Control device, drying device, and learning device

Abstract

The present invention provides a control device, a drying device, and a learning device that can efficiently dry a coating film formed on a workpiece using directional hot air. [Solution] The control device 100 comprises a first acquisition unit that acquires work information indicating the type of work in a drying oven that performs a drying operation to dry a coating film formed on a work, and a control unit that controls the direction of a plurality of ejection nozzles that eject hot air for drying the coating film based on the work information acquired by the first acquisition unit, and further comprises a second acquisition unit that acquires temperature data indicating the temperature of the work during the drying operation, and the control unit corrects the direction of the ejection nozzles based on the temperature data acquired by the second acquisition unit.

Description

Technical Field

[0001] The present invention relates to a control device, a drying device, and a learning device.

Background Art

[0002] Patent Document 1 discloses a coating drying device for locally drying a coating film. This coating drying device includes a drying furnace main body and hot air generation and supply means for supplying hot air into the drying furnace main body, and dries a wet coating film applied to an outer plate portion and a narrow portion while conveying an automobile body including the outer plate portion and the narrow portion. In the coating drying device, the drying furnace main body mainly blows hot air toward the narrow portion and includes a local drying region for locally drying the coating film applied to the narrow portion.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the technology of Patent Document 1, hot air is locally blown toward the coating surface of the narrow portion that is difficult to heat up. On the other hand, there is a problem in controlling the nozzle to locally blow hot air to locations other than the narrow portion.

[0005] The present invention provides a control device, a drying device, and a learning device that can efficiently dry a coating film formed on a workpiece with hot air having directivity.

Means for Solving the Problems

[0006] A control device according to a first aspect of the present invention comprises: a first acquisition unit that acquires work information indicating the type of work in a drying oven that performs a drying operation to dry a coating film formed on a work; and a control unit that controls the orientation of a plurality of ejection nozzles that eject hot air for drying the coating film based on the work information acquired by the first acquisition unit, wherein the control device further comprises a second acquisition unit that acquires temperature data indicating the temperature of the work during the drying operation, and the control unit corrects the orientation of the ejection nozzles based on the temperature data acquired by the second acquisition unit. [Effects of the Invention]

[0007] According to the present invention, a coating film formed on a workpiece can be efficiently dried using directional hot air. [Brief explanation of the drawing]

[0008] [Figure 1] This is a schematic block diagram of the control system according to the first embodiment. [Figure 2] This is a longitudinal cross-sectional view showing a paint drying oven according to the first embodiment. [Figure 3A] This is a longitudinal cross-sectional view showing a part of a paint drying oven according to the first embodiment. [Figure 3B] This is a view in the direction of arrow 3B in Figure 3A, showing a part of the configuration shown in Figure 3A. [Figure 4A] This is a cross-sectional view corresponding to Figure 3A, showing the inclination state of the ejection nozzle. [Figure 4B] This is a view in the direction of arrow 4B in Figure 4A, showing a part of the configuration shown in Figure 4A. [Figure 5] This is a block diagram showing the hardware configuration of the control device according to the first embodiment. [Figure 6] This is a block diagram showing an example of the functional configuration of a control device according to the first embodiment. [Figure 7] This is a block diagram showing an example of the functional configuration of a learning device according to the first embodiment. [Figure 8]This flowchart shows an example of the learning process flow of the learning device according to the first embodiment. [Figure 9] This flowchart shows an example of the control process flow of the control device according to the first embodiment. [Figure 10] This flowchart shows an example of the flow of the relearning process of the learning device according to the first embodiment. [Figure 11] This flowchart shows an example of the pretreatment flow according to a modified example of the first embodiment. [Figure 12] This is a schematic block diagram of the control system according to the second embodiment. [Figure 13] This is a longitudinal cross-sectional view showing a paint drying oven according to a second embodiment. [Figure 14] This is a schematic diagram of the control system according to the third embodiment. [Figure 15] This is a longitudinal cross-sectional view showing a paint drying oven according to the third embodiment. [Modes for carrying out the invention]

[0009] [First Embodiment] The control system 1 according to this embodiment will be described below with reference to the drawings. In this embodiment, the control system 1 will be described using as an example the case in which the control device 100 controls various devices of the paint drying oven 10 to bake and dry a wet paint film applied to the automobile body B of a vehicle. Note that the present invention is not limited to wet paint films formed on the body of a vehicle, but may be applied to any case of drying paint films formed on various workpieces. In each drawing, the same or equivalent components are given the same reference numerals. Also, the dimensional ratios in the drawings are exaggerated for explanatory purposes and may differ from the actual ratios. Also, the arrow UP indicated in each drawing indicates the upward direction, and the arrow R indicates the rightward direction. Furthermore, the present invention is not limited in any way to the following embodiments, and can be implemented with appropriate modifications within the scope of the object of the present invention.

[0010] (Control system) FIG. 1 is a block diagram showing an outline of a control system 1 according to a first embodiment. As shown in FIG. 1, the control system 1 of the present embodiment includes a control device 100, a learning device 200, and a painting drying furnace 10. The control device 100, the learning device 200, and the painting drying furnace 10 are connected via a network. The network may adopt any type of communication network such as a LAN (Local Area Network), the Internet, a VPN (Virtual Private Network), etc., regardless of being wired or wireless. In the present embodiment, the control device 100 and the learning device 200 are separate devices, but they may be configured as an integrated device. Also, the number of painting drying furnaces 10 may be one or plural. And each set of the control device 100 and the painting drying furnace 10 may be installed at each of a plurality of different sites.

[0011] The control device 100 is a device that controls a plurality of nozzle drive devices 23 and an air supply device 70 provided in the painting drying furnace 10 based on measurement data acquired from a measurement device 80 of the painting drying furnace 10 described later and information indicating the production specifications of the automobile body B (hereinafter also referred to as specification information). The measurement data of the present embodiment includes temperature data indicating the temperature of the automobile body B during drying (for example, the surface temperature of the automobile body B, the temperature inside the painting drying furnace 10, etc.), and environmental data indicating the environment outside the painting drying furnace 10 during drying (for example, air temperature, humidity, date and time, etc.). Also, the specification information of the present embodiment includes the vehicle type, paint color, dimensions, etc. of the automobile body B. The specification information is an example of "work information".

[0012] The control device 100 is, for example, a computer or a PLC (Programmable Logic Controller), and includes a learned model 142A and a control program 121. In the present embodiment, the control device 100 will be described as being a computer. The learned model 142A of the present embodiment is stored in a storage 140 (see FIG. 5) described later. Also, the control program 121 is stored in a ROM 120 (see FIG. 5) described later.

[0013] The learned model 142A of the present embodiment is a learned model obtained by deploying the learned model 242A learned by the learning device 200 described later to the control device 100. The learned model 142A outputs control data for controlling the nozzle drive device 23 included in the painting drying furnace 10. The control data of the present embodiment is data for controlling each of the plurality of nozzle drive devices 23 of the painting drying furnace 10. Based on the control data, the control device 100 changes the rotation angle and tilt degree of the ejection nozzle 18 (see FIG. 2), and the air volume and wind speed of the hot air ejected from the ejection nozzle 18. When there is insufficient data input to the learned model 142A, the control device 100 may supplement it with predetermined default values, average values, etc.

[0014] The control program 121 is a program that executes a process including a control process (see FIG. 9) described later. When the control program 121 is executed, the control device 100 uses various hardware resources (see FIG. 5) to execute the process based on the control program 121.

[0015] The learning device 200 is a device that causes a learned model that outputs control data to be learned using pre-prepared learning data. The learning data may include temperature temperature data such as temperature data, environmental data, control data, specification information, and evaluation data. The evaluation data is data indicating the evaluation of the coating film in the inspection process of the coating film after drying. Examples of inspection items for evaluating the coating film include gel fraction, baking rate, coating film hardness, MIBK adhesion test, and cross-cut test.

[0016] The learning device 200 is, for example, a computer, and includes learning data 241A, a learned model 242A, re-learning data 243A, and a learning program 221. The learning data 241A, the learned model 242A, and the re-learning data 243A of the present embodiment are stored in a storage 240 (see FIG. 5) described later. The learning program 221 is stored in a ROM 220 (see FIG. 5) described later.

[0017] The training data 241A includes temperature data acquired by the thermal camera 80A (see Figure 2) of the paint drying oven 10. The training data 241A also includes control data when the direction of the spray nozzle 18 was corrected to efficiently dry the paint film formed on the automobile body B. Furthermore, the training data 241A includes environmental data during the drying of the paint film on the automobile body B. Finally, the training data 241A includes evaluation data of the paint film formed on the automobile body B. The training data 241A may also include specification information of the automobile body B.

[0018] The trained model 242A is a trained model that has been trained using a machine learning algorithm with training data 241A as training data. The trained model 242A outputs control data when data including at least temperature data is input. The trained model 242A is trained to output control data when temperature data and environmental data are input, for example.

[0019] The retraining data 243A is training data for retraining the trained model 242A, and includes control data from when the control device 100 controlled the paint drying oven 10. The retraining data 243A also includes temperature data, environmental data, and evaluation data from when the paint film formed on the automobile body B was dried using the control data output from the trained model 142A. The retraining data 243A may also include specification information of the automobile body B.

[0020] The learning program 221 is a program that executes processes including the learning process (see Figure 8), the retraining process (see Figure 10), and the preprocessing (see Figure 11), which will be described later. When the learning program 221 is executed, the learning device 200 uses various hardware resources (see Figure 5) to execute processes based on the learning program 221.

[0021] In other words, the learning device 200 of this embodiment trains the trained model 242A using the training data 241A, and retrains the trained model 242A using the retraining data 243A.

[0022] The paint drying oven 10 is a device for drying the paint film formed on the automobile body B of a vehicle. In this embodiment, the paint drying oven 10 bakes and dries the wet paint film applied to the automobile body B by electrodeposition coating using hot air ejected from ejection nozzles 18 whose direction is changed by a nozzle drive device 23. The paint drying oven 10 also measures the temperature inside the paint drying oven 10, as well as the temperature and humidity outside the paint drying oven 10, using various sensors (not shown). The paint drying oven 10 is equipped with a plurality of nozzle drive devices 23, an air supply device 70, and a plurality of measuring devices 80.

[0023] The nozzle drive device 23 is a device for controlling the rotation angle and inclination of multiple hot air ejection nozzles 18 (see Figure 2) installed on the walls and ceiling of the paint drying oven 10. Multiple nozzle drive devices 23 are installed, one for each of the multiple ejection nozzles 18. The nozzle drive device 23 in this embodiment includes a rotary motor 24 (see Figure 2), which is a rotation drive source for changing the rotation angle of the ejection nozzle 18, and an inclination motor 26 (see Figure 2), which is an inclination drive source for changing the inclination. Hereinafter, in this embodiment, when it is not necessary to distinguish between the rotary motor 24 and the inclination motor 26, they may be collectively referred to as the nozzle drive device 23. Note that multiple ejection nozzles 18 may be controlled by a single nozzle drive device 23.

[0024] The air supply device 70 is a device for supplying hot air ejected from the discharge nozzle 18. The air supply device 70 includes at least an air supply fan (not shown) and a burner. The air supply device 70 sends air heated to a predetermined temperature by the burner into the air supply duct 14 (see Figure 2) using the air supply fan. The control device 100 in this embodiment controls the airflow rate and air velocity of the hot air supplied from the discharge nozzle 18 by controlling the air supply fan of the air supply device 70. Although only one air supply device 70 is shown in Figure 1, there may be multiple air supply devices 70.

[0025] The measuring device 80 is a device for measuring the temperature of the automobile body B during drying. Specifically, the measuring device 80 consists of various temperature sensors such as a thermal camera 80A, a high-temperature thermopile 80B, and a resistance thermometer 80C. In this embodiment, the measuring device 80 is a thermal camera 80A, which measures the surface temperature of various parts of the automobile body B.

[0026] In this embodiment of the control system 1, the control device 100 controls the multiple nozzle drive devices 23 and the air supply device 70 of the paint drying oven 10 based on measurement data acquired from the measuring device 80. In addition, in this embodiment of the control system 1, the learning device 200 retrains the trained model 242A based on learning data including the control data acquired from the control device 100, and deploys the retrained trained model 242A as trained model 142A to the control device 100.

[0027] (drying oven) The paint drying oven 10 according to the first embodiment of the present invention will be described below with reference to Figures 2 to 4B. In each figure, some reference numerals may be omitted for the sake of clarity. The following describes a cross-sectional view of the heating area of ​​the paint drying oven 10. The heating area of ​​the paint drying oven 10 is the area where the painted area is heated in order to bake and dry the wet paint film applied to the automobile body B.

[0028] As shown in Figure 2, the paint drying oven 10 according to this embodiment is one of the devices that make up the painting line for an automobile body B, and is a device for drying the wet paint film applied to the automobile body B while transporting the automobile body B mounted on the painting trolley T. The above drying includes drying the topcoat, drying the intermediate coat, drying the undercoat, etc.

[0029] This paint drying oven 10 includes a drying oven body 12 through which an automobile body B coated with a wet paint film is transported. The automobile body B corresponds to the "workpiece" in this invention. The automobile body B is coated with a wet paint film, for example, by electrodeposition coating. The drying oven body 12 is rectangular and has a top wall 12A, a floor 12B, a left wall 12C, and a right wall 12D.

[0030] A painting trolley T for transporting the automobile body B is mounted on the floor surface 12B. This painting trolley T transports the automobile body B from the back to the front in the direction perpendicular to the plane of the paper in Figure 2. An air supply duct 14 is provided inside each of the furnace sections: the top wall 12A, the left wall 12C, and the right wall 12D. In addition, an exhaust duct 16 is provided above the air supply duct 14 inside each of the furnace sections: the left wall 12C and the right wall 12D.

[0031] Hot air generated by an air supply device 70 (not shown) is supplied into each air supply duct 14. The air supply device 70 is composed of, for example, an air supply fan, an air supply filter connected to the intake side of the air supply fan, and a burner connected to the discharge side of the air supply fan. The air drawn in by the air supply fan is filtered by the air supply filter and heated to a predetermined temperature by the burner before being supplied into each air supply duct 14.

[0032] One or more discharge nozzles 18 are installed on the inside of each air supply duct 14. Each discharge nozzle 18 is cylindrical, and the inside of each air supply duct 14 and the inside of the drying oven body 12 are connected through the inside of each discharge nozzle 18. The high-temperature air supplied to each air supply duct 14 is blown into the drying oven body 12 through each discharge nozzle 18, and then exhausted outside the drying oven body 12 through each exhaust duct 16.

[0033] A thermal camera 80A is mounted on the outside of the drying oven body 12. The thermal camera 80A is installed through a glass surface that penetrates a portion of the left wall 12C of the drying oven body 12, and photographs the automobile body B, transmitting the captured image data to the control device 100. Note that the location where the thermal camera 80A is installed is not limited to the left side, but can be any location from which the automobile body B can be photographed.

[0034] (Spray nozzle) As shown in Figure 3A, the ejection nozzle 18 is attached to the inner (lower in Figure 3A) wall 14A of the air supply duct 14 via a gear cover 20. The gear cover 20 is formed in a roughly box shape from, for example, a press-formed metal plate. A circular through-hole (not shown in the reference numerals) is formed in the inner (lower in Figure 3A) wall of the gear cover 20, and the ejection nozzle 18 is attached to the edge of this through-hole.

[0035] The ejection nozzle 18 has a bellows-shaped base 18A on the gear cover 20 side and a truncated cone-shaped tip 18B on the opposite side of the gear cover 20. This ejection nozzle 18 is made of a heat-resistant and flexible material and is deformable at the base 18A. The tip 18B of the ejection nozzle 18 is formed to decrease in diameter towards the tip side (opposite the base 18A), and the tip of the tip 18B is the hot air outlet 18C.

[0036] A gear cover 22 is provided on the outer (upper in Figure 3A) wall 14B of the air supply duct 14. The gear cover 22 is located on the opposite side from the gear cover 20 via the air supply duct 14. This gear cover 22 is formed in a roughly box shape from, for example, a press-formed metal plate. A rotary motor 24, which is a rotation drive source, and a tilt motor 26, which is a tilt drive source, are mounted on the outer (upper in Figure 3A) side of the gear cover 22. The rotary motor 24 and the tilt motor 26 are gear motors and are connected to the ejection nozzle 18 via a drive force transmission unit 28.

[0037] The drive force transmission unit 28 includes a rotation transmission unit 30 that transmits the driving force of the rotary motor 24 to the ejection nozzle 18 and rotates the ejection nozzle 18, and a tilt transmission unit 32 that transmits the driving force of the tilt motor 26 to the ejection nozzle 18 and tilts the ejection nozzle 18. The rotation transmission unit 30 is composed of a pinion 34 and gear 36A arranged in the gear cover 22, a pinion 40 and gear 42A arranged in the gear cover 20, and a rotating shaft 46. The tilt transmission unit 32 is composed of a tilt shaft 48, an orthogonal gearbox 50, and a lever 54. The orthogonal gearbox 50 is, for example, a bevel box.

[0038] The pinion 34, the pinion 40, the rotating shaft 46, and the output shaft of the rotating motor 24 are arranged coaxially with respect to each other, with their respective axial directions perpendicular to the transport direction of the automobile body B (up and down in Figure 3A). The rotating shaft 46 penetrates the inner wall 14A and the outer wall 14B of the air supply duct 14, and is rotatably supported by bearings 45 and 47 on each wall 14A and 14B.

[0039] The rotating shaft 46 is connected to the output shaft of the rotating motor 24 so as to be coaxial and integrally rotatable. A pinion 34 is fixed to one end of the rotating shaft 46 (the end on the outside of the furnace), and a pinion 40 is fixed to the other end of the rotating shaft 46 (the end on the inside of the furnace). A gear 36A meshes with the pinion 34, and a gear 42A meshes with the pinion 40. As an example, a universal joint is provided in the middle of the rotating shaft 46, but a configuration without a universal joint in the middle of the rotating shaft 46 is also possible.

[0040] Gear 36A, gear 42A, tilting shaft 48, and the output shaft of the tilting motor 26 are arranged coaxially with each other, with their respective axial directions perpendicular to the transport direction of the automobile body B (up and down direction in Figure 3A). Gear 36A constitutes the outer ring of bearing 36, and the inner ring of bearing 36 is fixed to the wall portion 14B of the air supply duct 14 via a ring member 38 arranged concentrically with bearing 36. This gear 36A is meshed with pinion 34. Similarly, gear 42A constitutes the outer ring of bearing 42, and the inner ring of bearing 42 is fixed to the wall portion 14A of the air supply duct 14 via a ring member 44 arranged concentrically with bearing 42.

[0041] When the rotary motor 24 rotates, the rotary shaft 46, pinion 34, and pinion 40 rotate, and gears 36A and 42A also rotate. A rotating part 22A, which is part of the gear cover 22, is attached to gear 36A, and a rotating part 20A, which is part of the gear cover 20, is attached to gear 42A. The rotating part 22A is capable of rotating integrally with gear 36A, and the rotating part 20A is capable of rotating integrally with gear 42A. A tilting motor 26 is attached to the rotating part 22A, and a spray nozzle 18 is attached to the rotating part 20A. The tilting motor 26 and the spray nozzle 18 rotate integrally with gears 36A and 42A.

[0042] The tilting shaft 48 is connected to the output shaft of the tilting motor 26 so as to be able to rotate integrally with it. The tilting shaft 48 penetrates the inner wall 14A and the outer wall 14B of the air supply duct 14 and is arranged coaxially with the gears 36A and 42A. As an example, a universal joint is provided in the middle of the tilting shaft 48, but a configuration without a universal joint is also possible. The input shaft of the orthogonal gearbox 50 is fixed coaxially with one end of the tilting shaft 48 (the end on the inner side of the furnace).

[0043] The orthogonal gearbox 50 is attached to gear 42A (i.e., bearing 42) via a pair of stays 52 that extend radially to the gear 42A. The pair of stays 52 are made of, for example, L-shaped metal plates. The output shaft of the orthogonal gearbox 50 is oriented perpendicular to the axis of the inclined shaft 48. A lever 54 is fixed to this output shaft.

[0044] The lever 54 is located within the base 18A of the ejection nozzle 18. The lever 54 is made of a metal plate, for example, bent into a roughly U-shape, and has a main body 54A fixed to the output shaft of the orthogonal gearbox 50, and a shaft portion 54B extending from the center of the main body 54A toward the opposite side from the orthogonal gearbox 50. The shaft portion 54B is located coaxially with the ejection nozzle 18. This shaft portion 54B is connected to a stay 56 extending radially from the ejection nozzle 18 via a pair of bearings (not shown). The stay 56 is fixed to the intermediate portion of the ejection nozzle 18 between the base 18A and the tip 18B.

[0045] The rotational driving force of the tilt motor 26 is transmitted to the ejection nozzle 18 via the tilt shaft 48, the orthogonal gearbox 50, the lever 54, and the stay 56 (see arrow R1 in Figure 4A). As a result, the ejection nozzle 18 is configured to tilt, as shown in Figures 4A and 4B. In this case, the lever 54 swings around the output shaft of the orthogonal gearbox 50, causing the ejection nozzle 18 to deform at its base 18A, and the axis of the tip 18B of the ejection nozzle 18 to tilt with respect to the axis of the tilt shaft 48. The tilt shaft 48, the orthogonal gearbox 50, the lever 54, and the stay 56 constitute the tilt transmission unit 32.

[0046] Furthermore, in this embodiment, the rotational driving force of the rotary motor 24 is transmitted to the ejection nozzle 18 via the rotary shaft 46, pinion 40, gear 42A, stay 52, orthogonal gearbox 50, lever 54, and stay 52 (see arrow R2 in Figure 4A). As a result, the ejection nozzle 18 rotates around the axis of the gear 42A (see arrow R3 in Figure 4B). Therefore, by rotating the rotary motor 24 with the ejection nozzle 18 tilted as shown in Figure 4B, the direction of hot air discharge from the ejection nozzle 18 (see arrow F in Figure 4A) can be changed along the entire circumferential direction of the gear 42A.

[0047] The rotary motor 24 and tilting motor 26 described above are electrically connected to the control device 100 (not shown except in Figure 3A). In this embodiment, the rotary motor 24 and tilting motor 26 are used as the drive source for rotation and tilting, respectively, but this is not the only option. The drive source may be an air-driven source such as an air cylinder driven by compressed air.

[0048] (Operation of the drying oven) When the car body B is transported by the painting trolley T to a predetermined position (for example, the starting position of the heating area), the specifications of the car body B are transmitted to the control device 100. The control device 100 then controls a plurality of nozzle drive devices 23 according to the information indicating the type of car body B, and adjusts the direction of each ejection nozzle 18. Furthermore, the control device 100 controls the hot air supplied from the air supply duct 14 to be blown onto the car body B via the ejection nozzles 18. The atmosphere inside the drying oven is discharged through the exhaust duct 16.

[0049] While the automobile body B passes through the heating area, the thermal camera 80A transmits a thermal distribution image showing the surface temperature of the automobile body B to the control device 100. The control device 100 then controls multiple nozzle drive devices 23 according to the temperature data obtained from the received thermal distribution image, correcting the rotation angle and tilt of each ejection nozzle 18. The control device 100 also controls the air supply fan of the air supply device 70 to change the airflow volume and velocity of the hot air ejected from the ejection nozzles 18.

[0050] The control device 100 controls the flame of the burner of the air supply device 70 based on temperature information obtained by a temperature sensor (not shown) that measures the temperature inside the paint drying oven 10, and adjusts the temperature of the hot air supplied so that the temperature inside the oven reaches a predetermined temperature.

[0051] Then, once the car body B has finished passing through the heating area on the painting trolley T, the control device 100 adjusts the direction of the spray nozzle 18 and ends the blowing of hot air onto the car body B. The above process is repeated each time the painted car body B enters the heating area.

[0052] (Control device) Figure 5 is a block diagram showing the hardware configuration of the control device 100 according to the first embodiment. As shown in Figure 5, the control device 100 is composed of a CPU (Central Processing Unit) 110, ROM (Read Only Memory) 120, RAM (Random Access Memory) 130, storage 140, communication I / F 150, and input / output I / F 160. Each component is connected to the others so as to be able to communicate with each other via a bus 170. Note that the hardware configuration of the learning device 200 is the same as that of the control device 100, so its description is omitted.

[0053] The CPU 110 is a central processing unit that executes various programs and controls various parts. The ROM 120 stores various programs and data. The RAM 130 temporarily stores programs or data as a working area. That is, the CPU 110 reads a program from the ROM 120 and executes the program using the RAM 130 as a working area. Various programs and data may also be stored in the storage 140, which will be described later.

[0054] Storage 140 is a non-temporary recording medium consisting of an HDD (Hard Disk Drive), SSD (Solid State Drive), or flash memory, etc., and stores various programs, including the operating system, and various data.

[0055] The communication interface 150 is an interface for communicating with other devices. Specifically, the communication interface communicates with the learning device 200 and various devices of the paint drying oven 10 via a network. For this communication, a wired communication standard such as Ethernet® or FDDI, or a wireless communication standard such as 4G, 5G, or Wi-Fi® may be used.

[0056] The I / O I / F160 is an interface for connecting to input / output devices and is used for inputting and outputting various types of information.

[0057] (Functions of the control device) Figure 6 is a block diagram showing an example of the functional configuration of the control device 100 according to the first embodiment. In this embodiment, the control device 100 functions as an acquisition unit 110A, a control unit 110B, and a model acquisition unit 110C when the CPU 110 executes the control program 121.

[0058] The acquisition unit 110A has the function of acquiring work information indicating the type of work. In this embodiment, the acquisition unit 110A acquires specification information of the automobile body B when the automobile body B is transported to a predetermined location. The acquisition unit 110A may acquire the specification information from a transmitter mounted on the automobile body B, or it may acquire it from a database (not shown) that stores specification information for each work.

[0059] Furthermore, the acquisition unit 110A has the function of acquiring measurement data. Specifically, the acquisition unit 110A acquires temperature data measured in the paint drying oven 10 during the drying of the workpiece. The acquisition unit 110A acquires, for example, temperature data of each part of the automobile body B shown in the heat distribution image acquired by the thermal camera 80A. The acquisition unit 110A also acquires environmental data measured outside the paint drying oven 10 during the drying of the workpiece. The acquisition unit 110A acquires, for example, the temperature and humidity outside the paint drying oven 10 measured by a temperature sensor and a humidity sensor (not shown). The acquisition unit 110A may acquire the heat distribution image itself as temperature data.

[0060] The control unit 110B has the function of controlling various devices of the paint drying oven 10. Specifically, the control unit 110B controls a plurality of nozzle drive devices 23 and air supply devices 70 in the paint drying oven 10.

[0061] In this embodiment, when the automobile body B is transported to a predetermined position, the control unit 110B changes the orientation of the ejection nozzle 18 to a direction corresponding to the specification information of the automobile body B acquired by the acquisition unit 110A.

[0062] Furthermore, the control unit 110B controls the nozzle drive device 23 of the spray nozzle 18, which is set to a predetermined orientation, to correct the orientation of the spray nozzle 18. Specifically, the control unit 110B corrects the rotation angle and inclination of the spray nozzle 18 based on the temperature data of each part of the automobile body B acquired by the acquisition unit 110A during drying. For example, the control unit 110B controls the nozzle drive device 23 based on control data obtained by inputting the temperature data acquired by the acquisition unit 110A into the learned model 142A, and corrects the rotation angle and inclination of the spray nozzle 18. Alternatively, the control unit 110B may transmit the temperature data acquired by the acquisition unit 110A to the learning device 200, and control the nozzle drive device 23 based on control data obtained by inputting the temperature data into the learned model 242A.

[0063] Furthermore, the control unit 110B controls the air supply device 70. Specifically, the control unit 110B acquires the internal temperature of the paint drying oven 10 and provides feedback control to the temperature of the hot air supplied from the air supply device 70 so that the internal temperature of the oven reaches a predetermined temperature. The control unit 110B may also control the air supply device 70 using control data output from the trained model 142A.

[0064] The model acquisition unit 110C has the function of acquiring a trained model. In this embodiment, the model acquisition unit 110C acquires the trained model 242A from the learning device 200 and stores it as the trained model 142A.

[0065] (Functions of the learning device) Figure 7 is a block diagram showing an example of the functional configuration of the learning device 200 according to the first embodiment. In this embodiment, the learning device 200 functions as a performance acquisition unit 210A, an evaluation acquisition unit 210B, a pre-processing unit 210C, a learning unit 210D, and a storage unit 210E when the CPU 210 executes the learning program 221.

[0066] The performance acquisition unit 210A has the function of acquiring performance data acquired during the drying of the workpiece. The performance data includes a series of temperature data, environmental data, and control data acquired during the drying of the workpiece. In this embodiment, the performance acquisition unit 210A acquires temperature data, environmental data, and control data output from the trained model 142A during the drying of the automobile body B from the control device 100. The performance data may also include workpiece specification information.

[0067] The evaluation acquisition unit 210B has the function of acquiring evaluation data. For example, the evaluation acquisition unit 210B acquires the inspection results of the dried coating as evaluation data. In this embodiment, the evaluation acquisition unit 210B acquires the coating hardness of the dried coating on the automobile body B.

[0068] The preprocessing unit 210C has the function of generating training data. For example, the preprocessing unit 210C generates retraining data 243A by associating the performance data acquired by the performance acquisition unit 210A with the evaluation data acquired by the evaluation acquisition unit 210B.

[0069] Furthermore, the preprocessing unit 210C has a function for selecting training data. For example, the preprocessing unit 210C labels training data whose evaluation data exceeds a standard value as normal training data.

[0070] The learning unit 210D has the function of training a pre-trained model using a machine learning algorithm. Specifically, the learning unit 210D trains the pre-trained model 242A to learn from the accumulated training data 241A and output control data. The machine learning method performed by the learning unit 210D is not particularly limited as long as it is supervised learning using training data. Examples of models or algorithms used in supervised learning include regression analysis, decision trees, support vector machines, neural networks, ensemble learning, and random forests. Alternatively, classification may be performed in advance, and then supervised learning may be performed for each class. In this case, the classification may be either supervised or unsupervised. The training data used when training the pre-trained model 242A may include retraining data 243A.

[0071] Here, we will explain the correlation between training data and control data. The training data in this embodiment may include temperature data, environmental data, control data, specification information, and evaluation data.

[0072] This section explains the case where temperature data is included in the training data. In this embodiment, the temperature data represents the temperature of various parts of the automobile body B. The temperature of each part of the automobile body B changes depending on the direction of the ejection nozzle 18, which is controlled by the control data. For example, the heating rate of the part of the automobile body B that the ejection nozzle 18 is pointing towards will increase. Thus, there is a correlation between the temperature data and the control data.

[0073] This section explains the case where environmental data is included in the training data. In this embodiment, the environmental data represents the temperature and humidity outside the paint drying oven 10. The heating rate of each part of the automobile body B changes according to the external environment. Furthermore, the temperature of each part of the automobile body B changes depending on the direction of the spray nozzle 18, which is controlled by the control data. Thus, there is a correlation between the environmental data and the control data.

[0074] This section explains the case where the learning data includes specification information. In this embodiment, the specification information represents the vehicle type, paint color, and dimensions of the automobile body B. The heating rate of each part of the automobile body B changes according to the vehicle type, paint color, and dimensions of the automobile body B. Furthermore, the temperature of each part of the automobile body B changes depending on the direction of the ejection nozzle 18, which is controlled by the control data. Thus, there is a correlation between the specification information and the control data.

[0075] This section explains the case where evaluation data is included in the training data. In this embodiment, the evaluation data represents the hardness of the coating after drying. The hardness of the coating on the automobile body B changes according to the baking drying process (heating rate, drying time, etc.). The baking drying process of the automobile body B changes depending on the direction of the ejection nozzle 18, which is controlled by the control data. Thus, there is a correlation between the evaluation data and the control data.

[0076] Furthermore, the learning unit 210D has a function to retrain the trained model using a machine learning algorithm. Specifically, the learning unit 210D retrains the trained model 242A using the retraining data 243A. The retraining methods performed by the learning unit 210D include batch learning, mini-batch learning, or online learning.

[0077] The storage unit 210E has the function of storing the trained model. In this embodiment, the storage unit 210E deploys the trained model 242A, which has been trained or retrained by the learning unit 210D, to the control device 100 as trained model 142A. The control device 100 then stores the deployed trained model 142A in the storage 140. The trained model 142A may be the same model as the trained model 242A, or it may be a lightweight version of the trained model 242A. Methods for lightweighting include distillation, pruning, and quantization.

[0078] (flowchart) Figure 8 is a flowchart showing an example of the learning process flow of the learning device 200 according to the first embodiment. This learning process is performed by the CPU 210 of the learning device 200 reading the learning program 221 from the ROM 220 and executing it in the RAM 230 (see Figure 5). The learning process is performed, for example, when a user operates the learning device 200 and the learning device 200 receives a learning process instruction signal.

[0079] In step S100 of Figure 8, the CPU 210 acquires the training data 241A. Specifically, the CPU 210 acquires the training data 241A stored in the storage 240.

[0080] In step S101, the CPU 210 generates a trained model 242A. Specifically, the CPU 210 uses a supervised learning method to train a machine learning model for outputting control data from temperature data based on the training data 241A acquired in step S100, thereby generating the trained model 242A.

[0081] In step S102, the CPU 210 deploys the trained model 242A. Specifically, the CPU 210 deploys the trained model 242A to the control device 100 as trained model 142A. Then, the CPU 210 terminates the training process.

[0082] Figure 9 is a flowchart showing an example of the control process flow of the control device 100 according to the first embodiment. This control process is performed by the CPU 110 of the control device 100 reading the control program 121 from the ROM 120 and executing it in the RAM 130 (see Figure 5). The control process is performed, for example, when the painted automobile body B is transported to the heating area of ​​the paint drying oven 10.

[0083] In step S200 of Figure 9, the CPU 110 acquires specification information of the painted automobile body B. The CPU 110 acquires specification information of the automobile body B from, for example, a transmitter (not shown) mounted on the automobile body B.

[0084] In step S201, the CPU 110 changes the rotation angle and inclination of the ejection nozzle 18 based on the acquired specification information. Specifically, the CPU 110 controls each of the nozzle drive devices 23 of the multiple ejection nozzles 18 so that the rotation angle and inclination correspond to the specification information acquired in step S200.

[0085] In step S202, the CPU 110 acquires temperature data of the painted surface of the automobile body B. Specifically, the CPU 110 acquires temperature data for each part of the painted surface of the automobile body B from the heat distribution image of the thermal camera 80A.

[0086] In step S203, the CPU 110 determines whether or not it has acquired environmental data from outside the paint drying oven 10. If the CPU 110 determines that it has acquired environmental data from outside the paint drying oven 10 (step S203: YES), it proceeds to step S204. On the other hand, if the CPU 110 determines that it has not acquired environmental data from outside the paint drying oven 10 (step S203: NO), it proceeds to step S205.

[0087] In step S204, the CPU 110 inputs the acquired temperature data and environmental data into the trained model 142A. Specifically, the CPU 110 inputs the temperature data acquired in step S202 and the environmental data acquired in step S203 into the trained model 142A stored in the storage 140.

[0088] In step S205, the CPU 110 inputs the acquired temperature data into the trained model 142A. Specifically, the CPU 110 inputs the temperature data acquired in step S202 into the trained model 142A stored in the storage 140. The CPU 110 may also input predetermined default values ​​of environmental data into the trained model 142A along with the acquired temperature data.

[0089] In step S206, the CPU 110 acquires the control data output from the trained model 142A.

[0090] In step S207, the CPU 110 corrects the rotation angle and inclination of the multiple discharge nozzles 18 based on the control data, and changes the air volume and air velocity. Specifically, the CPU 110 corrects the rotation angle and inclination of the multiple discharge nozzles 18 and changes the air volume and air velocity by controlling the nozzle drive unit 23 and the air supply unit 70 based on the control data acquired in step S206.

[0091] In step S208, the CPU 110 obtains the internal temperature of the paint drying oven 10. The CPU 110 obtains the internal temperature of the paint drying oven 10 from a temperature sensor (not shown) installed inside the paint drying oven 10.

[0092] In step S209, the CPU 110 controls the supply air temperature so that the acquired furnace temperature reaches a predetermined temperature. For example, the CPU 110 provides feedback control to the temperature of the hot air supplied from the supply air device 70 so that the furnace temperature acquired in step S208 reaches 180 degrees.

[0093] In step S210, the CPU 110 determines whether the car body B has passed through the heating area. If the CPU 110 determines that the car body B has passed through the heating area (step S210: YES), it terminates the control process. On the other hand, if the CPU 110 determines that the car body B has not passed through the heating area (step S210: NO), it returns to step S202.

[0094] Figure 10 is a flowchart showing an example of the relearning process flow of the learning device 200 according to the first embodiment. This relearning process is performed by the CPU 210 of the learning device 200 reading the learning program 221 from the ROM 220 and executing it in the RAM 230 (see Figure 5). The relearning process is performed, for example, when a user operates the learning device 200 and the learning device 200 receives a signal to initiate the relearning process.

[0095] In step S300 of Figure 10, the CPU 210 acquires the retraining data 243A. Specifically, the CPU 210 acquires the retraining data 243A stored in the storage 240.

[0096] In step S301, the CPU 210 selects one actual data set from among multiple actual data sets included in the retraining data 243A.

[0097] In step S302, the CPU 210 obtains the coating hardness corresponding to the selected historical data.

[0098] In step S303, the CPU 210 determines whether the coating hardness exceeds a standard value. For example, the CPU 210 determines whether the coating hardness exceeds 2H. If the CPU 210 determines that the coating hardness exceeds the standard value (step S303: YES), it proceeds to step S304. On the other hand, if the CPU 210 determines that the coating hardness does not exceed the standard value (step S303: NO), it proceeds to step S305.

[0099] In step S304, the CPU 210 successfully labels the selected historical data.

[0100] In step S305, the CPU 210 labels the selected historical data as abnormal.

[0101] In step S306, the CPU 210 determines whether all of the multiple performance data have been labeled. If the CPU 210 determines that all of the multiple performance data have been labeled (step S306: YES), it proceeds to step S307. On the other hand, if the CPU 210 determines that not all of the multiple performance data have been labeled (step S306: NO), it returns to step S301.

[0102] In step S307, the CPU 210 retrains the trained model 242A using properly labeled historical data.

[0103] In step S308, the CPU 210 stores the retrained trained model 242A. Specifically, the CPU 210 stores the trained model 242A, which was retrained in step S307, as trained model 142A in the storage 140 of the control device 100. Then, the CPU 210 terminates the retraining process.

[0104] (Modified version of the first embodiment) The learning device 200, a modified version of the first embodiment, stores the actual data as retraining data 243A when the number of corrections to the rotation angle and inclination of the ejection nozzle 18 exceeds a predetermined number. The differences from the first embodiment are described below. As other aspects are the same as the first embodiment, detailed explanations are omitted.

[0105] (Function of the learning device for modification) The pre-processing unit 210C in Figure 7 stores as retraining data actual data where the number of corrections for the rotation angle and inclination of the spray nozzle 18 based on control data acquired during the drying of the automobile body B exceeds a predetermined number.

[0106] Figure 11 is a flowchart showing an example of the preprocessing flow for the relearning data 243A according to a modified example of the first embodiment. This preprocessing is performed by the CPU 210 of the learning device 200 reading the learning program 221 from the ROM 220 and executing it in the RAM 230 (see Figure 5). Preprocessing is, for example, a process that is performed before the relearning process described above is executed.

[0107] In step S400 of Figure 11, the CPU 210 acquires multiple historical data.

[0108] In step S401, the CPU 210 selects one performance data from multiple performance data. Specifically, the CPU 210 selects one performance data from multiple performance data acquired in step S400.

[0109] In step S402, the CPU 210 obtains the number of corrections for the rotation angle and tilt of the ejection nozzle 18 from the selected historical data.

[0110] In step S403, the CPU 210 determines whether the number of corrections obtained exceeds a predetermined number. If the CPU 210 determines that the number of corrections obtained exceeds a predetermined number (step S403: YES), it proceeds to step S404. On the other hand, if the CPU 210 determines that the number of corrections obtained does not exceed a predetermined number (step S403: NO), it proceeds to step S405.

[0111] In step S404, the CPU 210 labels the selected historical data as retraining data.

[0112] In step S405, the CPU 210 determines whether all of the multiple performance data have been labeled. If the CPU 210 determines that all of the multiple performance data have been labeled (step S405: YES), it proceeds to step S406. On the other hand, if the CPU 210 determines that not all of the multiple performance data have been labeled (step S405: NO), it returns to step S401.

[0113] In step S406, the CPU 210 stores the actual data labeled as retraining data as retraining data 243A. Specifically, the CPU 210 stores the actual data labeled as retraining data in step S404 as retraining data 243A in storage 240. Then, the CPU 210 finishes the preprocessing.

[0114] [Second Embodiment] The control device 100 of the second embodiment acquires temperature data indicating the partial surface temperature of the automobile body B measured by the high-temperature thermopile 80B. Furthermore, the control device 100 of the second embodiment controls the ejection nozzle 18 using the trained model 142A learned by the learning device 200 of the first embodiment. The differences from the first embodiment are described below. Other aspects are the same as in the first embodiment, so detailed explanations are omitted.

[0115] Figure 12 is a schematic block diagram of the control system 1 according to the second embodiment. As shown in Figure 12, the learning device 200 includes learning data 241A, a trained model 242A, and retraining data 243B.

[0116] The retraining data 243B in this embodiment is training data that includes control data obtained by inputting temperature data indicating the temperature measured by the high-temperature thermopile 80B into the trained model 142A.

[0117] Figure 13 is a longitudinal cross-sectional view showing a paint drying oven 10 according to the second embodiment. As shown in Figure 13, the paint drying oven 10 of the second embodiment is equipped with a high-temperature thermopile 80B and a signal converter 81B connected to the high-temperature thermopile 80B by a wire, instead of the thermal camera 80A of the first embodiment. The high-temperature thermopile 80B of this embodiment is installed in the upper and central parts of the inner wall of the air supply duct 14 on the left side inside the paint drying oven 10. The signal converter 81B is installed on the outside of the drying oven body 12, connected to each of the high-temperature thermopiles 80B. Note that the location where the high-temperature thermopile 80B and signal converter 81B are installed is not limited to the above location, but may be any location where the temperature of a part of the automobile body B can be measured.

[0118] (Functions of the control device) The acquisition unit 110A in Figure 6 acquires the surface temperature of a part of the automobile body B by receiving an electrical signal output from the high-temperature thermopile 80B of the paint drying oven 10 via a signal converter 81B.

[0119] The model acquisition unit 110C has the function of acquiring learned models for different measuring devices. In this embodiment, the model acquisition unit 110C acquires a learned model 142A for the thermal camera 80A. The model acquisition unit 110C may also be configured to adjust the learned model 142A so that control data is output when temperature data indicating the temperature measured by the high-temperature thermopile 80B is input.

[0120] The control unit 110B has a function to correct the orientation of the ejection nozzle 18 based on a learned model for a different measuring device. In this embodiment, the control unit 110B inputs temperature data indicating the temperature of the automobile body B measured by the high-temperature thermopile 80B to the learned model 142A acquired by the model acquisition unit 110C. The control unit 110B then controls the nozzle drive device 23 based on the control data obtained from the learned model 142A to correct the rotation angle and inclination of the ejection nozzle 18. Alternatively, the control unit 110B may adjust the temperature data indicating the temperature measured by the high-temperature thermopile 80B to temperature data corresponding to the learned model 142A and input it to the learned model 142A.

[0121] In other words, the control device 100 of the second embodiment corrects the rotation angle and inclination of the ejection nozzle 18 based on control data obtained by inputting temperature data acquired by the high-temperature thermopile 80B into a trained model 142A that has been trained using training data including temperature data acquired by the thermal camera 80A. Therefore, the control device 100 of this embodiment eliminates the need to train a new trained model.

[0122] (Functions of the learning device) The learning unit 210D in Figure 7 retrains the trained model 242A using the retraining data 243B. Known methods such as transfer learning and fine-tuning may be used for retraining. Therefore, the learning device 200 of this embodiment allows a trained model, trained using learning data accumulated in different environments, to be adapted to a new environment, thereby improving its accuracy.

[0123] [Third Embodiment] The control device 100 of the third embodiment acquires the temperature difference between the supply air temperature (hereinafter also simply referred to as supply air temperature) measured near the ejection nozzle 18 of the paint drying oven 10, which is measured by a resistance thermometer 80C, and the exhaust temperature (hereinafter also simply referred to as exhaust temperature) measured near the exhaust port of the exhaust duct 16. The differences from the first embodiment are described below. As other aspects are the same as in the first embodiment, a detailed explanation is omitted.

[0124] Figure 14 is a schematic block diagram of the control system 1 of the third embodiment. As shown in Figure 14, the control device 100 includes a trained model 142C, and the learning device 200 includes training data 241C, a trained model 242C, and retraining data 243C.

[0125] The trained model 142C in this embodiment is a trained model obtained by deploying the trained model 242C, which was trained on the learning device 200, to the control device 100.

[0126] The training data 241C in this embodiment is training data that includes temperature data showing the temperature difference between the supply air temperature and the exhaust air temperature measured by the resistance thermometer 80C. The training data 241C also includes control data when the direction of the ejection nozzle 18 is corrected so that the temperature difference between the supply air temperature and the exhaust air temperature is large. The trained model 242C is a machine learning model trained using the training data 241C as training data. The retrained data 243C is training data that includes control data output from the trained model 142C.

[0127] Figure 15 is a longitudinal cross-sectional view showing a paint drying oven 10 according to the third embodiment. As shown in Figure 15, the paint drying oven 10 of the third embodiment is equipped with a plurality of resistance thermometers 80C instead of the thermal camera 80A described in the first embodiment, and a temperature sensing junction 81C connected to each of the resistance thermometers 80C by a wire. The temperature sensing junction 81C in this embodiment is installed near the hot air outlet 18C of the ejection nozzle 18 installed in the lower left corner and to the left of the ceiling. The temperature sensing junction 81C is also installed near the exhaust port installed in the exhaust duct 16 in the upper left corner. Note that the location where the temperature sensing junction 81C is installed is not limited to the above locations, but may be any location where the hot air supply temperature or exhaust temperature can be measured.

[0128] (Functions of the control device) The acquisition unit 110A in Figure 6 acquires the supply air temperature and exhaust air temperature in the paint drying oven 10 by receiving the electrical signal output from the resistance thermometer 80C of the paint drying oven 10. The acquisition unit 110A then acquires temperature data showing the temperature difference between the supply air temperature and the exhaust air temperature.

[0129] The control unit 110B corrects the orientation of the ejection nozzle 18 based on temperature data indicating the temperature difference between the supply air temperature and the exhaust air temperature. Specifically, the control unit 110B controls the nozzle drive device 23 based on control data obtained by inputting the temperature data indicating the temperature difference acquired by the acquisition unit 110A into the learned model 142C, thereby correcting the rotation angle and inclination of the ejection nozzle 18.

[0130] In other words, the control device 100 of the third embodiment acquires temperature data indicating the temperature difference between the supply air temperature and the exhaust air temperature, and corrects the rotation angle and inclination of the ejection nozzle 18 based on the acquired temperature data. Therefore, according to the control device 100 of this embodiment, by correcting the orientation of the ejection nozzle 18 so that the temperature difference between the supply air temperature and the exhaust air temperature is increased, the amount of heat supplied to the workpiece is increased, and the coating film formed on the workpiece can be dried efficiently.

[0131] (Functions of the learning device) The learning unit 210D in Figure 7 retrains the trained model 242C using the retraining data 243C. In this embodiment, the retraining data 243C is training data that includes control data based on the temperature difference between the supply air temperature and the exhaust air temperature. Therefore, according to the learning device 200 of this embodiment, the trained model 242C can be retrained to output control data that increases the temperature difference between the supply air temperature and the exhaust air temperature.

[0132] [Other embodiments] In the first embodiment, control data was acquired using a trained model 142A. However, the control device 100 of this embodiment may also feedback control the rotation angle and inclination of the spray nozzle 18 so as to heat the part of the automobile body B where the temperature indicated by the temperature data is lower than a predetermined temperature.

[0133] In the first embodiment, the control device 100 changed the orientation of the ejection nozzle 18 based on the specification information of the automobile body B. However, the control device is not limited to this, and the specification information of the automobile body B may be estimated by performing image recognition processing on an image of the automobile body B, and the orientation of the ejection nozzle 18 may be changed accordingly. According to the control device 100 of this embodiment, the orientation of the ejection nozzle 18 can be changed even if the specification information cannot be obtained.

[0134] In the first embodiment, the control device 100 controlled the direction of the ejection nozzle 18 based on control data obtained by inputting measurement data into a learned model 142A. However, the control device 100 is not limited to this, and may also obtain control data by inputting specification information of the automobile body B into the learned model 142A. According to the control device 100 of this embodiment, the direction of the ejection nozzle 18 can be controlled by control data estimated based on the specification information.

[0135] In the first embodiment, the learning device 200 trained a pre-trained model 242A through supervised learning using training data 241A. However, the learning device 200 of this embodiment may also train the pre-trained model 242A through reinforcement learning using time-series data showing how the temperature indicated by the temperature data rises to a predetermined temperature. For example, the learning device 200 trains the pre-trained model 242A using deep reinforcement learning, where the reward is set higher the shorter the time it takes for the temperature indicated by the temperature data to rise to the predetermined temperature. According to the learning device 200 of this embodiment, the pre-trained model 242A can be trained to output control data that can quickly raise the surface temperature of the automobile body B to a predetermined temperature.

[0136] In the first embodiment, the control device 100 changed the rotation angle and tilt of the ejection nozzle 18 by controlling the shafts provided in the nozzle drive device 23. Here, the control device 100 may control the order in which each shaft is moved. Specifically, the control device 100 may drive the rotary motor 24 first, and then the tilt motor 26. Alternatively, the control device 100 may drive the tilt motor 26 first, and then the rotary motor 24. Furthermore, the control device 100 may drive the rotary motor 24 and the tilt motor 26 simultaneously. The order and amount of change for changing the rotation angle and tilt may correspond to the measured temperature data. Specifically, the control device 100 may set the direction in which the outlet 18C faces while changing the orientation of the ejection nozzle 18 to correspond to the temperature data indicated by the surface temperature of the automobile body B. The control device 100 controls, for example, the order and amount of changes in the rotation angle and inclination so that hot air is directed to areas where the surface temperature has not yet risen to a predetermined value while the direction of the ejection nozzle 18 is being changed. According to the control device 100 of this embodiment, the paint film formed on the automobile body B can be dried more efficiently.

[0137] In the first embodiment, the case in which one automobile body B is being transported was described as an example. However, the control device 100 of this embodiment is not limited to this and may be applied when multiple automobile bodies B are being transported in succession. Furthermore, when multiple automobile bodies B are being transported in succession, the control device 100 of this embodiment may control each of the multiple nozzle drive devices 23 so that the multiple spray nozzles 18 face in directions corresponding to the multiple automobile bodies B. For example, when the first automobile body B is transported beyond the spraying range of one spray nozzle 18, the control device 100 controls the nozzle drive device 23 so that the direction of the one spray nozzle 18 is such that it can spray onto the second automobile body B. The control device 100 then applies the control of the nozzle drive devices 23 corresponding to the multiple automobile bodies B being transported in succession to each of the multiple nozzle drive devices 23. According to the control device 100 of this embodiment, each of the coating films formed on the multiple automobile bodies B can be dried more efficiently.

[0138] The learning data of the first embodiment may include specification information for automobile body B. However, the learning data is not limited to this and may include specification information for multiple automobile body B being transported in succession. For example, the learning data may include specification information for the automobile body B being transported immediately before and immediately after. By including specification information for multiple automobile body B being transported in succession in the learning data 241A of this embodiment, the trained model 242A can output control data for cases where automobile body B of different vehicle types are transported in succession. Therefore, the control device 100 of this embodiment can more efficiently dry each of the coatings formed on automobile body B of different vehicle types being transported in succession.

[0139] Furthermore, the configurations of the control device 100, learning device 200, and paint drying oven 10 described in the above embodiment are examples and may be modified as needed without departing from the main purpose.

[0140] The first acquisition unit, the second acquisition unit, and the third acquisition unit may be an acquisition unit 110A as a functional unit, or they may be hardware such as a communication I / F 150 or an input / output I / F 160. Furthermore, for example, the first acquisition unit may be a receiver for acquiring work information, the second acquisition unit may be a measuring device 80, and the third acquisition unit may be various sensors such as a temperature sensor and a humidity sensor.

[0141] Furthermore, the program processing flow described in the above embodiment is just one example, and unnecessary steps may be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[0142] Furthermore, the various processes that the CPU 110 of the control device 100 reads and executes in the above embodiment may be executed by various processors other than the CPU. Examples of such processors include PLDs (Programmable Logic Devices) such as FPGAs (Field-Programmable Gate Arrays) whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits that are processors with circuit configurations specifically designed to execute specific processes, such as ASICs (Application Specific Integrated Circuits). In addition, the various processes may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, multiple FPGAs, and a combination of a CPU and an FPGA). More specifically, the hardware structure of these various processors is an electrical circuit that combines circuit elements such as semiconductor elements. The same applies to the learning device 200 as described above.

[0143] Furthermore, although the above embodiments describe a configuration in which various programs are pre-stored (installed) on storage, the invention is not limited to this. Programs may be provided in the form of recordings on recording media such as CD-ROMs (Compact Disc Read Only Memory), DVD-ROMs (Digital Versatile Disc Read Only Memory), and USB (Universal Serial Bus) memory. Programs may also be provided in the form of downloads from external devices via a network. Moreover, the present invention can also be applied to programs and program products.

[0144] Furthermore, the operation of the processor in the above embodiment may not be performed by a single processor, but may be performed by multiple processors located in physically separate locations working together. Also, the order of the processor operations is not limited to the order described in the above embodiment, but may be changed as appropriate. [Explanation of Symbols]

[0145] 1. Control System 10 Drying oven 14. Air supply duct 16 Exhaust duct 18 Spray nozzles 23 Nozzle drive device 24-speed motor 26. Inclination motor 70 Air supply system 80 Measuring device 80A Thermal Camera 80B High-temperature thermopile 81B Signal Converter 80C resistance thermometer 81C Temperature sensing junction 100 Control device 110A Acquisition Department 110B Control Unit 110C Model Acquisition Unit 121 Control Program 142A Pre-trained model 142C Pre-trained Model 200 Learning Devices 210A Performance Acquisition Department 210B Evaluation Acquisition Unit 210C Pre-processing Unit 210D Learning Department 210E Storage Unit 221 Learning Programs 241A Training data 242A Pre-trained model 243A Retraining Data 243B Retraining Data 241C Training Data 242C Pre-trained Model 243C Retraining Data T Painted trolley B. Automobile Body

Claims

1. A first acquisition unit acquires work information indicating the type of work in a drying oven that performs a drying operation to dry a coating film formed on a workpiece, A control device comprising: a control unit that controls the orientation of a plurality of ejection nozzles that eject hot air for drying the coating film based on the work information acquired by the first acquisition unit, The system further includes a second acquisition unit that acquires temperature data indicating the temperature of the workpiece during the drying operation, The control device is characterized in that the control unit corrects the orientation of the ejection nozzle based on the temperature data acquired by the second acquisition unit.

2. The control unit, The system controls the spray nozzle based on control data obtained by inputting the temperature data acquired during drying to a trained model that has been trained by associating the accumulated temperature data with the evaluation data of the coating film. The control device according to claim 1.

3. The trained model is learned by associating the accumulated temperature data and evaluation data with accumulated environmental data representing the external environment of the drying oven. It includes a third acquisition unit for acquiring the aforementioned environmental data during drying, The control unit controls the ejection nozzle based on the control data obtained by inputting the acquired temperature data and environmental data to the trained model. The control device according to claim 2.

4. The second acquisition unit acquires temperature data indicating the temperature of the surface of the workpiece on which the coating film is formed, as the temperature data. The control device according to claim 1.

5. The second acquisition unit acquires temperature data, which represents the temperature difference between the supply air temperature and the exhaust air temperature of the hot air ejected from the ejection nozzle, as the temperature data. The control device according to claim 1.

6. The ejection nozzle is equipped with multiple axes for changing direction, The control unit controls at least one of the plurality of axes of the ejection nozzle. The control device according to claim 1.

7. The control unit controls the direction of each nozzle, as well as the amount of hot air ejected from each nozzle. The control device according to claim 1.

8. The second acquisition unit acquires the internal temperature of the drying oven, The control unit performs feedback control of the air supply device to adjust the temperature of the hot air so that the acquired furnace temperature reaches a predetermined temperature. The control device according to claim 1.

9. A drying oven that performs a drying operation to dry the coating film formed on the workpiece, A control device according to any one of claims 1 to 8, A drying apparatus equipped with the following features.

10. A performance acquisition unit acquires performance data, including control data output from a trained model that takes temperature data indicating the temperature of the workpiece during drying in a drying oven that performs a drying operation to dry a coating film formed on the workpiece as input, and outputs control data that corrects the direction of each of the multiple nozzles that emit hot air to dry the coating film. An evaluation unit that acquires evaluation data regarding the coating film of the workpiece, A learning unit that retrains the trained model using retraining data obtained by associating the aforementioned performance data with the aforementioned evaluation data, A learning device equipped with the following features.

11. The performance acquisition unit acquires, in addition to the control data, environmental data indicating the external environment of the drying oven during drying as performance data. The learning device according to claim 10.

12. The performance acquisition unit acquires performance data from the performance data when the correction of the direction of the ejection nozzle indicated by the control data exceeds a predetermined number of times. The learning device according to claim 10.

Images