JP2023008250A - Learning method, prediction model, temperature estimation method, temperature adjustment method, printing method, and printer - Google Patents

Abstract

To provide a technique capable of maintaining the temperature of ink in a head within a desired range even when a consumption amount of the ink discharged from the head varies.SOLUTION: A printer comprises a conveyance mechanism, processing units 21-24, a supply unit, a heating unit 40, a first measurement unit 51, a second measurement unit 52, and a learning unit 92. The processing unit discharges a processing material from a head to a surface of a base material. The supply unit supplies the processing material in an equivalent amount according to the amount of the processing material discharged from the head to the head. The heating unit heats the processing material supplied to the head. The first measurement unit acquires a first measurement value by measuring the amount or the equivalent amount of the processing material discharged from the head. The second measurement unit acquires a second measurement value by measuring the temperature of the processing material in the head. The learning unit machine-learns a learning model with the first measurement value as an input variable, the second measurement value after a prescribed time as teacher data, and the temperature of the processing material in the head as a target variable.SELECTED DRAWING: Figure 5

Description

TECHNICAL FIELD The present invention relates to a technique for estimating the temperature of a treatment substance in the head after a predetermined time while ejecting a treatment substance from a head onto the surface of the substrate while transporting a long belt-like substrate in the longitudinal direction.

2. Description of the Related Art Conventionally, there has been known a printing apparatus that discharges a treatment substance such as ink from a plurality of heads onto the surface of a base material while conveying a long belt-shaped base material in the longitudinal direction. In this type of printing apparatus, the viscosity of the treatment substance may increase when the temperature of the treatment substance in the head drops due to the low frequency of discharge of the treatment substance from the head. In this case, at the time of printing, there is a possibility that the ejection position of the treatment substance on the substrate may be shifted, or streaks may occur. Also, if the temperature of the substance to be processed inside the head rises excessively, there is a possibility that the substance to be processed will deteriorate. Techniques for controlling the temperature of ink before being ejected are therefore described in, for example, Japanese Patent Laid-Open Nos. 2003-100000 and 2000-200403.

Japanese Patent Application Laid-Open No. 2020-59129 Japanese Patent Application Laid-Open No. 2020-13671

The ink jet recording apparatus of Patent Document 1 has an ink circulation system that circulates the ink between the recording head and the ink tank in order to suppress thickening of the ink in the vicinity of the ejection port. Further, the ink jet recording apparatus of Patent Document 1 starts circulation of ink and heats the ink when a print job is received. Further, the ink jet printer disclosed in Patent Document 2 uses a thyristor to control power supplied to a rubber heater that heats ink conveyed to the ink jet head.

However, the temperature of the ink stored in each head before being ejected is irregular due to fluctuations in the consumption of ink ejected from each head and changes in the amount of newly stored ink accompanying the ejection. change to Therefore, even if the ink is heated uniformly, it is difficult to maintain the desired temperature.

The present invention has been made in view of such circumstances, and even if the consumption of ink ejected from each head fluctuates or the amount of ink newly stored along with the ejection changes, An object of the present invention is to provide a technique capable of maintaining the temperature of ink stored in each head or the like within a desired range.

In order to solve the above-mentioned problems, the first invention of the present application is a learning method, comprising: b) supplying a corresponding amount of the processing substance from a supply unit to the head according to the amount of the processing substance discharged from the head; c) from the supply unit heating the treatment substance supplied to the head; and d) the amount of the treatment substance ejected from the head in step a) or the equivalent amount of the treatment substance supplied in step b). and e) measuring the temperature of the material to be processed in the head to obtain a second measurement value, f) the first measurement a step of learning a learning model using a machine learning algorithm with the value as an input variable and the temperature of the material to be treated in the head at a time after a predetermined time from the measurement time of the first measured value as an objective variable; and, in the step f), using as teacher data the acquisition result of the second measured value at the time after the predetermined time from the measurement time of the first measured value, the teacher data and the output value from the learning model A plurality of parameters included in the learning model are adjusted and updated and stored so that the difference between is small.

A second invention of the present application is the learning method of the first invention, wherein in the step a), ink as the treatment substance is ejected from the head to record or print an image on the surface of the base material, In the step d), the amount of the treatment substance ejected from the head in the step a) is calculated from the print pattern or print amount included in the image recorded on the surface of the base material. 1 Get the measurement value.

A third invention of the present application is the learning method of the first invention or the second invention, comprising g) the step of measuring the temperature of the material to be treated supplied from the supply unit to the head to obtain a third measured value. and in the step f), learning processing is performed on the learning model while using the third measured value as an input variable.

A fourth invention of the present application is the learning method according to any one of the first invention to the third invention, further comprising the step of h) measuring the environmental temperature outside the head to obtain a fourth measured value. In the step f), learning processing is performed on the learning model while using the fourth measurement value as an input variable.

A fifth invention of the present application is a prediction model generated by the learning method of any one of the first to fourth inventions.

A sixth aspect of the invention of the present application is a method for estimating temperature, comprising: p) conveying a long strip-shaped substrate in the longitudinal direction along a predetermined conveying path, while applying the treatment substance from the head to the surface of the substrate; q) supplying the corresponding amount of the processing substance corresponding to the amount of the processing substance discharged from the head from the supply unit to the head; r) from the supply unit to the head s) the amount of the processing substance discharged from the head in the step p), or the amount of the processing substance supplied in the step q). a step of measuring a corresponding amount to obtain the first measured value; and t) entering the first measured value into the prediction model of the fifth invention as an input variable, thereby obtaining and calculating a predicted value of the temperature of the material to be treated.

A seventh aspect of the invention of the present application is a temperature adjustment method, in which feedback control of the heating unit is performed based on a difference between the predicted value calculated by the temperature estimation method of the sixth aspect and a preset target value.

The eighth invention of the present application is a printing method, wherein the steps p) to t) are performed in parallel, and the heating by the heating unit in the step r) is performed based on the temperature adjustment method of the seventh invention. to control.

A ninth invention of the present application is a printing apparatus, comprising: a transport mechanism for transporting a long strip-shaped base material in a longitudinal direction along a predetermined transport path; a processing unit that discharges a processing substance from a head to the head; a supply unit that supplies a considerable amount of the processing substance corresponding to the amount of the processing substance discharged from the head to the head; A heating unit that heats the supplied processing substance and the amount of the processing substance discharged from the head or the equivalent amount of the processing substance supplied from the supply unit to the head are measured, and a first a first measuring unit that acquires a measured value; a second measuring unit that measures the temperature of the material to be processed in the head and acquires a second measured value; a learning unit that performs learning processing using a machine learning algorithm on a learning model whose target variable is the temperature of the material to be treated in the head at a time after a predetermined time from the measurement time of the measurement value; and from the measurement time of the first measurement value. a head temperature prediction unit that calculates a predicted value of the temperature of the material to be processed in the head at a time after the predetermined time, wherein the learning unit calculates the temperature after the predetermined time from the time when the first measurement value is measured; While adjusting a plurality of parameters included in the learning model so that the difference between the teaching data and the output value from the learning model is small, using the acquisition result of the second measurement value at the time of By updating and saving, the learning model is learned.

A tenth invention of the present application is the printing apparatus according to the ninth invention, wherein the processing unit ejects ink as the processing substance from the head to record or print an image on the surface of the base material. and the first measurement unit calculates the amount of the treatment substance ejected from the head from the print pattern or print amount included in the image recorded on the surface of the base material, thereby calculating the first Get measurements.

An eleventh invention of the present application is the printing apparatus according to the ninth or tenth invention, wherein the temperature of the processing material supplied from the supply unit to the head is measured to obtain a third measurement value. a unit, wherein the learning model further uses the third measured value as an input variable.

A twelfth invention of the present application is the printing apparatus according to any one of the ninth to eleventh inventions, further comprising a fourth measuring unit for measuring the environmental temperature outside the head to obtain a fourth measured value. The learning model further has the fourth measured value as an input variable.

A thirteenth invention of the present application is the printing apparatus according to any one of the ninth invention to the twelfth invention, wherein the heating unit is fed back based on a difference between the predicted value and a preset target value. It further has a temperature controller for controlling.

According to the first to thirteenth inventions of the present application, by using a prediction model obtained by machine learning, it is possible to predict the temperature of the material to be processed inside the head after a predetermined period of time. By heating the processing substance supplied to the head based on the prediction result, the temperature of the processing substance in the head can be controlled with high accuracy. As a result, even if the consumption of the processing substance ejected from the head fluctuates or the amount of the processing substance newly stored along with the ejection changes, the temperature of the processing substance stored in each head can be kept as desired. can be maintained in the range of

1 is a diagram conceptually showing the configuration of a printing apparatus; FIG. 4 is a partial top view of the printing device in the vicinity of the printing section; FIG. 1 is a diagram conceptually showing a configuration of part of a printing apparatus; FIG. 2 is a block diagram showing connections between the computer and each part of the printing apparatus; FIG. 2 is a block diagram conceptually showing functions in a computer; FIG. 4 is a flowchart showing the flow of learning processing; FIG. 2 is a diagram conceptually showing the configuration of a part of a printing apparatus when performing print processing; 4 is a flowchart showing the flow of print processing;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<1. First Embodiment>
<1-1. Configuration of Printing Apparatus>
FIG. 1 is a diagram conceptually showing the configuration of a printing apparatus 1 according to an embodiment of the invention. The printing apparatus 1 records an image on the surface of the base material 9 by ejecting ink droplets from a plurality of heads 21 to 24 toward the base material 9 while conveying the base material 9 having a long strip shape. It is an inkjet type device that prints on the surface of the substrate 9. The base material 9 may be printing paper, or may be a resin film. Also, the substrate 9 may be a substrate made of cardboard, metal foil, or glass.

As shown in FIG. 1, the printing apparatus 1 includes a transport mechanism 10, a printing unit 20, a supply unit 30, a heating unit 40, a plurality of first measuring units 51, a plurality of second measuring units 52, and A plurality of third measurement units 53 , a fourth measurement unit 54 , an encoder 60 , a camera 70 , a plurality of various sensors 80 , and a computer 90 are provided.

The transport mechanism 10 is a mechanism that transports the base material 9 along a predetermined transport path in the transport direction along the longitudinal direction thereof. The conveying mechanism 10 of this embodiment has an unwinding section 11 , a plurality of conveying rollers 12 , and a winding section 13 . The base material 9 is stretched over the plurality of rollers. Further, the base material 9 is unwound from the unwinding section 11 and conveyed along a conveying path composed of a plurality of conveying rollers 12 . Each transport roller 12 guides the substrate 9 to the downstream side of the transport path by rotating around an axis extending in a direction perpendicular to the transport direction. The substrate 9 after transport is recovered to the winding section 13 . In addition, tension is applied to the base material 9 in the transport direction. This suppresses slackness and wrinkles of the base material 9 during transportation.

The transport mechanism 10 further has a motor 14 that rotates some rollers (hereinafter referred to as "drive rollers"). The transport mechanism 10 may include multiple motors 14 . Motor 14 is electrically connected to computer 90 . When the motor 14 is driven, a command value for rotationally driving the motor 14 is input from a motor driving section 95 of the computer 90, which will be described later. Then, the motor 14 is driven according to the command value, and the drive roller rotates. As a result, the base material 9 is conveyed from the unwinding section 11 toward the winding section 13 .

The printing unit 20 is a processing unit that ejects ink droplets (hereinafter referred to as “ink droplets”) onto the substrate 9 conveyed by the conveying mechanism 10 . The printing unit 20 of this embodiment has a first head 21 , a second head 22 , a third head 23 and a fourth head 24 . The first head 21, the second head 22, the third head 23, and the fourth head 24 of this embodiment have the same structure. Also, the first head 21 , the second head 22 , the third head 23 , and the fourth head 24 are arranged at intervals along the conveying direction of the substrate 9 . The substrate 9 moves substantially parallel to the arrangement direction of the plurality of heads 21-24 under the plurality of heads 21-24. At this time, the surface (printing surface) of the substrate 9 faces upward (toward the heads 21 to 24).

Also, each of the first head 21 , the second head 22 , the third head 23 , and the fourth head 24 has an internal space capable of storing ink and a plurality of nozzles 201 . FIG. 2 is a partial top view of the printing apparatus 1 in the vicinity of the printing unit 20. FIG. As indicated by broken lines in FIG. 2, a plurality of nozzles 201 are arranged in parallel with the width direction of the substrate 9 on the lower surfaces of the heads 21 to 24 . Further, each of the plurality of nozzles 201 has a piezo element as a pressure generating element (not shown) and an ejection port communicating with the internal space of each of the heads 21-24. When the ink is ejected, the ink flows down from the internal space to the vicinity of the ejection port, and the ink in the vicinity of the ejection port is pressurized by the action of the piezoelectric element, and the ink is ejected from the ejection port as droplets. However, the nozzle 201 may be of a so-called thermal system in which a heater is used as a pressure generating element to heat the ink in the vicinity of the ejection port to generate bubbles.

Each of the heads 21 to 24 directs from a plurality of nozzles 201 toward the upper surface of the substrate 9, each of K (black), C (cyan), M (magenta), and Y (yellow), which are color components of a multicolor image. of ink droplets are respectively ejected. That is, the first head 21 ejects K-color ink droplets as a processing substance onto the upper surface of the substrate 9 at the first printing position P1, which is the processing position on the transport path. The second head 22 ejects C-color ink droplets as a processing substance onto the upper surface of the substrate 9 at a second printing position P2, which is a processing position downstream of the first printing position P1. The third head 23 ejects M-color ink droplets as a processing substance onto the upper surface of the substrate 9 at a third printing position P3, which is a processing position downstream of the second printing position P2. The fourth head 24 ejects Y-color ink droplets as a processing substance onto the upper surface of the substrate 9 at a fourth printing position P4, which is a processing position downstream of the third printing position P3.

In this embodiment, the first printing position P1, the second printing position P2, the third printing position P3, and the fourth printing position P4 are arranged at equal intervals along the conveying direction of the base material 9. FIG. The four heads 21 to 24 respectively record monochromatic images on the top surface of the substrate 9 by ejecting ink droplets. A multicolor image is formed on the upper surface of the substrate 9 by superimposing the four monochromatic images. Further, each of the heads 21 to 24 can print on the upper surface of the substrate 9 by ejecting ink droplets.

The supply unit 30 is a unit that supplies ink as a processing substance to the four heads 21-24. FIG. 3 is a diagram conceptually showing a configuration of part of the printing apparatus 1. As shown in FIG. As shown in FIG. 3, the supply unit 30 includes a first tank 311, a second tank 312, a third tank 313, and a fourth tank 314, a first pipe 321, a second pipe 322, a third pipe 323, and It has a fourth pipe 324 , a first pump 331 , a second pump 332 , a third pump 333 and a fourth pump 334 , and four flowmeters 34 .

K-color ink is stored in the first tank 311 in advance. C-color ink is stored in advance in the second tank 312 . M-color ink is stored in advance in the third tank 313 . The fourth tank 314 stores yellow ink in advance.

One end of the first pipe 321 is connected to the internal space of the first head 21 . The other end of the first pipe 321 is connected to the first tank 311 . Further, an electromagnetic valve and a degassing module (not shown) and a first pump 331 are inserted in the first pipe 321 of the present embodiment. The electromagnetic valve switches opening and closing of the ink flow path in the first pipe 321 . The first pump 331 generates a flow of ink in the first pipe 321 from the first tank 311 toward the first head 21 . A pump equipped with a brushless motor, for example, is used for the first pump 331 . Ink is supplied from the first tank 311 to the first head 21 when the first pump 331 is driven while the ink flow path in the first pipe 321 is opened by switching the electromagnetic valve.

When the K-color ink droplets are ejected from the first head 21 toward the substrate 9, the amount of K-color ink in the first head 21 decreases. At this time, a command value for driving the first pump 331 is input from the later-described pump control section 97 of the computer 90 . As a result, the first pump 331 is driven according to the command value, and an amount of ink corresponding to the decrease in the amount of K-color ink in the first head 21 is supplied from the first tank 311 to the first head 21 . be. That is, the supply unit 30 supplies the first head 21 with an amount of K-color ink corresponding to the amount of K-color ink ejected from the first head 21 . The K-color ink supplied from the first tank 311 to the first head 21 is deaerated by the deaeration module on the way.

Similarly, by driving the second pump 332 , C-color ink is supplied from the second tank 312 to the second head 22 . The supply unit 30 supplies the second head 22 with an amount of C-color ink corresponding to the amount of C-color ink ejected from the second head 22 . Similarly, by driving the third pump 333 , M-color ink is supplied from the third tank 313 to the third head 23 . The supply unit 30 supplies the third head 23 with an amount of M-color ink corresponding to the amount of the M-color ink ejected from the third head 23 . Similarly, by driving the fourth pump 334 , Y ink is supplied from the fourth tank 314 to the fourth head 24 . The supply unit 30 supplies the fourth head 24 with an amount of Y-color ink corresponding to the amount of the Y-color ink ejected from the fourth head 24 .

Furthermore, a flow meter 34 is interposed in each of the four pipes 321 to 324 of this embodiment. The flowmeter 34 measures the flow rate of ink flowing through each of the pipes 321-324. That is, the flow meter 34 measures the equivalent amount of each color ink supplied from the supply unit 30 to each of the heads 21 to 24, and acquires the measurement value Mf as the measurement result. The flowmeter 34 is also electrically connected to the computer 90 . The flowmeter 34 then outputs data relating to the measured value Mf to the computer 90 .

The heating unit 40 heats each color ink supplied from the supply unit 30 to each of the heads 21 to 24 . A heater, for example, is used for the heating unit 40 . Further, the heating unit 40 of this embodiment has a configuration for heating each color ink stored in each of the tanks 311 to 314 . However, the heating unit 40 may be configured to heat each color of ink flowing through each of the pipes 321-324. Further, the heating section 40 may have a configuration for heating each color of ink near each of the heads 21 to 24 . Furthermore, each of the pipes 321 to 324 may be provided with a reservoir for temporarily storing the ink of each color, and the heating unit 40 may have a configuration for heating the ink of each color stored in the reservoir. . Further, the printing apparatus 1 may have a plurality of heating units 40 provided at these plurality of locations.

A first measurement unit 51 is mounted on each of the four heads 21 to 24 . The first measurement unit 51 is configured by, for example, an electric circuit board. As will be described later, each of the heads 21 to 24 is controlled by a later-described print instruction unit 96 of the computer 90, and records an image on the surface of the base material 9 by ejecting ink droplets onto the top surface of the base material 9, and The surface of the base material 9 is printed. The first measuring unit 51 mounted on the first head 21 measures the amount of the K-color ink as the processing substance ejected from the first head 21 according to the print pattern included in the image recorded on the surface of the substrate 9. and the amount of printing, and obtain a first measurement value Me1 as a calculation result. Also, the first measuring section 51 is electrically connected to the computer 90 . The first measurement unit 51 outputs data related to the first measurement value Me1 to the computer 90 .

Similarly, the first measuring unit 51 mounted on the second head 22 includes the amount of C-color ink as the processing substance ejected from the second head 22 in the image recorded on the surface of the substrate 9. The first measurement value Me1 is calculated from the printed pattern or the printed amount, and the data related to the first measurement value Me1 as the calculation result is output to the computer 90. In addition, the first measurement unit 51 mounted on the third head 23 measures the amount of the M-color ink as the processing substance ejected from the third head 23 so that it is included in the image recorded on the surface of the substrate 9. It is calculated from the print pattern or the print amount, and the data related to the first measurement value Me1 as the calculation result is output to the computer 90 . In addition, the first measuring unit 51 mounted on the fourth head 24 measures the amount of the Y ink as the processing substance ejected from the fourth head 24 so that it is included in the image recorded on the surface of the substrate 9. It is calculated from the print pattern or the print amount, and the data related to the first measurement value Me1 as the calculation result is output to the computer 90 .

A second measurement unit 52 is mounted on each of the four heads 21 to 24 . The second measurement unit 52 is configured by, for example, a temperature sensor. The second measurement unit 52 mounted on the first head 21 measures the temperature of the K-color ink in the first head 21 and acquires a second measurement value Mt2 as the measurement result. Also, the second measuring unit 52 is electrically connected to the computer 90 . The second measuring section 52 outputs data relating to the second measured value Mt2 to the computer 90 .

Similarly, the second measurement unit 52 mounted on the second head 22 measures the temperature of the C-color ink in the second head 22, and outputs data related to the second measurement value Mt2 as the measurement result to the computer 90. Output to Further, the second measurement unit 52 mounted on the third head 23 measures the temperature of the M-color ink in the third head 23, and transmits data related to the second measurement value Mt2 as the measurement result to the computer 90. Output. Further, the second measurement unit 52 mounted on the fourth head 24 measures the temperature of the Y-color ink in the fourth head 24, and transmits data relating to the second measurement value Mt2 as the measurement result to the computer 90. Output.

A third measurement unit 53 is inserted near each of the four heads 21 to 24 of the four pipes 321 to 324 of this embodiment. The third measurement unit 53 is configured by, for example, a temperature sensor. A third measurement unit 53 inserted in the first pipe 321 measures the temperature of the K color ink supplied from the supply unit 30 to the first head 21, and obtains a third measurement value Mt3 as a measurement result. Also, the third measuring section 53 is electrically connected to the computer 90 . The third measuring section 53 outputs data relating to the third measured value Mt3 to the computer 90 .

Similarly, the third measurement unit 53 inserted in the second pipe 322 measures the temperature of the C color ink supplied from the supply unit 30 to the second head 22, and the third measurement value Mt3 as the measurement result is Such data is output to the computer 90 . Further, the third measurement unit 53 inserted in the third pipe 323 measures the temperature of the M-color ink supplied from the supply unit 30 to the third head 23, and determines the temperature of the third measurement value Mt3 as the measurement result. Data is output to computer 90 . Further, the third measurement unit 53 inserted in the fourth pipe 324 measures the temperature of the Y-color ink supplied from the supply unit 30 to the fourth head 24, and determines the temperature of the third measurement value Mt3 as the measurement result. Data is output to computer 90 . However, the third measuring section 53 may be provided inside each of the tanks 311-314. The third measuring section 53 may measure the temperature of each color ink in each of the tanks 311-314.

The fourth measuring section 54 is arranged outside the four heads 21-24. The fourth measurement unit 54 is configured by, for example, a temperature sensor. The fourth measurement unit 54 measures the ambient temperature outside the four heads 21 to 24 and acquires a fourth measurement value Mt4 as a measurement result. Also, the fourth measuring section 54 is electrically connected to the computer 90 . The fourth measurement unit 54 outputs data regarding the fourth measurement value Mt4 to the computer 90 .

A fixing section for fixing ink to the printing surface (front surface) of the substrate 9 may be further provided on the downstream side of the four heads 21 to 24 in the transport direction. The fixing unit, for example, blows heated gas toward the substrate 9 to dry the ink adhering to the substrate 9 . However, the fixing unit may cure the ink by irradiating the UV curable ink with ultraviolet rays.

The encoder 60 is attached to the axis of one of the plurality of transport rollers 12 (the transport roller 121 in the example of FIG. 1). The encoder 60 detects the rotational drive amount of the transport roller 121 and outputs a continuous pulse signal synchronized with the rotation of the transport roller 121 to the computer 90 . The continuous pulse signal is data that reflects changes over time in the transport speed of the base material 9 transported by the plurality of transport rollers 12 including the transport roller 121 . However, the encoder 60 does not necessarily have to be provided.

The camera 70 is an imaging device that captures an image of the printed surface (surface) of the base material 9 that has passed through the printing unit 20 . The camera 70 is arranged to face the printing surface of the base material 9 at a photographing position P5 downstream of the four heads 21 to 24 in the transport path. For the camera 70, for example, a line sensor in which a plurality of imaging devices such as CCD and CMOS are arranged in the width direction is used. While the substrate 9 is being transported by the transport mechanism 10 , the camera 70 acquires image data of the printed substrate 9 by constantly photographing the printed surface of the substrate 9 . The camera 70 then outputs the obtained image data to the computer 90 . However, the camera 70 does not necessarily have to be provided.

A plurality of various sensors 80 are measuring instruments for measuring the conveying state of the substrate 9 in addition to those mentioned above. A plurality of various sensors 80 are provided at a plurality of measurement points on the transport path of the base material 9 . Various sensors 80 acquire measured values at respective measurement locations. The measurement items of the various sensors 80 include, for example, the vertical displacement of the base material 9 (displacement amount in the direction perpendicular to the base material 9), the tension applied to the base material 9, and the edge 9E of the base material 9 (see FIG. 2). position in the width direction, etc. Various sensors 80 that measure the same item may be arranged at a plurality of positions on the transport route. While the substrate 9 is being conveyed by the conveying mechanism 10, the plurality of various sensors 80 always measure the state of each measurement point. The various sensors 80 then output data indicating the obtained measured values to the computer 90 . However, the various sensors 80 may not necessarily be provided.

The computer 90 is an information processing device for controlling the printing device 1 . FIG. 4 is a block diagram showing connections between the computer 90 and each part of the printing apparatus 1. As shown in FIG. As conceptually shown in FIG. 4, the computer 90 has a processor 901 such as a CPU, a memory 902 such as a RAM, and a storage unit 903 such as a hard disk drive. The storage unit 903 stores a computer program 90P for executing learning processing and printing processing, which will be described later.

Further, as shown in FIG. 4, the computer 90 controls the motor 14 of the transport mechanism 10, the four heads 21 to 24 of the printing unit 20, the four pumps 331 to 334 and the four flow meters 34 of the supply unit 30, Communicate with the heating unit 40, the four first measurement units 51, the four second measurement units 52, the four third measurement units 53, the fourth measurement unit 54, the encoder 60, the camera 70, and the plurality of various sensors 80, respectively. connected as possible. The computer 90 controls the operations of these units according to a computer program 90P. As a result, learning processing and printing processing, which will be described later, proceed.

<1-2. Computer functions>
The computer 90 of the printing apparatus 1 has a function of predicting the future temperature of the ink in each of the heads 21 to 24 and feedback-controlling the heating section 40 based on the predicted value. FIG. 5 is a block diagram conceptually showing the functions of the computer 90. As shown in FIG. As shown in FIG. 5 , the computer 90 has a data storage section 91 , a learning section 92 , a head temperature prediction section 93 , a temperature control section 94 , a motor drive section 95 , a print instruction section 96 and a pump control section 97 . Each of these functions is realized by the computer 90 operating according to the computer program 90P. In addition, the computer 90 may further have a function of controlling the driving of the solenoid valve and the degassing module of the supply section 30 .

As described above, the motor drive unit 95 inputs a command value for rotating the motor 14 to the motor 14 . Thereby, the motor 14 is driven according to the command value, and the driving roller rotates. As a result, the base material 9 is conveyed from the unwinding section 11 toward the winding section 13 .

The print instruction unit 96 also controls the operation of ejecting ink droplets from the heads 21 to 24 based on the received image data I. FIG. As a result, the four heads 21 to 24 record monochromatic images on the upper surface of the base material 9 by ejecting ink droplets at each processing position of each of the heads 21 to 24 . A multicolor image is formed on the upper surface of the substrate 9 by superimposing the four monochromatic images. Further, each of the heads 21 to 24 can print on the upper surface of the substrate 9 by ejecting ink droplets.

Further, as described above, when ink droplets are ejected from each of the heads 21 to 24 toward the substrate 9, the amount of ink in each of the heads 21 to 24 decreases. The first measuring unit 51 mounted on each of the heads 21 to 24 calculates the amount of ejected ink from the printed pattern and the printed amount included in the image recorded on the surface of the base material 9, and the calculation result is the first Data related to one measured value Me1 is output to the computer 90. The pump control unit 97 inputs a command value for driving each of the pumps 331-334 based on the first measured value Me1. As a result, each of the pumps 331 to 334 is driven according to the command value, and an amount of ink corresponding to the amount of ink ejected from each of the heads 21 to 24 is supplied from the supply section 30 to each of the heads 21 to 24. supplied.

The data accumulation unit 91 is a processing unit for accumulating data used for learning processing, which will be described later. A first measured value Me1 input from the four first measuring units 51, a second measured value Mt2 input from the four second measuring units 52, and a third measured value Mt3 input from the four third measuring units 53. , and the fourth measurement value Mt4 input from the fourth measurement unit 54 are stored in the data accumulation unit 91 . The data accumulation unit 91 stores these measured values together with the measurement times when these measured values were obtained.

The data accumulation unit 91 stores measured values Mf input from the four flowmeters 34, continuous pulse signals input from the encoder 60, image data input from the camera 70, and input from a plurality of various sensors 80. Various data may be further stored together with these measured values and the measurement times when the data were acquired.

The learning unit 92 is a processing unit for learning the relationship between the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 and the temperature of the ink in each of the heads 21-24. The learning unit 92 reads the learning data Dl from the data storage unit 91 . The learning data Dl includes a first measured value Me1, a third measured value Mt3, and a fourth measured value Mt4 at time t1, and each head at time t2 after a predetermined time (for example, several seconds) after time t1. A second measurement Mt2, which is the actual measurement of the temperature of the ink within 21-24, is included. The learning unit 92 reads a large number of such learning data Dl from the data storage unit 91 at different times.

The learning unit 92 uses the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 as input variables, and from the measurement times of the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4, A learning model M0 whose target variable is a second measured value Mt2, which is the actual measurement value of the temperature of ink in each of the heads 21 to 24 after a predetermined time, is learned by a supervised machine learning algorithm. The learning unit 92 repeats such learning processing until a predetermined termination condition is satisfied. As a result, the learning unit 92 uses the first measurement value Me1, the third measurement value Mt3, and the fourth measurement value Mt4 to output the temperature of the ink in each of the heads 21 to 24 after a predetermined time, based on the prediction model M1. can be generated. Details of the learning process will be described later.

The machine learning algorithm used in the learning unit 92 may be a simple regression algorithm or a tree model represented by random forest or gradient boosting. Also, the machine learning algorithm used in the learning unit 92 may be a deep learning algorithm such as a convolutional neural network. Also, an appropriate combination of machine learning algorithms and input variables may be selected for each type and characteristic of ink used in the printing apparatus 1 .

Based on the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4, the head temperature prediction unit 93 calculates the measurement times of the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4. is a processing unit for calculating the predicted value of the temperature of the ink in each of the heads 21 to 24 at the time after a predetermined time from . The head temperature prediction unit 93 uses the prediction model M1 generated by the learning unit 92 to predict the temperature of the ink inside each of the heads 21-24. The head temperature prediction unit 93 inputs the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 to the prediction model M1. Then, the prediction model M1 outputs a prediction value Df of the ink temperature in each of the heads 21 to 24 corresponding to the first measurement value Me1, the third measurement value Mt3, and the fourth measurement value Mt4. The head temperature prediction section 93 inputs the prediction value Df output from the prediction model M1 to the temperature control section 94 .

The temperature control section 94 is a processing section for controlling the operation of the heating section 40 . The temperature control section 94 calculates a control value by performing feedback control based on the difference between the predicted value Df input from the head temperature prediction section 93 and the preset target value Ds. The temperature control section 94 then outputs a control signal Dc including the calculated control value to the heating section 40 . As a result, the heating unit 40 is driven to heat each color ink stored in the four tanks 311 to 314 . As a result, the ink in the four tanks 311-314, the four pipes 321-324, and the four heads 21-24 can be adjusted to a predetermined temperature.

<1-3. About the learning process>
Next, learning processing executed in the printing apparatus 1 will be described. FIG. 6 is a flowchart showing the flow of learning processing.

As shown in FIG. 6, when performing the learning process, first, the transport mechanism 10 is operated to transport the base material 9 in the longitudinal direction along a predetermined transport path, while each head 21 is placed on the surface of the base material 9. 24 to record or print an image on the surface of the substrate 9 (step S1). In addition, the pumps 331 to 334 are driven to supply a corresponding amount of ink from the supply section 30 to the heads 21 to 24 according to the amount of ink ejected from the heads 21 to 24 (step S2). Also, the heating unit 40 is operated to heat the ink supplied from the supply unit 30 to each of the heads 21 to 24 (step S3). In this embodiment, step S1 to step S5, which will be described later, are performed in parallel.

Further, the four first measurement units 51, the four second measurement units 52, the four third measurement units 53, and the fourth measurement unit 54 obtain the first measurement value Me1, the second measurement value Mt2, the third measurement value Mt3 and the fourth measurement value Mt4 are acquired (step S4).

Specifically, the first measuring unit 51 calculates the amount of ink ejected from each of the heads 21 to 24 in step S1 from the print pattern and print amount included in the image recorded on the surface of the base material 9. , data relating to the first measured value Me1 is output to the computer 90 as a calculation result. However, in step S2, the equivalent amount of ink of each color supplied from the supply unit 30 to each of the heads 21 to 24 is measured by the flow meter 34, and the measured value Mf is measured by the first measuring unit 51 as the first measured value Me1 Alternatively, or together with the first measurement value Me1 by the first measurement unit 51, it may be used as the “first measurement value”.

The second measurement unit 52 also measures the temperature of the ink in each of the heads 21 to 24 and outputs data related to the second measurement value Mt2 to the computer 90 as the measurement result. Further, the third measurement unit 53 measures the temperature of each color ink supplied from the supply unit 30 to each of the heads 21 to 24, and outputs data related to the third measurement value Mt3 to the computer 90 as the measurement result. Further, the fourth measurement unit 54 measures the ambient temperature outside the four heads 21 to 24, and outputs data related to the fourth measurement value Mt4 to the computer 90 as the measurement result.

Furthermore, the encoder 60 may detect the rotational drive amount of the transport roller 121 and output a continuous pulse signal synchronized with the rotation of the transport roller 121 to the computer 90 . Further, the camera 70 may acquire image data of the printed base material 9 by constantly photographing the printed surface of the base material 9 and output the image data to the computer 90 . In addition, a plurality of various sensors 80 measure the vertical displacement of the base material 9, the tension applied to the base material 9, the position of the edge 9E of the base material 9 in the width direction, etc., and transmit the data related to the measurement results to the computer 90. may be output. These data input to the computer 90 may be stored in the data storage unit 91 in step S5 to be described later, acquired as learning data Dl in step S6, and used as input variables in the learning process in step S7. .

The data storage unit 91 receives the first measured value Me1 input from the four first measuring units 51, the second measured value Mt2 input from the four second measuring units 52, and the four third measuring units 53. The third measured value Mt3 obtained and the fourth measured value Mt4 input from the fourth measuring unit 54 are stored together with the measurement time when these measured values were acquired (step S5).

When the data necessary for learning are sufficiently accumulated, the learning section 92 acquires the learning data Dl from the data accumulation section 91 (step S6). The learning data Dl includes a first measured value Me1, a third measured value Mt3, and a fourth measured value Mt4 at time t1, and each head at time t2 after a predetermined time (for example, several seconds) after time t1. A second measurement Mt2, which is the actual measurement of the temperature of the ink within 21-24, is included.

Subsequently, a learning model M0 is prepared as a prototype of a prediction model M1, which will be described later. Learning model M0 is installed in storage unit 903 . Then, the learning unit 92 uses the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 at a certain measurement time, which are included in the learning data Dl, as input variables, The above-described learning model M0, whose objective variable is the second measured value Mt2 at a time after a predetermined time from the measurement time of the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4, is a supervised machine learning algorithm. Then, learning processing is performed (step S7). At this time, the learning unit 92 uses the acquisition result of the second measured value Mt2 at a time after a predetermined time from the measurement times of the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 as teacher data. Learning processing of the learning model M0 is performed by updating and saving while adjusting a plurality of parameters included in the learning model M0 so that the difference between the data and the output value from the learning model M0 becomes small.

After that, the learning unit 92 determines whether or not a predetermined termination condition is satisfied (step S8). The end condition is, for example, the output value (prediction result) from the learning model M0 when the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 are input to the learning model M0, and The difference between the obtained result (actual value) of the second measured value Mt2 at the time after a predetermined time from the measurement time of the input first measured value Me1, third measured value Mt3, and fourth measured value Mt4 is set in advance. be smaller than the set threshold. Alternatively, the termination condition may be that the number of repetitions of steps S6 and S7 reaches a preset threshold value. If the termination condition is not satisfied (step S8: No), the learning section 92 repeats the processes of steps S6 and S7. At that time, the learning data Dl may be used at a time different from that of the previous time.

By repeating the learning process of steps S6 to S7, the prediction accuracy of the learning model M0 is improved. When the end condition is eventually satisfied (step S8: Yes), the learning unit 92 ends the learning process. In addition, below, suppose that the learning model M0 after learning processing is completed is called the prediction model M1. That is, the learning method described above can predict (output) the temperature of the ink in each of the heads 21 to 24 after a predetermined time based on the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4. , a trained prediction model M1 is generated. As a result, even if the consumption of ink ejected from each of the heads 21 to 24 onto the substrate 9 fluctuates or the amount of ink newly stored in each of the heads 21 to 24 changes, can be predicted (output) with high accuracy by using the prediction model M1. Note that the learning unit 92 provides the generated prediction model M1 to the head temperature prediction unit 93 .

Note that, when performing the above learning process, a predetermined number of learning data Dl may be used as one learning data set. Then, learning processing may be performed on a plurality of learning data sets. In this case, while the processing of steps S6 to S7 is performed for a predetermined number of learning data Dl included in one learning data set, the processing of steps S6 to S7 is performed without determining the end condition of step S8. You can repeat. Then, the end condition of step S8 may be determined when the processing of steps S6 to S7 is completed for all the learning data Dl included in one learning data set. If the termination condition is not satisfied in step S8, the learning process of steps S6 and S7 may be performed on another learning data set.

Further, when performing the above learning process, the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 input to the learning model M0 may have different measurement times. . In this case, the above-mentioned "predetermined time (after) (from the measurement time)" is a different value for each measurement time of the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4. There may be.

<1-4. About print processing>
Next, the printing process performed by the printing apparatus 1 after the learning process described above will be described. FIG. 7 is a diagram conceptually showing a partial configuration of the printing apparatus 1 when performing print processing. FIG. 8 is a flowchart showing the flow of print processing.

As shown in FIGS. 7 and 8, when performing the printing process, first, the transport mechanism 10 is operated to transport the base material 9 along a predetermined transport path in the longitudinal direction, while the surface of the base material 9 is coated with An image is recorded or printed on the surface of the substrate 9 by ejecting ink droplets from each of the heads 21 to 24 (step S9). In addition, the pumps 331 to 334 are driven to supply a corresponding amount of ink from the supply unit 30 to the heads 21 to 24 according to the amount of ink ejected from the heads 21 to 24 (step S10). Also, the heating unit 40 is operated to heat the ink supplied from the supply unit 30 to each of the heads 21 to 24 (step S11).

Further, the four first measurement units 51, the four third measurement units 53, and the fourth measurement unit 54 acquire the first measurement value Me1, the third measurement value Mt3, and the fourth measurement value Mt4 (step S12 ).

Specifically, the first measuring unit 51 calculates the amount of ink ejected from each of the heads 21 to 24 in step S9 from the print pattern and print amount included in the image recorded on the surface of the base material 9. , data relating to the first measured value Me1 is output to the computer 90 as a calculation result. However, in step S10, the equivalent amount of each color ink supplied from the supply unit 30 to each of the heads 21 to 24 is measured using the flow meter 34, and the measured value Mf is measured by the first measuring unit 51 as the first measured value Me1 Alternatively, or together with the first measurement value Me1 by the first measurement unit 51, it may be used as the “first measurement value”.

Further, the third measurement unit 53 measures the temperature of each color ink supplied from the supply unit 30 to each of the heads 21 to 24, and outputs data related to the third measurement value Mt3 to the computer 90 as the measurement result. Further, the fourth measurement unit 54 measures the ambient temperature outside the four heads 21 to 24, and outputs data related to the fourth measurement value Mt4 to the computer 90 as the measurement result.

In the learning process described above, if machine learning is performed using the continuous pulse signal from the encoder 60, the image data from the camera 70, or the measurement results from the plurality of various sensors 80, then in the printing process as well, these It acquires signals and data and outputs them to the computer 90 . Then, in step S13, which will be described later, these signals and data are input to the prediction model M1 as input variables.

The head temperature prediction unit 93 calculates the current first measured value Me1, third measured value Mt3, and fourth measured value Mt4 input from the first measuring unit 51, the third measuring unit 53, and the fourth measuring unit 54. is input to the prediction model M1 as an input variable. Then, the prediction model M1 calculates (outputs) a predicted value Df of the ink temperature in each of the heads 21 to 24 after a predetermined time (step S13). Thereby, the predicted value Df of the temperature of the ink in each of the heads 21 to 24 after a predetermined time can be estimated with high accuracy. The head temperature prediction section 93 inputs the obtained predicted value Df to the temperature control section 94 .

The temperature control unit 94 controls the operation of the heating unit 40 based on the predicted value Df of the temperature of the ink in each of the heads 21 to 24 after a predetermined period of time, which is calculated by the above estimation method (step S14). A target value Ds for the temperature of the ink in each of the heads 21 to 24 is preset in the temperature control section 94 . The temperature control section 94 feedback-controls the heating section 40 so that the predicted value Df input from the head temperature prediction section 93 approaches the target value Ds. That is, the temperature control unit 94 feedback-controls the heating unit 40 based on the difference between the predicted value Df and the preset target value Ds to control the tanks 311 to 314, the pipes 321 to 324, and The temperature of the ink in each head 21-24 can be adjusted to a predetermined temperature.

PID control, for example, is used for feedback control. That is, the temperature control unit 94 calculates the control value based on the deviation between the predicted value Df and the target value Ds, the differential value of the deviation, and the integral value of the deviation. Then, a control signal Dc including the control value is output to the heating section 40 . Thereby, the heating of the heating unit 40 is controlled, and each color ink stored in each of the tanks 311 to 314 is heated. As a result, the temperature of the ink in each of the pipes 321 to 324 connected to each of the tanks 311 to 314 and in each of the heads 21 to 24 can be adjusted to a predetermined temperature.

The printing apparatus 1 performs steps S9 to S13 in parallel, and controls the heating by the heating unit 40 in step S11 based on the ink temperature adjustment method in step S14. As a result, while recording or printing an image on the surface of the base material 9, the temperature of the ink in each of the heads 21 to 24 can be maintained at an appropriate temperature, so high print quality can be achieved.

After that, the computer 90 determines whether or not to end the printing process (step S15). If image data to be printed remains, the computer 90 continues the printing process (step S15: No). In this case, the computer 90 executes the processes of steps S9 to S14 again. In this way, the computer 90 repeats the processes of steps S9 to S14, thereby maintaining the temperature of the ink in each of the heads 21 to 24 at an appropriate temperature, and advancing the printing process while conveying the substrate 9. be able to.

Eventually, when there is no more image data to be printed, the computer 90 stops the operation of each unit and ends the printing process (step S15: Yes).

As described above, in the printing apparatus 1, the prediction model M1 obtained by machine learning is used to predict the temperature of the ink in each of the heads 21 to 24 after a predetermined time. 24 to heat each color of ink supplied. As a result, the temperature of the ink inside each of the heads 21 to 24 can be controlled with high accuracy. As a result, even if the consumption of ink ejected from each head 21 to 24 fluctuates or the amount of ink newly stored in each head 21 to 24 changes as ink is ejected, each head Since the temperature of each color ink stored in 21 to 24 can be maintained within an appropriate range, high print quality can be achieved.

Further, in the printing apparatus 1 of the present embodiment, it is possible to additionally execute the learning process of FIG. 6 while executing the printing process of FIG. In this way, the prediction model M1 can be updated at any time while executing the printing process. Therefore, even when the installation environment of the printing apparatus 1 changes or the state of the printing apparatus 1 changes, the prediction accuracy of the prediction model M1 can always be maintained at a high level.

<2. Variation>
Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment.

In the above embodiment, the first measured value Me1 input from the four first measuring units 51, the third measured value Mt3 input from the four third measuring units 53, and the fourth measuring unit 54 are used as the learning model. It was used as an input variable for M0 and prediction model M1. However, the input variables of the learning model M0 and the prediction model M1 are not limited to this. The input variables of the learning model M0 and the prediction model M1 should include at least the first measurement value Me1 input from the first measurement unit 51 .

Also, the first measurement values Me1 input from the four first measurement units 51 are different values depending on the heads 21 to 24 on which the first measurement units 51 are mounted. However, in order to suppress the amount of calculation in the computer 90, only the first measurement value Me1 by the first measurement unit 51 mounted on any one of the four heads 21 to 24 is used as the learning model M0 and the prediction model M1. It may be an input variable. Similarly, only the third measurement value Mt3 by the third measurement unit 53 inserted in any one of the four pipes 321-324 may be used as the input variable for the learning model M0 and the prediction model M1. Alternatively, only the second measurement value Mt2 obtained by the second measurement unit 52 mounted on any one of the four heads 21 to 24 may be used as teacher data used in the above learning process.

Further, in the above embodiment, the first measured value Me1, the second measured value Mt2, the third measured value Mt3, and the fourth measured value Mt4 are used as they are as input variables or teacher data. However, these measured values may be subjected to processing such as filtering, and the processed measured values may be used as input variables or teacher data.

In the above-described embodiment, four flowmeters 34 are inserted in each of the pipes 321-324 to measure the amount of ink of each color supplied from the supply unit 30 to each of the heads 21-24. However, instead of the four flowmeters 34 or in addition to the four flowmeters 34, four pressure sensors for measuring the pressure of each color ink supplied to each head 21 to 24 from the supply unit 30 are It may be inserted in each of the pipes 321-324.

In the embodiment described above, the ink of each color is supplied in one direction from the supply unit 30 to each of the heads 21 to 24 . However, the printing apparatus 1 may have a structure in which each color of ink circulates between the supply unit 30 and each of the heads 21 to 24, and each color of ink is heated on the circulation path. . Further, the printing apparatus 1 may have a configuration in which the ink of each color circulates in each of the heads 21 to 24, and the ink of each color is heated on the circulation path.

Moreover, in the above embodiment, PID control was given as an example of feedback control. However, the feedback control may be another method such as PI control.

In the above embodiment, the first measured value Me1, the third measured value Mt3, and the fourth measured value Mt4 are used as input variables, and the temperature of the ink in each of the heads 21 to 24 after a predetermined time is actually measured. The learning model M0 with the second measured value Mt2 as the objective variable is learned by a supervised machine learning algorithm. However, instead of performing supervised machine learning, reinforcement learning is performed so that the control value for controlling the heating unit can be output with high accuracy in order to maintain the temperature of the ink in each of the heads 21 to 24 within an appropriate range. you can go That is, in order to maintain the second measured value Mt2, which is the actually measured value of the temperature of the ink in each of the heads 21 to 24, within an appropriate range, the control value for controlling the heating unit is updated and stored through reinforcement learning. You can do it.

Further, in the above embodiment, the printing apparatus 1 that ejects ink onto the surface of the base material 9 has been described. However, the printing apparatus of the present invention may eject a treatment substance other than ink onto the surface of the substrate conveyed by the conveying mechanism. For example, the processing apparatus may apply a resist solution or the like to the surface of the base material transported by the transport mechanism.

Also, the elements appearing in the above embodiments and modifications may be appropriately combined as long as there is no contradiction.

1 Printing Device 9 Base Material 10 Conveying Mechanism 20 Printing Section 21 First Head 22 Second Head 23 Third Head 24 Fourth Head 30 Supply Section 34 Flowmeter 40 Heating Section 51 First Measuring Section 52 Second Measuring Section 53 Third Measurement unit 54 Fourth measurement unit 90 Computer 91 Data accumulation unit 92 Learning unit 93 Head temperature prediction unit 94 Temperature control unit 311 First tank 312 Second tank 313 Third tank 314 Fourth tank 321 First pipe 322 Second pipe 323 3rd pipe 324 4th pipe 331 1st pump 332 2nd pump 333 3rd pump 334 4th pump Dc Control signal (for controlling heating unit) Df Predicted value (temperature of ink in head) Dl For learning Data Ds Target value (temperature of ink in head) M0 Learning model M1 Prediction model Me1 First measured value Mf Measured value (by flow meter) Mt2 Second measured value Mt3 Third measured value Mt4 Fourth measured value

Claims (13)

a) a step of ejecting a treatment substance from a head onto the surface of a long strip-shaped substrate while transporting the substrate along a predetermined transport path in the longitudinal direction;
b) supplying a corresponding amount of the processing substance from a supply unit to the head according to the amount of the processing substance ejected from the head;
c) heating the material to be treated that is supplied from the supply unit to the head;
d) measuring the amount of the treatment substance ejected from the head in step a) or the equivalent amount of the treatment substance supplied in step b) to obtain a first measured value;
e) measuring the temperature of the material to be treated in the head to obtain a second measurement;
has
f) using a machine learning algorithm to learn a learning model using the first measured value as an input variable and the temperature of the material to be treated in the head at a time after a predetermined time from the measurement of the first measured value as an objective variable; a step of processing;
has
In the step f), the acquisition result of the second measurement value at the time after the predetermined time from the measurement time of the first measurement value is used as teacher data, and the difference between the teacher data and the output value from the learning model. A learning method that updates and saves while adjusting a plurality of parameters included in the learning model so that . The learning method according to claim 1,
In the step a), an image is recorded or printed on the surface of the base material by ejecting ink as the treatment substance from the head,
In the step d), the amount of the treatment substance ejected from the head in the step a) is calculated from the print pattern or print amount included in the image recorded on the surface of the base material. A learning method that acquires 1 measured value. The learning method according to claim 1 or claim 2,
g) further comprising measuring the temperature of the material to be treated supplied from the supply unit to the head to obtain a third measured value;
In the step f), the learning method further performs learning processing on the learning model while using the third measured value as an input variable. The learning method according to any one of claims 1 to 3,
h) further comprising measuring the ambient temperature outside the head to obtain a fourth measured value;
In the step f), the learning method performs learning processing on the learning model while using the fourth measurement value as an input variable. A prediction model generated by the learning method according to any one of claims 1 to 4. p) ejecting the treatment substance from the head onto the surface of the base material while conveying the long belt-like base material along the predetermined conveying path in the longitudinal direction;
q) supplying to the head from the supply unit the equivalent amount of the processing substance corresponding to the amount of the processing substance discharged from the head;
r) heating the treatment substance supplied from the supply unit to the head with a heating unit;
s) measuring the amount of the treatment substance ejected from the head in step p) or the equivalent amount of the treatment substance supplied in step q) to obtain the first measured value; ,
t) inputting the first measured value into the predictive model according to claim 5 as an input variable to calculate a predicted value of the temperature of the material to be treated in the head from the predictive model;
A temperature estimation method comprising: A temperature adjustment method, wherein feedback control is performed on the heating unit based on a difference between the predicted value calculated by the temperature estimation method according to claim 6 and a preset target value. A printing method, wherein the heating by the heating unit in the step r) is controlled based on the temperature adjustment method according to claim 7, while performing the steps p) to t) in parallel. a transport mechanism for transporting a long strip-shaped base material in the longitudinal direction along a predetermined transport path;
a processing unit that ejects a processing substance from a head onto the surface of the substrate at a processing position on the transport path;
a supply unit that supplies the head with an amount of the processing substance corresponding to the amount of the processing substance ejected from the head;
a heating unit that heats the processing material supplied from the supply unit to the head;
a first measurement unit that measures the amount of the processing substance ejected from the head or the equivalent amount of the processing substance supplied from the supply unit to the head to obtain a first measurement value;
a second measuring unit that measures the temperature of the material to be processed in the head to obtain a second measured value;
A machine learning algorithm learns a learning model using the first measured value as an input variable and the objective variable as the temperature of the material to be treated in the head at a predetermined time after the measurement of the first measured value. the learning department;
a head temperature prediction unit that calculates a predicted value of the temperature of the material to be processed in the head at the time after the predetermined time from the measurement time of the first measurement value;
has
The learning unit uses the acquisition result of the second measured value at the time after the predetermined time from the measurement time of the first measured value as teacher data, and the difference between the teacher data and the output value from the learning model is A printing device that performs a learning process on the learning model by updating and saving a plurality of parameters included in the learning model so as to reduce the size of the learning model. 10. The printing device according to claim 9,
The processing unit is a printing unit that records or prints an image on the surface of the substrate by ejecting ink as the processing substance from the head,
The first measurement unit calculates the first measurement value by calculating the amount of the treatment substance ejected from the head from a print pattern or print amount included in an image recorded on the surface of the base material. The printer to retrieve. The printing device according to claim 9 or 10,
further comprising a third measurement unit that measures the temperature of the material to be processed supplied from the supply unit to the head to obtain a third measurement value;
The printing apparatus, wherein the learning model further uses the third measurement value as an input variable. The printing device according to any one of claims 9 to 11,
further comprising a fourth measurement unit that measures the environmental temperature outside the head to obtain a fourth measurement value;
The printing apparatus, wherein the learning model further uses the fourth measurement value as an input variable. The printing device according to any one of claims 9 to 12,
A printing apparatus, further comprising a temperature control section that feedback-controls the heating section based on a difference between the predicted value and a preset target value.

Images