data 112, which are operation parameters of the winding device necessary for manufacturing the winding roll, are input from an acquisition unit such as a sensor (not shown) provided in the winding device 101, and the winding roll at the time of manufacture is input. A first roll internal state calculated value, which is a calculated value of the internal state of the roll, is calculated based on a first algorithm 113 described later (not shown, but provided corresponding to the first algorithm 113) It has an algorithm calculation part.
In this way, in the manufacturing process of the winding roll, not only the preliminary data 111 but also the in-manufacturing data 112 regarding the winding roll being manufactured from the acquisition unit such as the sensor provided in the winding device 101 is sent to the winding device 101. By inputting and outputting a first roll internal state calculation value, which is a calculated value of the internal state of the winding roll, it is possible to grasp the internal state of the winding roll being manufactured in real time.

Examples of the pre-data 111 and the in-manufacturing data 112 are given below.
・Preliminary data (input data that must be obtained in advance before manufacturing the winding roll)
Mandatory data on web: Young's modulus (radial and circumferential), thickness Mandatory data on core: outer radius, Young's modulus Optional data on web (preferred height): Coefficient of linear expansion (radial and circumferential) (*1) ), viscoelastic properties (radial and circumferential)
Optional data on core (preferred height): Coefficient of linear expansion Optional data on web (preferred): Surface roughness, Poisson's ratio, coefficient of static friction between webs Optional data on core (preferred): Wall thickness, Poisson's ratio Optional on nip roller Data (middle priority): outer radius, rubber hardness (or Young's modulus of rubber), Poisson's ratio Optional data on web (preferred low): air permeability, thermal conductivity (*1), specific heat (*1), density ( *1)
Optional data on the core (preferred low): thermal conductivity (*1), specific heat (*1), density (*1)
・Data during manufacturing (input data detected by the sensor during manufacturing of the winding roll)
Required data: tension, number of web layers (or winding length)
Optional data (preferred height): indoor temperature (*1), indoor humidity (*1), production time (*2)
Optional data (medium priority): Line speed, web width, nip load Optional data (low priority): Web thickness profile in width direction (*1 is storage environment temperature and humidity, *2 is data related to storage time)

In the second embodiment of the present invention, the winding roll quality control device 102 further includes a first roll internal state calculated value calculated by the first algorithm calculation unit and a pass/fail judgment regarding the manufacture of the winding roll. The correlation between the resulting first pass/fail judgment result is learned by the first pre-learning AI model to generate the first trained AI model 114 (not shown, the first trained AI model 114).
Here, the first trained AI model 114 that the first learning unit learns is the first pre-learning AI model, which is an untrained neural network (not shown), with the first roll internal state calculated value and It is generated by learning the correlation between the first pass/fail judgment result, which is the pass/fail judgment result regarding the manufacture of the winding roll.
As the neural network, an autoencoder, a CNN (Convolutional Neural Network), an RNN (Recurrent Neural Network), or the like can be appropriately selected and used. As a learning method by the first learning unit, for example, it is possible to adopt supervised learning in which the first pass/fail judgment result, which is the pass/fail judgment result regarding the manufacture of the winding roll, is used as teacher data (label). can. This first learned AI model 114 can be used to make inferences regarding pass/fail decisions regarding the manufacture of the winding roll being manufactured.

Further, in the third embodiment of the present invention, the winding roll quality control device 102 includes the first roll internal state calculated value calculated by the first algorithm calculation unit and the pass/fail judgment result regarding manufacture of the winding roll. Let the first pre-learned AI model learn the correlation between the first pass/fail judgment result and, after generating the first trained AI model 114, the first roll for the winding roll being manufactured The internal state calculation value is input to the first trained AI model 114, and the first pass/fail judgment result regarding the manufacture of the winding roll is output (not shown, provided corresponding to the first trained AI model 114). has a first inferencing unit).

Next, when it is inferred that the first acceptance/rejection determination result regarding the manufacture of the winding roll is acceptable, the first inference section generates the production-in-progress data 112 corresponding to the input first roll internal state calculation value. By reflecting the data in the manufacturing data 117, which are the next operation parameters of the winding device 101, the optimum production of the winding roll can be performed with the optimum operation parameters.

Further, the first inference unit outputs an error as the first notification output 115 when it is inferred that the first pass/fail judgment result regarding the manufacture of the winding roll is rejected. Furthermore, by automatically stopping the production line when the error is output, it is possible to quickly stop the production of winding rolls whose quality may be rejected, thereby reducing wasteful costs.

In the fourth embodiment of the present invention, the winding roll quality control device 102 includes prior data 111, which are the physical property parameters of the winding roll, and In-manufacturing data 112, which are operation parameters of the winding device necessary for manufacturing the winding roll, are input from an acquisition unit provided in the winding device, and the first data 112, which is the calculated value of the internal state of the winding roll based on a first algorithm 113 (not shown, but provided corresponding to the first algorithm 113), a first algorithm calculation unit, and a first roll internal state calculation A first trained AI model 114 that has learned the correlation between the value and the first next-process suggestion data 116, which is data suggesting the manufacturing parameter of the next process, is generated (not shown, but the first learned AI model 114 is It has a first learning unit (provided corresponding to the AI model 114).
This first trained AI model 114 can be used to make inferences about first next step suggestion data 116 that suggests next step manufacturing parameters.

Further, the winding roll quality control device 102 inputs the first roll internal state calculation value of the winding roll calculated by the first algorithm calculation unit based on the first algorithm 113, and obtains the first learned value. It has a first inference unit (not shown, provided corresponding to the first trained AI model 114) that outputs first next step suggestion data 116 using the AI model 114. FIG.
Based on the internal state of the winding roll, the first inference section outputs the first next-process suggestion data 116, which is data suggesting the manufacturing parameters of the manufacturing line of the next process, and outputs the next-process of the winding roll. can be optimally manufactured in

Next, in the fifth embodiment of the present invention, the winding roll quality control device 102 inputs the first roll internal state calculation value as the market feedback data for the winding roll in which the defect occurred in the market, The cause of the defect including the calculated value of the first roll internal state, the in-manufacturing data which is the operation parameter of the winding device necessary for manufacturing the winding roll, and the preliminary data which is the physical property parameter of the winding roll Output first cause analysis data 119 using the first trained AI model 114 that has learned the correlation with the first cause analysis data, which is data that has become (provided corresponding to the trained AI model 114).
This first reasoning unit can input the first roll internal state calculation value as feedback data from the market, and output the first cause analysis data that caused the defect, and can analysis becomes easier.

In the sixth embodiment of the present invention, the take-up roll quality control device 102 inputs transport/storage data 121, which are environmental parameters relating to the transport or storage of the take-up roll, when transporting or storing the take-up roll. Then, a second roll internal state calculated value, which is a calculated value of the internal state of the winding roll during transportation or storage, is calculated based on a second algorithm 122 described later (not shown, but the second algorithm 122 a second algorithm calculation unit provided corresponding to the second algorithm calculation unit, the second roll internal state calculation value calculated by the second algorithm calculation unit, and the result of the pass/fail judgment regarding the export or storage of the winding roll The correlation between the pass/fail judgment result is learned by the first pre-learning AI model, and the second trained AI model 123 is generated (not shown, provided corresponding to the second trained AI model 123). ) and a second learning unit.
This second trained AI model 123 can be used for reasoning regarding whether to pass or fail to transport or store a winding roll during transport or storage.

Here, an example of the transportation storage data 121 is given below.
Storage data (input data as assumed values for the transportation and storage state of the winding roll: input values for estimating the future transportation and storage state of the winding roll)
Mandatory item: time Optional item (priority): temperature (or temperature difference) (*1), humidity (or humidity difference) (*1), storage time (*2)
Optional items (high priority): temperature change speed, humidity change speed, heat transfer coefficient inside and outside of winding roll Environmental temperature/humidity, *2 data related to storage time)

Further, in the seventh embodiment of the present invention, the winding roll quality control device 102 further inputs a second roll internal state calculation value of the winding roll during transportation or storage, and generates a second learned AI It has a second inference unit (not shown, but provided corresponding to the second trained AI model 123) that outputs a second pass/fail judgment result regarding transportation or storage of the winding roll using the model 123. .

Then, the second inference unit outputs an error as the second notification output 124 when it is inferred that the second pass/fail judgment result regarding the transportation or storage of the winding roll is rejected.
Furthermore, by stopping transportation or storage when the error is output, wasteful costs can be reduced by immediately stopping transportation or storage of the winding roll whose quality may be rejected.

Next, in the eighth embodiment of the present invention, the take-up roll quality control device 102, when transporting or storing the take-up roll, stores the transport storage data 121, which are the environmental parameters related to the transport or storage of the take-up roll. input, and a second roll internal state calculated value, which is a calculated value of the internal state of the take-up roll, is calculated based on the second algorithm 122 (not shown, but provided corresponding to the second algorithm 122 ) A first learning that learns the correlation between the second algorithm calculation unit and the second calculated value of the roll internal state and the second next process suggestion data 125 that is data that suggests the manufacturing parameters of the next process. It has a second learning unit (not shown, provided corresponding to the second trained AI model 123) that generates a trained AI model 123.
This second trained AI model 123 can be used for reasoning about second next-process suggestive data 125 suggesting manufacturing parameters for the next process.

In addition, the winding roll quality control device 102 inputs the second roll internal state calculation value of the winding roll calculated by the second algorithm calculation unit based on the second algorithm 122, and calculates the second learned value. It has a second inference unit (not shown, provided corresponding to the first trained AI model 123) that outputs second next step suggestion data 125 using the AI model 123.
Based on the internal state of the winding roll during transportation or storage, the second inference unit outputs second next-process suggestion data 125, which is data suggesting manufacturing parameters for the production line of the next process, and Manufacturing in the next step of the take-up roll can be optimally performed.

Furthermore, in the ninth embodiment of the present invention, the winding roll quality control device 102 inputs a second roll internal state calculation value for a winding roll with a problem in the market, A second process that learns the correlation between the internal state calculated value and the second cause analysis data that is the data that caused the defect including the transport storage data that is the environmental parameter in the transport or storage of the winding roll It has a second inference unit (not shown, provided corresponding to the second trained AI model 123) that outputs second cause analysis data 127 using the trained AI model 123.
This second reasoning unit can input the second roll internal state calculation value as feedback data from the market, output the second cause analysis data that caused the defect, and can output the second cause analysis data for the winding roll. Or, it becomes easier to analyze defects during storage.

FIG. 2 shows a conceptual diagram of a center-driven winding device having a nip roller 203, which is one of the basic winding methods. A web 201 such as a plastic film is wound around a core 202 while being applied with a winding tension Tw and a pressing load L by a nip roller 203. As the winding progresses, a new layer is added. The stress in the wound portion changes sequentially.
At the same time, due to the viscoelastic properties of the web 201 and the effect of thermal strain associated with changes in roll temperature, the internal stress changes every moment from the start of winding, during winding, and even after completion of winding.
Note that the nip roller 203 is not essential to the winding device, and can be omitted if an appropriate winding tension Tw can be obtained without the nip roller 203 .

Also, the web 201 is normally wound in the atmosphere. At that time, ambient air flows between the outermost layer during winding and the already wound portion due to its viscous action, as a result of which an air layer is formed between the web 201 in the winding roll.
If such an air layer exists, the apparent stiffness of the take-up roll 204 is significantly reduced, and the internal stress changes greatly. On the other hand, the nip roller 203 is used for the purpose of limiting the amount of entrained air and adjusting the apparent rigidity of the take-up roll 204 .

The first algorithm will be described below with respect to the conceptual diagram of the winding roll in FIG.
Here, when formulating the winding model, (1) the winding roll maintains a perfect cylindrical shape, and the thickness, width, surface roughness, etc. of the web are uniform, and (2) the web is spirally wound. (3) Radial stress and circumferential stress are dominant inside the winding roll, and axial stress is considered. Make an assumption that it is not necessary.

・・・（１）
Ｅ_θ：ウェブと空気層を複合した等価層としたときの円周方向のヤング率（Ｐａ）
Ｅ_c：巻芯のヤング率（Ｐａ）
ν：ウェブのポアソン比
ｒ_c：前記巻芯の半径（ｍ）

・・・（２）
ｒ_out：巻取りロールの最外層半径（ｍ）

・・・（３）
ｒ：巻取りロールの半径（ｍ）
Ｅ_r：ウェブと空気層を複合した等価層における半径方向のヤング率（Ｐａ）
First, the calculation of the first radial stress _σr , which is one of the first internal state calculation values in the radial direction, by the first algorithm when manufacturing the winding roll is as follows.
The radial stress increment Δσ _r in the winding roll due to winding a new web on the outermost layer of the winding roll is expressed by the following equation (1) at the boundary of the innermost layer of the winding roll. It is calculated based on the conditions, the boundary condition of the outermost layer represented by the following equation (2), and the following basic equation (3).

... (1)
E _θ : Young's modulus (Pa) in the circumferential direction when the equivalent layer is a composite of the web and the air layer
E _c : Young's modulus of winding core (Pa)
ν: Poisson's ratio of the web r _c : Radius of the winding core (m)

... (2)
r _out : outermost layer radius of winding roll (m)

... (3)
r: Radius of winding roll (m)
E _r : Young's modulus (Pa) in the radial direction in the equivalent layer combining the web and air layer

・・・（４）
σ_r,j.i：第ｉ層まで巻き取った巻取りロールにおける第j層の半径方向応力（Ｐａ）
Then, the first radial stress _{σ r,ji} in the j-th layer in the winding roll wound up to the i-th layer, which is one of the first internal state calculation values in the radial direction by the first algorithm The calculation is as follows.

... (4)
σ _r,ji : Radial stress (Pa) of the j-th layer in the take-up roll wound up to the i-th layer

・・・（５）
Further, when the winding roll is manufactured, the calculation of the first circumferential stress σ _θ , which is one of the first internal state calculation values in the first algorithm, is given by the following equation (5) as σ _r can be calculated by substituting the calculated values of and r.

... (5)

・・・（６）
ｒ_out：巻取りロールの最外層半径（ｍ）
η：空気の粘度（Ｐａ・ｓ）
Ｖ：巻取り速度（ｍ／ｓ）
Ｔ_W：巻取り張力（Ｎ／ｍ）
λ：ウェブの幅Ｗ（ｍ）に関する無次元パラメータであり以下の式（７）で定義される

・・・（７）
そして、時間増分Δｔ_iにおける第i層のウェブにおける解析対象の空気層 の厚さｈ_a.i（ｍ）を、以下の式（８）に基づいて算出する。

・・・（８）
Next, at the time of manufacturing the winding roll, regarding the calculation of the first air layer thickness _ha , which is one of the first internal state calculation values in the first algorithm, the winding roll and the newly wound Calculate the thickness h _a0 (m) of the initial air layer formed between the web layer and the web layer below based on the following equation (6), and calculate the thickness of the air layer to be analyzed in the i-th web layer h _ai (m) can be calculated based on Equation (8) below.

... (6)
r _out : outermost layer radius of winding roll (m)
η: Viscosity of air (Pa s)
V: Winding speed (m/s)
T _W : Winding tension (N/m)
λ: a dimensionless parameter for the web width W (m) defined by the following equation (7)

... (7)
Then, the thickness h _ai (m) of the air layer to be analyzed in the web of the i-th layer at the time increment Δt _i is calculated based on the following equation (8).

... (8)

Next, the second algorithm will be described below with respect to the conceptual diagram of the winding roll in FIG.
Here, when formulating the winding model, (1) the winding roll maintains a perfect cylindrical shape, and the thickness, width, surface roughness, etc. of the web are uniform, and (2) the web is spirally wound. (3) Radial stress and circumferential stress are dominant inside the winding roll, and axial stress is considered. It is the same as the first algorithm that assumes that it is not necessary.

・・・（９）
Ｊ_θ（Δｔ_i）：ウェブの円周方向クリープ関数（１／Ｐａ）
ｒ：巻取りロールの半径（ｍ）
α_θ：ウェブの円周方向膨張係数（１／℃）
Ｔ_f.k（ｔ_i）：巻取りロールの第k番目のウェブ層の時間ｔ_iにおけるウェブの温度（℃）
ΔＴ_f.k（ｔ_i）：巻取りロールの第k番目のウェブ層の時間ｔ_iにおけるウェブの温度増分（℃）
ΔＴ_c（ｔ_i）：時間ｔ_iにおける巻芯の温度増分（℃）
ｒ_c：巻芯の半径（ｍ）
Ｅ_c：巻芯のヤング率（Ｐａ）
α_c：巻芯の線膨張係数（１／℃）

・・・（１０）
ｒ_out：巻取りロールの最外層半径（ｍ）

・・・（１１）
α_ｒ：ウェブの半径方向線膨張係数（１／℃）
Ｊ_ｒ（Δｔ_i）：時間増分Δtにおけるウェブの半径方向クリープ関数（１／Ｐａ）

・・・（１２）
First, the calculation of the second radial stress _σr , which is one of the second internal state calculated values in the radial direction by the second algorithm during transport or storage of the winding roll, is as follows.
By using the innermost layer boundary condition equation (9) and the outermost layer boundary condition equation (10) in the following basic equation (11) regarding the radial stress increment of the winding roll, the winding at the time increment Δt _i Determine the stress increment Δσ _rk (t _i ) occurring in the kth layer of the roll and apply the determined stress increment Δσ _{r .k} (t _i ) to equation (12) to obtain the radial stress σ at time t _i We can find _{r .k} (t _i ).

... (9)
J _θ (Δt _i ): Circumferential creep function of the web (1/Pa)
r: Radius of winding roll (m)
α _θ : Web expansion coefficient in the circumferential direction (1/°C)
T _fk (t _i ): web temperature (°C) at time t _i of the kth web layer of the winding roll
ΔT _fk (t _i ): web temperature increment at time t _i of the kth web layer of the take-up roll (°C)
ΔT _c (t _i ): Core temperature increment at time t _i (°C)
r _c : core radius (m)
E _c : Young's modulus of winding core (Pa)
α _c : Coefficient of linear expansion of winding core (1/°C)

(10)
r _out : outermost layer radius of winding roll (m)

(11)
α _r : Web radial expansion coefficient (1/° C.)
J _r (Δt _i ): web radial creep function (1/Pa) at time increment Δt

(12)

Also, during transportation or storage of the winding roll, the calculation of the second circumferential stress σ _θ , which is one of the second internal state calculation values in the second algorithm, is performed using the formula ( Since the calculation can be performed in the same manner as the calculation using 5), the explanation thereof is omitted here.

Next, when transporting or storing the winding roll, regarding the calculation of the second air layer thickness _ha , which is one of the second internal state calculation values in the circumferential direction by the second algorithm, the first Since the calculation can be performed in the same manner as the calculation using the formula (8) in the algorithm of (1), the explanation thereof is omitted here.

FIG. 3 illustrates the operation of the take-up roll quality control device, the take-up roll quality control method, or the take-up roll quality control program according to the first embodiment of the present embodiment over time. 10 is a flowchart for calculating roll internal state calculation values at the time of manufacture using algorithm 1. FIG.

The operation of the quality control device starts from the start (S301).
First, prior data 111, which are physical property parameters at the time of manufacture of the take-up roll, are input to the quality control device 102 in FIG. 1 (S302).
Also, in the winding roll, a variable n (n is a positive integer) that designates the layer number for which the calculated value of the internal state of the winding roll is to be initialized to 1 (S303).
When the quality control device 102 is connected to the winding device 101, the in-manufacturing data 112 from the winding device 101 during manufacturing of the winding roll is sent to the first algorithm calculation unit of the quality control device 102 in real time. It is input (S304).
Next, the first algorithm calculation unit calculates the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113 in real time (S305).

After completing the calculation of the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113, it is determined whether the winding of the roll is completed (S306).
In step S306, if it is determined that the winding of the roll has not been completed (in the case of NO), the variable n specifying the layer number is incremented by 1, the process moves to step S304, and the steps after step S304 are performed for winding the roll. This is executed until the acquisition is completed (S307).
Alternatively, if it is determined in step S306 that the winding of the roll has been completed (if YES), the operation of the quality control device 102 ends (S308).

FIG. 4 shows the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the second embodiment of the present embodiment. 1 is a flowchart for calculating roll internal state calculation values at the time of manufacture using algorithm No. 1;

The operation of the quality control device starts from the start (S401).
First, prior data 111, which are physical property parameters at the time of manufacture of the take-up roll, are input to the quality control device 102 in FIG. 1 (S402).
Also, in the winding roll, a variable n (n is a positive integer) that designates the layer number for which the calculated value of the internal state of the winding roll is to be initialized to 1 (S403).
When the quality control device 102 is connected to the winding device 101, the in-manufacturing data 112 from the winding device 101 during manufacturing of the winding roll is sent to the first algorithm calculation unit of the quality control device 102 in real time. It is input (S404).
Next, the first algorithm calculation unit calculates the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113 in real time (S405).
After the first roll internal state calculated value is calculated, the first roll internal state calculated value calculated by the first algorithm calculation unit and the first pass/fail judgment result regarding the manufacture of the winding roll The first untrained AI model is trained to generate the first trained AI model 114 (S406).

After completing the calculation of the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113, it is determined whether the winding of the roll is completed (S407).
If it is determined in step S407 that the winding of the roll has not been completed (in the case of NO), the variable n specifying the layer number is incremented by 1, and step S404 and subsequent steps are executed until the winding of the roll is completed. (S408).
Alternatively, if it is determined in step S407 that the winding of the roll has been completed (if YES), the operation of the quality control device 102 ends and the generation of the first learned AI model 114 is completed (S409). .

FIG. 5 shows the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the third embodiment of the present embodiment. 10 is a flow chart in the case of making the first AI model learn the roll internal state calculation value at the time of manufacturing using the algorithm No. 1. FIG.

The operation of the quality control device starts from the start (S501).
First, prior data 111, which are physical property parameters at the time of manufacture of the take-up roll, are input to the quality control device 102 in FIG. 1 (S502).
Also, in the winding roll, a variable n (n is a positive integer) that designates the layer number for which the calculated value of the internal state of the winding roll is to be initialized to 1 (S503).
When the quality control device 102 is connected to the winding device 101, the in-manufacturing data 112 from the winding device 101 during manufacturing of the winding roll is sent to the first algorithm calculation unit of the quality control device 102 in real time. It is input (S504).
Next, the first algorithm calculation unit calculates the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113 in real time (S505).
After the first roll internal state calculated value is calculated, the first roll internal state calculated value calculated by the first algorithm calculation unit and the first pass/fail judgment result regarding the manufacture of the winding roll Using the first trained AI model 114 that has learned the correlation between (S506).

As a result of inference with respect to the first roll internal state calculated value by the first learned AI model 114 in step S506, it is determined whether or not the product quality of the winding roll is normal (S507).
In step S507, if it is inferred that the product quality of the winding roll is normal (if YES), the production data is reflected in the data during production, which is the operating parameter of the winding device, so that the production line is fed back (S508).
Alternatively, when it is inferred that the product quality of the take-up roll is abnormal in step S507 (in the case of NO), an error is output as the first notification output 115. FIG. Furthermore, when the error is output, the production line is automatically stopped (S509).

After completing the calculation of the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113, it is determined whether the winding of the roll is completed (S510).
If it is determined in step S510 that the winding of the roll has not been completed (in the case of NO), the variable n specifying the layer number is incremented by 1, and step S504 and subsequent steps are executed until the winding of the roll is completed. (S511).
Alternatively, if it is determined in step S510 that the winding of the roll has been completed (in the case of YES), the operation of the quality control device 102 ends (S512).

FIG. 6 illustrates the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the fourth embodiment of the present embodiment. 10 is a flow chart in the case of causing a first AI model to infer suggested data for the next process during manufacturing using algorithm No. 1. FIG.

The operation of the quality control device starts from the start (S601).
First, prior data 111, which are physical property parameters at the time of manufacture of the take-up roll, are input to the quality control device 102 in FIG. 1 (S602).
Also, in the winding roll, a variable n (n is a positive integer) that designates the layer number for which the calculated value of the internal state of the winding roll is to be initialized to 1 (S603).
When the quality control device 102 is connected to the winding device 101, the in-manufacturing data 112 from the winding device 101 during manufacturing of the winding roll is sent to the first algorithm calculation unit of the quality control device 102 in real time. It is input (S604).
Next, the first algorithm calculation unit calculates the first roll internal state calculation values for the first layer to the n-th layer using the first algorithm 113 in real time (S605).
After the first roll internal state calculated value is calculated, the first roll internal state calculated value calculated by the first algorithm calculation unit and the first next process suggestion data which is data suggesting the manufacturing parameter of the next process Using the first trained AI model 114 that has learned the correlation between Inference is made in real time, and the first next step suggestion data 116 is output (S606).

Next, in step S607, it is determined whether or not the first next process suggestion data 116 inferred by the first inference unit belongs to a predetermined range (S607).
In step S607, when it is determined that the first next-process suggestion data 116 belongs to the predetermined range (if YES), the first next-process suggestion data 116 is changed to the manufacturing data of the next process in the production line. By reflecting, feedback to the production line of the next process is performed (S608).
Alternatively, if it is determined in step S607 that the first next-step suggestive data 116 does not fall within the predetermined range (NO), an error is output as the first notification output 115. FIG. Furthermore, when the error is output, the production line is automatically stopped (S609).

After completing the inference of the first next-process suggestive data 116 from the first layer to the n-th layer using the first algorithm 113, it is determined whether winding of the roll is completed (S610).
If it is determined in step S610 that the winding of the roll has not been completed (in the case of NO), the variable n specifying the layer number is incremented by 1, and step S604 and subsequent steps are executed until the winding of the roll is completed. (S611).
Alternatively, if it is determined in step S610 that the winding of the roll has been completed (in the case of YES), the operation of the quality control device 102 ends (S612).

FIG. 7 illustrates the operation of the take-up roll quality control device, the take-up roll quality control method, or the take-up roll quality control program according to the fifth embodiment of the present embodiment. 10 is a flow chart in the case of causing the first AI model to infer suggested data for the next process during manufacturing using the algorithm No. 1. FIG.

The operation of the quality control device starts from the start (S701).
First, regarding the winding roll that has caused a problem in the market, the winding device 101 is provided with the first roll internal state calculation value as market feedback data, the advance data 111 that is the physical property parameter at the time of manufacturing the winding roll, or / and , manufacturing data 112 and the like from the winding device 101 at the time of manufacture of the winding roll are input (S702).

Next, the quality control device 102 inputs the first roll internal state calculation value as market feedback data for the winding roll in which the defect occurred in the market, and the first inference unit calculates the first roll internal state Data that caused the problem, including calculated values, data during production, which are the operating parameters of the winding device necessary for manufacturing the winding roll, and prior data, which are the physical property parameters of the winding roll. First causal analysis data 119 is output using the first trained AI model 114 that has learned the correlation with one causal analysis data (S703).
After outputting the first cause analysis data 119 by the first inference unit, the operation of the quality control device 102 ends, and the output first cause analysis data 119 is (S704).

FIG. 8 illustrates the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the sixth embodiment of the present embodiment. 2 is a flow chart in the case of making the second AI model learn the roll internal state calculation value during transportation and storage using the algorithm No. 2. FIG.

The operation of the quality control device starts from the start (S801).
First, the internal state of the take-up roll on which the web has been taken up to the n-th layer calculated by the first algorithm 113 is input. (S802)
Also, a variable s (s is 0 or a positive integer) that specifies the step number for which the calculated value of the internal state of the winding roll is to be initialized to 0 (S803).
Then, the transportation and storage data 121 of the first layer and the n-th layer of the winding roll at the time of transportation or storage of the winding roll are input to the second algorithm calculation unit of the quality control device 102 (S804).
Next, the second algorithm calculation unit calculates second roll internal state calculation values for the first to n-th layers using the second algorithm 122 (S805).
After the second roll internal state calculated value is calculated, the second roll internal state calculated value calculated by the second algorithm calculation unit and the result of pass/fail judgment regarding the transportation or storage of the winding roll The correlation between the pass/fail judgment result is learned by the second pre-learning AI model to generate the second trained AI model 123 (S806).

In step S807, s is the final step number SMAX (SMAX is a positive integer ) is determined (S807).
If it is determined in step S807 that s is not equal to the final number of steps SMAX (in the case of NO), the variable s specifying the step number is incremented by 1, and step S804 and subsequent steps are repeated until s reaches the final number of steps SMAX. Execute (S808).
Alternatively, if it is determined in step S807 that s is equal to the final number of steps SMAX (if YES), the operation of the quality control device 102 ends and generation of the second trained AI model 123 is completed (S809 ).

FIG. 9 shows the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the seventh embodiment of the present embodiment. 2 is a flow chart in the case of making the second AI model learn the roll internal state calculation value during transportation and storage using the algorithm No. 2. FIG.

The operation of the quality control device starts from the start (S901).
First, the internal state of the take-up roll on which the web has been taken up to the n-th layer calculated by the first algorithm 113 is input. (S902)
Also, a variable s (s is 0 or a positive integer) that specifies the step number for which the calculated value of the internal state of the winding roll is to be initialized to 0 (S903).
Then, the transportation and storage data 121 of the first layer and the n-th layer of the winding roll at the time of transportation or storage of the winding roll are input to the second algorithm calculating section of the quality control device 102 (S904).
Next, the second algorithm calculation unit calculates second roll internal state calculation values for the first to n-th layers using the second algorithm 122 (S905).
After the second roll internal state calculated value is calculated, the second roll internal state calculated value calculated by the second algorithm calculation unit and the second pass/fail result of pass/fail judgment regarding transport or storage of the winding roll The second inference unit infers the second roll internal state calculated value calculated by the second algorithm calculation unit using the second trained AI model 123 that has learned the correlation between the judgment result and the (S906).

As a result of inference with respect to the second roll internal state calculated value by the second learned AI model 123 in step S906, it is determined whether or not the product quality of the winding roll is normal (S907).
If it is inferred in step S907 that the product quality of the take-up roll is normal (if YES), it is determined whether s is equal to the maximum layer number final step number SMAX (SMAX is a positive integer) (S908 ).
Alternatively, if it is inferred in step S907 that the product quality of the winding roll is abnormal (in the case of NO), an error is output as the second notification output 124 (S909).
If it is determined in step S908 that s is not equal to the maximum number of layers final step number SMAX (in the case of NO), the variable s specifying the step number is incremented by 1, and steps S904 and subsequent steps are performed so that s is the final step number SMAX. (S910).
If it is determined in step S908 that s is equal to the final number of steps SMAX (if YES), the operation of quality control device 102 ends (S911).

FIG. 10 illustrates the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the eighth embodiment of the present embodiment. 2 is a flow chart for inferring a roll internal state calculated value during transportation and storage to a second AI model using algorithm No. 2. FIG.

The operation of the quality control device starts from the start (S1001).
First, the internal state of the take-up roll on which the web has been taken up to the n-th layer calculated by the first algorithm 113 is input. (S1002)
, a variable s (s is 0 or a positive integer) that designates the step number for which the calculated value of the internal state of the winding roll is to be initialized to 0 (S1003).
Then, the transportation and storage data 121 of the first layer and the n-th layer of the winding roll at the time of transportation or storage of the winding roll are input to the second algorithm calculation unit of the quality control device 102 (S1004).
Next, the second algorithm calculation unit calculates second roll internal state calculation values from the first layer to the n-th layer using the second algorithm 122 (S1005).
After the second roll internal state calculated value is calculated, the second roll internal state calculated value calculated by the second algorithm calculation unit and the second next process suggestion which is data suggesting the transportation and storage parameters of the next process Using the second trained AI model 123 that has learned the correlation between the data 125, the second reasoning unit generates the second next step suggestion data 125 based on the second roll internal state calculation value is inferred, and the second next step suggestion data 125 is output (S1006).

Next, in step S1006, it is determined whether or not the second next process suggestion data 125 inferred by the second inference section belongs to a predetermined range (S1007).
In step S1007, when it is determined that the second next-process suggestion data 125 belongs to the predetermined range (if YES), the second next-process suggestion data 125 is transferred to the next-process transport/storage process in the transport or storage process. Feedback to the transportation and storage process is performed by reflecting it in the data (S1008).
Alternatively, if it is determined in step S1007 that the second next step suggestion data 125 does not fall within the predetermined range (NO), an error is output as the second notification output 124 (S1010).

After completing the execution of the inference of the second next step suggestion data 125 from the first layer to the n-th layer using the second algorithm 122, s is the maximum layer number final step number (SMAX is a positive integer) It is determined whether they are equal (S1008).
If it is determined in step S1009 that s is not equal to the final number of steps SMAX (if NO), the variable s specifying the step number is incremented by 1, and step S1004 and subsequent steps are repeated until s reaches the final number of steps SMAX. Execute (S1011).
If it is determined in step S1009 that s is equal to the final number of steps SMAX (if YES), the operation of quality control device 102 ends (S1012).

FIG. 11 is a diagram illustrating the operation of the winding roll quality control device, the winding roll quality control method, or the winding roll quality control program according to the ninth embodiment of the present embodiment. FIG. 10 is a flow chart in the case of inferring data during transportation and storage that causes a problem using feedback data using a second AI model; FIG.

The operation of the quality control device starts from the start (S1101).
First, the quality control device 102 receives a second roll internal state calculation value as market feedback data for a winding roll that has a problem in the market (S1102).

Next, the second inference unit learns the correlation between the second roll internal state calculated value and the second cause analysis data, which is the data that caused the defect including the transportation and storage data. The second cause analysis data 127 is output using the second learned AI model 123 (S1103).
After the second reasoning unit outputs the second cause analysis data 127, the operation of the quality control device 102 ends, and the output second cause analysis data 127 is (S1104).

As described in detail above, according to the present embodiment, in the technical field of web handling, regarding the winding roll being manufactured in the manufacturing process of the winding roll, the roll internal state calculation, which is the calculated value of the internal state of the winding roll By outputting the value in real time and inferring various data using the generated learned AI model, the quality control of the winding roll can be performed.

101 winding device 102 quality control device 113 first algorithm 114 first AI model 122 second algorithm 123 second AI model 201 web 202 winding core 203 nip roller 204 winding roll

Claims (25)

In a winding roll quality control device that performs quality control of a winding roll,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. A winding having a first algorithm calculation unit for inputting certain manufacturing data and calculating a first roll internal state calculation value, which is a calculation value of the internal state of the winding roll, based on a first algorithm. Roll quality control equipment. The winding roll quality control device further includes:
A first trained AI model that has learned a correlation between the first roll internal state calculated value and a first pass/fail judgment result that is a pass/fail judgment result regarding manufacturing of the winding roll. 2. The winding roll quality control device according to claim 1, having a learning unit of . The winding roll quality control device further includes:
A first inference unit that inputs the first roll internal state calculation value and outputs a first pass/fail judgment result regarding manufacture of the winding roll using the first learned AI model. 3. The winding roll quality control device according to 1 or 2. In the winding roll quality control device,
The first inference unit
When the first pass/fail judgment result regarding the manufacture of the winding roll is inferred to be acceptable, the data during manufacture corresponding to the input calculated value of the internal state of the roll is transferred to the next operation of the winding device. 4. The take-up roll quality control device according to claim 3, wherein the parameters are reflected in manufacturing data. In the winding roll quality control device,
The first inference unit
5. The winding roll quality control device according to claim 3, wherein an error is output when the first acceptance/rejection judgment result regarding the manufacture of the winding roll is inferred to be rejected. In the winding roll quality control method for quality control of the winding roll,
the computer
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Enter some manufacturing data and
A winding roll quality control method for calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm. In the winding roll quality control program for quality control of the winding roll,
to the computer,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Input certain manufacturing data,
A winding roll quality control program for calculating a first roll internal state calculation value, which is a calculation value of the internal state of the winding roll, based on a first algorithm. In a winding roll quality control device that performs quality control of a winding roll,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. a first algorithm calculation unit for inputting certain manufacturing data and calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
A first learning unit that generates a first trained AI model that has learned a correlation between the first roll internal state calculated value and first suggestion data that is data that suggests a manufacturing parameter for the next process. A winding roll quality control device having The winding roll quality control device further includes:
9. The winding roll according to claim 8, further comprising a first inference unit that inputs the first roll internal state calculation value and outputs the first suggested data using the first trained AI model. quality control equipment. In the winding roll quality control method for quality control of the winding roll,
the computer
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Enter some manufacturing data and
calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
Winding roll quality for generating a first trained AI model that has learned a correlation between the first roll internal state calculated value and first suggested data that is data that suggests a manufacturing parameter for the next process. management device. In the winding roll quality control program for quality control of the winding roll,
to the computer,
Preliminary data, which are physical property parameters of the winding roll, and operation parameters of the winding device necessary for manufacturing the winding roll from an acquisition unit provided in the winding device for winding the winding roll. Input certain manufacturing data,
calculating a first roll internal state calculated value, which is a calculated value of the internal state of the winding roll, based on a first algorithm;
Winding roll quality for generating a first trained AI model that has learned a correlation between the first roll internal state calculated value and first suggested data that is data that suggests a manufacturing parameter for the next process. management program. In a winding roll quality control device that performs quality control of a winding roll,
A first calculated roll internal state value of the winding roll in which the defect has occurred is input, and the first calculated roll internal state calculation value and operating parameters of the winding device necessary for manufacturing the winding roll are obtained. A first learned AI that has learned the correlation between first cause analysis data that is data that caused the defect, including data during manufacturing and prior data that is the physical property parameter of the winding roll A winding roll quality control device having a first inference unit that outputs first cause analysis data using a model. In the winding roll quality control method for quality control of the winding roll,
the computer
inputting the first roll internal state calculation value of the winding roll in which the defect occurred,
Defects including the first roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and prior data that are physical property parameters of the winding roll A winding roll quality control method that outputs first cause analysis data using a first learned AI model that has learned the correlation between the first cause analysis data that is the data that caused the . In the winding roll quality control program for quality control of the winding roll,
to the computer,
Input the first roll internal state calculation value of the winding roll in which the problem occurred,
Defects including the first roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and prior data that are physical property parameters of the winding roll Winding roll quality control program that outputs the first cause analysis data using the first learned AI model that has learned the correlation between the first cause analysis data that is the data that caused the . In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. a second algorithm calculation unit for
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; A winding roll quality control device having a second learning unit. The winding roll quality control device further includes:
a second inference unit that inputs the second roll internal state calculated value and outputs a second pass/fail judgment result regarding transportation or storage of the winding roll using the second learned AI model; The winding roll quality control device according to claim 14 or 15. In the winding roll quality control method for quality control of the winding roll,
the computer
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. death,
generating a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll; Quality control method for winding rolls. In the winding roll quality control program for quality control of the winding roll,
to the computer,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. let
Generate a second trained AI model that has learned a correlation between the second roll internal state calculated value and a second pass/fail judgment result that is a pass/fail judgment result regarding transportation or storage of the winding roll. Quality control program for winding rolls. In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. a second algorithm calculation unit for
A second learning unit that generates a second trained AI model that has learned a correlation between the second roll internal state calculated value and second suggested data that is data that suggests a manufacturing parameter for the next process. A winding roll quality control device having The winding roll quality control device further includes:
20. The winding roll according to claim 19, further comprising a second inference unit that inputs the second roll internal state calculated value and outputs the second suggested data using the second trained AI model. quality control equipment. In the winding roll quality control method for quality control of the winding roll,
the computer
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. death,
Winding roll quality generating a second trained AI model that has learned the correlation between the second roll internal state calculated value and the second suggested data that is data that suggests the manufacturing parameters of the next process Management method. In the winding roll quality control program for quality control of the winding roll,
to the computer,
Transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll, are input, and a second roll internal state calculation value, which is a calculation value of the internal state of the winding roll, is calculated based on a second algorithm. let
Winding roll quality for generating a second trained AI model that has learned the correlation between the second roll internal state calculated value and the second suggested data that is data that suggests the manufacturing parameters of the next process. management program. In a winding roll quality control device that performs quality control of a winding roll,
Transportation and storage data, which are environmental parameters related to the transportation or storage of the winding roll, are input, and the second roll internal state calculated value and the operating parameters of the winding device necessary for manufacturing the winding roll are used. A second trained AI model that learns the correlation between first cause analysis data, which is data that caused the defect, including certain in-manufacturing data and prior data, which is the physical property parameter of the winding roll. A winding roll quality control device having a second inference unit that outputs second cause analysis data using In the winding roll quality control method for quality control of the winding roll,
the computer
inputting transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll;
Defects including the second roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and preliminary data that are physical property parameters of the winding roll A winding roll quality control method that outputs second cause analysis data using a second learned AI model that has learned the correlation between the first cause analysis data that is the data that caused the . In the winding roll quality control program for quality control of the winding roll,
to the computer,
input transportation and storage data, which are environmental parameters related to transportation or storage of the winding roll;
Defects including the second roll internal state calculated value, in-manufacturing data that are operating parameters of the winding device necessary for manufacturing the winding roll, and advance data that are physical property parameters of the winding roll A winding roll quality control program that outputs second cause analysis data using a second learned AI model that has learned the correlation between the first cause analysis data that is the data that caused the .

Images