JP7329754B2 - Repair welding system and repair welding method - Google Patents

Description

The present disclosure relates to repair welding systems and methods.

Patent Document 1 discloses a shape inspection apparatus for inspecting the shape of an object to be inspected using an imaging optical system, which includes projection means for projecting slit light onto the object to be inspected, and scanning the slit light to scan the object to be inspected. a point cloud data acquiring means for acquiring the three-dimensional shape of the object to be inspected as point cloud data based on the captured data of the shape lines formed sequentially; Cutting line setting means for setting a cutting line on the object to be inspected displayed based on the point cloud data according to an input; and cross-sectional shape calculation means for calculating the cross-sectional shape.

JP 2012-037487 A

The present disclosure provides a repair welding system and method capable of performing repair welding.

The present disclosure includes an inspection device that visually inspects a welded portion of a work, and a robot control device that controls a robot that welds the work. When it is determined that there is a defective part in the welded part of the work and that the defective part can be corrected, based on the state determination result of the welded part, correction of repair welding for correcting the defective part A type indicating a method and a parameter used for repairing the defective portion are generated, and the inspection device holds a program generation logic table that associates the state determination result with the type and parameter of the repair welding, and performs the state determination. Based on the result and the program generation logic table, not only the type and the parameters of the repair welding are generated, but also whether or not the repair welding can be automatically performed by the robot is determined. When the inspection device determines that the repair welding can be automatically performed by the robot, the robot control device determines the welding location of the workpiece according to the type and the parameters of the repair welding . a repair welding system that controls the robot to automatically correct the repair welding to

Further, the present disclosure provides a repair welding system including an inspection device that visually inspects a welded portion of a work and a robot control device that controls a robot that welds the work, wherein the inspection device detects the state of the welded portion of the work. If it is determined that a defective portion exists in the welded portion of the workpiece and that the defective portion can be corrected as the state determination, the defective portion is corrected based on the state determination result of the welded portion. A type indicating a repair welding correction method for performing repair welding and a parameter used to correct the defective portion are generated , and the inspection device generates a program generation logic table that associates the state determination result with the type and parameter of the repair welding. and not only generating the type and parameters of the repair welding based on the state determination result and the program generation logic table, but also whether it is possible for the robot to automatically correct the repair welding. If the inspection device determines that the repair welding can be automatically performed by the robot, the robot control device determines whether the repair welding is performed according to the type and the parameters. The present invention provides a repair welding method for controlling the robot so as to automatically correct the repair welding to the welding location of the workpiece.

According to the present disclosure , repair welding can be performed.

Schematic diagram illustrating an example use case of the automatic repair welding system 1000 according to the present disclosure. FIG. 2 is a diagram showing an internal configuration example of an automatic repair welding system 1000a regarding control of the inspection/repair welding robot MC23 according to Embodiment 1; Flowchart showing a first operation procedure example of automatic repair welding by the automatic repair welding system 1000a according to Embodiment 1 4 is a flowchart showing a second operation procedure example of automatic repair welding by the automatic repair welding system 1000a according to Embodiment 1. FIG. 10 is a diagram showing an internal configuration example of an automatic repair welding system 1000b regarding control of an inspection robot MC2 and a repair welding robot MC3 according to Embodiment 2; Flowchart showing an example of an operation procedure of automatic repair welding by the automatic repair welding system 1000b according to Embodiment 2 Flowchart showing an example of an operation procedure of automatic repair welding by the automatic repair welding system 1000a according to Modification 1 FIG. 10 is a diagram showing an internal configuration example of an automatic repair welding system 1000c related to control of the inspection/repair welding robot MC23 according to Modification 2; FIG. 9 is a diagram showing a configuration example of the data processing unit 35 in the internal configuration example of FIG. 8; Conceptual diagram showing comparison between master data and weld bead shape data to be inspected FIG. 10 is a diagram showing a program generation logic table, showing (a) an example of using the defect type and characteristic data as keys, and (b) an example of using the defect type and inspection score as keys.

(Background leading up to this disclosure)
The technique of Patent Literature 1 can determine the quality of the shape of the welded portion after the final welding is performed, using an appearance inspection device. However, if the shape is not good, the judgment of whether or not it can be corrected by re-welding (repair welding) and the re-welding for correction (repair welding) are performed by human welding workers. is currently doing. Therefore, there was a potential problem that the quality was not stable due to the skill level difference of workers and misjudgment.

Furthermore, in recent years, workpieces to be welded have been diversified. In proportion to the diversification of works, the burden on welders who perform repair welding is increasing.

In addition, even if the work after main welding can be repaired by repair welding, it is automatically determined whether this repair can be performed by automatic repair welding or whether welding by the welding operator (worker repair welding) is required. cannot be determined by

Furthermore, the criteria for the above various determinations and the correction method by repair welding may differ for each user environment such as a factory using welding technology. Therefore, it would be more preferable if the data generated in relation to the main welding/repair welding work in one's own factory or another factory could be effectively used to obtain more appropriate judgment criteria.

Therefore, in the present disclosure, it is automatically determined whether or not repair welding can be performed on the defective shape of the workpiece on which the main welding is performed, and further, whether automatic repair welding or operator repair welding is possible is automatically determined. do in As a result, the quality of the welded work can be improved and stabilized.

Further, by using the data generated in relation to the main welding/repair welding work of the own factory and other factories as feedback information, it is possible to adapt to the user environment and improve the accuracy and efficiency of the above determination.

Hereinafter, embodiments specifically disclosing the configuration and operation of an automatic repair welding system and an automatic repair welding method according to the present disclosure will be described in detail with reference to the drawings as appropriate. However, more detailed description than necessary may be omitted. For example, detailed descriptions of well-known matters and redundant descriptions of substantially the same configurations may be omitted. This is to avoid unnecessary verbosity in the following description and to facilitate understanding by those skilled in the art. It should be noted that the accompanying drawings and the following description are provided to allow those skilled in the art to fully understand the present disclosure and are not intended to limit the claimed subject matter.

FIG. 1 is a schematic diagram illustrating an example use case for an automated repair welding system 1000 according to the present disclosure. The automatic repair welding system 1000 according to the present disclosure inspects welded portions of the work Wk that have been fully welded based on information input by the user or preset welding information, and determines whether the welded portions are defective. It is a system that automatically performs correction welding (repair welding) of judged defective points. The system may perform final welding in addition to the inspection and repair welding described above.

The automatic repair welding system 1000 is roughly divided into three parts: a robot (RB0) used for welding and inspection of welding results, a controller for controlling the robot and its inspection function, and a host device for the controller. good.

More specifically, the automatic repair welding system 1000 includes a main welding robot MC1 that performs the main welding, an inspection robot MC2 that performs a visual inspection of the welded portion after the main welding, and a defective portion in the welded portion after the main welding. and a repair welding robot MC3 for performing repair welding if included. The welding system may also include a robot control device 2a, an inspection device 3, and a robot control device 2b as controllers for controlling the above-described various robots and inspection functions of the robots. Also, the automatic repair welding system 1000 may comprise a host device 1 for the controller described above. The host device 1 may be connected to the monitor MN1, the interface UI1, and the external storage ST.

Although not shown, the host device 1 or various control devices included in the controller may have a communication interface (wired or wireless) for communicating with an external network. When these devices are connected to an external network, they can communicate with other devices (typically servers, PCs, various sensor devices, etc.) on the external network.

In FIG. 1, the main welding robot MC1 is shown as a separate robot from the repair welding robot MC3. However, when main welding is performed using another system, or when main welding is performed manually and then the automatic repair welding system 1000 performs inspection and repair welding, the main welding robot MC1 may be omitted.

Further, the main welding robot MC1 may be integrated with each of the repair welding robot MC3 or the inspection robot MC2. For example, the repair welding robot MC3 may use the same robot to perform the main welding of the workpiece Wk and the repair welding of correcting a defective portion among the welded portions welded by the main welding. Further, for example, the inspection robot MC2 may use the same robot to perform the final welding of the workpiece Wk and the inspection of whether or not there is a defective portion among the welded portions welded by the final welding.

The inspection robot MC2 and the repair welding robot MC3 may be integrated into one robot, and the main welding robot MC1, the inspection robot MC2, and the repair welding robot MC3 may be integrated into one robot.

In the automatic repair welding system 1000 shown in FIG. 1, the respective numbers of main welding robots MC1, inspection robots MC2, and repair welding robots MC3 are not limited to the numbers shown in FIG. For example, the number of main welding robots MC1, inspection robots MC2, and repair welding robots MC3 may be plural or may not be the same. For example, the automatic repair welding system 1000 may include one main welding robot MC1, three inspection robots MC2, and two repair welding robots MC3. Thereby, the automatic repair welding system 1000 can be adaptively configured according to the processing range or processing speed of each robot as required.

The host device 1 is communicably connected among the monitor MN1, the interface UI1, the external storage ST, the robot controller 2a, and the robot controller 2b. Also, the host device 1 shown in FIG. 1 is connected to the inspection device 3 via the robot control device 2b, but may be directly connected to the inspection device 3 without the robot control device 2b.

The host device 1 may be a terminal device P1 integrally configured including the monitor MN1 and the interface UI1, or may be integrally configured including the external storage ST. In this case, the terminal device P1 is, for example, a PC (Personal Computer) used by a user (worker) to perform welding. The terminal device P1 is not limited to the PC described above, and may be a computer having a communication function such as a smart phone, a tablet terminal, or a PDA (Personal Digital Assistant).

The host device 1 is for performing final welding, inspection of welding points, and repair welding of defective points on the workpiece Wk based on input operations by a user (operator) or information preset by a user (operator). Generate each of the control signals. The host device 1 transmits to the robot control device 2a a control signal for performing main welding on the generated workpiece Wk and a control signal for performing repair welding of a defective portion. In addition, the host device 1 transmits a control signal for inspecting the welded portion welded by the final welding to the robot control device 2b.

The host device 1 may collect the inspection results of the welding points received from the inspection device 3 via the robot control device 2b. The host device 1 transmits the received test results to the external storage ST and the monitor MN1. Although the inspection device 3 shown in FIG. 1 is connected to the host device 1 via the robot control device 2b, it may be directly connected so as to be communicable.

The monitor MN1 may be configured using a display such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence). The monitor MN1 displays the weld inspection results and alerts received from the inspection device 3 . Further, the monitor MN1 may be configured using a speaker (not shown), for example, and may notify the alert by voice when receiving the alert. That is, the form of notification is not limited to notification by visual information.

The interface UI1 is a user interface (UI) that detects an input operation by a user (operator), and is configured using a mouse, keyboard, touch panel, or the like. The interface UI1 transmits an input operation based on a user's input operation to the host device 1 . The interface UI1 receives, for example, the input of weld lines, the setting of inspection criteria according to the weld lines, the operation of starting or ending the operation of the automatic repair welding system 1000, and the like.

The external storage ST is configured using, for example, a hard disk (HDD: Hard Disk Drive) or a solid state drive (SSD: Solid State Drive). The external storage ST may store the inspection results of the welding points received from the host device 1 .

The robot control device 2a is communicably connected to the host device 1, the main welding robot MC1, and the repair welding robot MC3. The robot control device 2a receives control information regarding final welding from the host device 1, controls the final welding robot MC1 based on the received control information, and causes the final welding to be performed on the workpiece Wk.

In addition, the robot control device 2a receives control information regarding repair welding received from the host device 1, controls the repair welding robot MC3 based on the received control information, and determines that the welding portion is defective by the inspection device 3. Repair welding is performed on the determined defective portion.

The robot control device 2a shown in FIG. 1 controls each of the main welding robot MC1 and the repair welding robot MC3. However, the automatic repair welding system 1000 according to Embodiment 1 is not limited to this, and for example, the main welding robot MC1 and the repair welding robot MC3 may be controlled using different controllers. Furthermore, the automatic repair welding system 1000 according to Embodiment 1 may control the main welding robot MC1, the inspection robot MC2, and the repair welding robot MC3 with one control device.

The robot control device 2b is communicably connected to the host device 1, the inspection device 3, and the inspection robot MC2. The robot control device 2b receives information about the welding location (for example, position information of the welding location) received from the host device 1 . Note that the welded portion includes a welded portion to the workpiece Wk (that is, a portion welded by final welding) and a welded portion corrected by repair welding. The robot control device 2b controls the inspection robot MC2 on the basis of the received information on the welding point to detect the shape of the weld bead at the welding point. Further, the robot control device 2b transmits the received information on the welded portion to the inspection device 3 that inspects the shape of the welded portion. The robot control device 2 b transmits the inspection result received from the inspection device 3 to the host device 1 .

The inspection device 3 is communicably connected to the robot control device 2b and the inspection robot MC2. The inspection device 3 inspects ( judge. The inspection device 3 transmits, as an inspection result, information on the defective portion determined to be defective among the welded portions (for example, it may include the defective section, the position information of the defective section, the cause of the defect, etc.) to the robot control device 2b. . In addition, when it is determined that the defective portion can be automatically repair-welded, the inspection device 3 also transmits information such as the type of repair and parameters for performing repair welding to the robot control device 2b as the inspection result. you can Note that the inspection device 3 may be directly connected to the host device 1 so as to be communicable. In this case, the inspection device 3 may be capable of transmitting the above information to the host device 1 without going through the robot control device 2b.

Although the robot control device 2b and the inspection device 3 are described as separate units in FIG. 1, the robot control device 2b and the inspection device 3 may be integrated into a single device.

The main welding robot MC1 is a robot that is communicably connected to the robot control device 2a and performs welding (main welding) on workpieces that have not been welded. The main welding robot MC1 performs the main welding on the workpiece Wk based on the control signal received from the robot controller 2a.

The inspection robot MC2 is communicably connected to the robot control device 2b and the inspection device 3. FIG. The inspection robot MC2 acquires the shape data of the weld bead at the welding location based on the control signal received from the robot controller 2b.

The repair welding robot MC3 is communicably connected to the robot controller 2a. The repair welding robot MC3 performs repair welding on the defective portion based on the inspection result of the welded portion received from the robot control device 2a (that is, information on the defective portion).

<Embodiment 1>
FIG. 2 is a diagram showing an internal configuration example of an automatic repair welding system 1000a regarding control of the inspection/repair welding robot MC23 according to the first embodiment. The inspection/repair welding robot MC23 shown in FIG. 2 is a robot in which the inspection robot MC2 and the repair welding robot MC3 shown in FIG. 1 are integrated. In order to make the explanation easier to understand, configurations related to the monitor MN1, the interface UI1, and the external storage ST are omitted.

(Configuration example of inspection/repair welding robot MC23)
Based on the control signal received from the robot controller 2, the inspection/repair welding robot MC23 inspects the welded portion of the workpiece Wk after the main welding has been performed. In addition, the inspection/repair welding robot MC23 automatically performs repair welding on defective welding points in the welding points of the workpiece Wk based on control signals received from the robot control device 2 .

In this example, the inspection/repair welding robot MC23 is a robot that performs arc welding. However, the inspection/repair welding robot MC23 may be a robot that performs, for example, laser welding other than arc welding. In this case, although illustration is omitted, instead of the welding torch 400, a laser head may be connected to the laser oscillator via an optical fiber.

In this example, an inspection/repair welding robot MC23 that performs arc welding includes a manipulator 200, a wire feeder 300, a welding wire 301, a welding torch 400, and a shape detector 500.

The manipulator 200 has a multi-joint arm, and this arm is movable based on control signals received from the robot controller 26 of the robot controller 2 . As a result, the positions of welding torch 400 and shape detection unit 500 can be controlled. The angle of the welding torch 400 with respect to the workpiece Wk can also be changed by moving the arm.

The wire feeding device 300 controls the feeding speed of the welding wire 301 based on control signals received from the robot control device 2 . Wire feeding device 300 may include a sensor capable of detecting the remaining amount of welding wire 301 .

Welding wire 301 is held by welding torch 400, and when power is supplied to welding torch 400 from welding power supply 4, an arc is generated between the tip of welding wire 301 and workpiece Wk, and the arc is generated. Welding is done. For convenience of explanation, illustration and explanation of the configuration for supplying the shielding gas to the welding torch 400 are omitted.

A shape detection unit 500 included in the inspection/repair welding robot MC23 detects the shape of the weld bead at the welding location based on the control signal received from the robot controller 2, and based on the detection result, generates shape data for each weld bead. to get The inspection/repair welding robot MC23 transmits the acquired shape data for each weld bead to the inspection device 3 .

The shape detection unit 500 is, for example, a three-dimensional shape measurement sensor, and has a laser light source (not shown) configured to scan the welding point on the workpiece Wk based on the position information of the welding point received from the robot control device 2. ), and a camera ( (not shown). The shape detection unit 500 transmits shape data (image data) of the welded portion based on the laser light imaged by the camera to the inspection device 3 .

The camera (not shown) described above includes at least a lens (not shown) and an image sensor (not shown). The image sensor is, for example, a CCD (Charged-Coupled Device) or CMOS (Complementary Metal Oxide Semiconductor) solid-state imaging device, and converts an optical image formed on an imaging surface into an electrical signal.

(upper device)
Next, the host device 1 will be described. A host device 1 generates a control signal for performing repair welding based on an input operation by a user (worker) or information preset by a user (worker), and controls the generated control signal by a robot. Send to device 2. The host device 1 includes a communication unit 10, a processor 11, and a memory 12. FIG.

The communication unit 10 is communicably connected to the robot control device 2 . The communication unit 10 transmits a control signal for executing repair welding to the robot control device 2 . The control signal for performing repair welding referred to here may include a control signal for controlling each of manipulator 200 , wire feeder 300 and welding power supply 4 .

The processor 11 is configured using, for example, a CPU (Central Processing Unit) or FPGA (Field Programmable Gate Array), and cooperates with the memory 12 to perform various types of processing and control. Specifically, the processor 11 refers to the programs and data held in the memory 12 and executes the programs, thereby implementing the functions of the cell control unit 13 .

The cell control unit 13 performs repair welding based on an input operation by a user (operator) using the interface UI 1 and information preset by the user (operator) and stored in the external storage ST. to generate a control signal for A control signal generated by the cell control unit 13 is transmitted to the robot control device 2 via the communication unit 10 .

The memory 12 includes, for example, a RAM (Random Access Memory) as a work memory used when executing each process of the processor 11, and a ROM (Read Only Memory) for storing programs and data defining the operation of the processor 11. have. The RAM temporarily stores data or information generated or acquired by the processor 11 . A program that defines the operation of the processor 11 is written in the ROM.

In addition, the memory 12 stores information types related to the work Wk, a work S/N (Serial Number) assigned in advance to each work Wk, a welding line ID assigned to each welding point (welding line) set by the user, and the like. Remember.

(Robot control device 2)
Next, the robot control device 2 will be explained. Robot controller 2 controls manipulator 200 , wire feeder 300 , and welding power supply 4 based on control signals received from host device 1 . The robot control device 2 includes a communication section 20 , a processor 21 and a memory 22 . The processor 21 includes a program editing section 23a, a program calling section 23b, a program storage section 23c, a computing section 24, an inspection device control section 25, a robot control section 26, and a welding power source control section 27. Configured.

The communication unit 20 is communicably connected to the host device 1 . The communication unit 20 receives, from the host device 1 , a control signal for performing repair welding or visual inspection by the inspection device 3 .

The processor 21 is configured using, for example, a CPU or FPGA, and cooperates with the memory 22 to perform various types of processing and control. Specifically, the processor 21 refers to the programs and data held in the memory 22 and executes the programs to realize the functions of the respective units. Each unit is a program editing unit 23a, a program calling unit 23b, a program storage unit 23c, a computing unit 24, an inspection device control unit 25, a robot control unit 26, and a welding power source control unit 27. Functions of each part are, for example, a function of editing and calling a pre-stored repair welding program for executing repair welding, and manipulator 200, wire feeder 300 and welding power supply based on the called repair welding program. 4, and the like is a function of generating a control signal for controlling each of them.

The memory 22 has, for example, a RAM as a work memory used when executing each process of the processor 21 and a ROM for storing programs and data defining the operation of the processor 21 . Data or information generated or obtained by the processor 21 is temporarily stored in the RAM. A program that defines the operation of the processor 21 is written in the ROM.

The program editing unit 23a generates a program (control signal ). The program editing unit 23a refers to the repair welding basic program for executing repair welding, which is stored in advance in the program storage unit 23c. parameters), etc., to edit the repair welding program. The edited repair welding program may be stored in the program storage unit 23c, or may be stored in the RAM in the memory 22 or the like.

It should be noted that the repair welding program referred to here includes, in performing repair welding, current, Parameters such as voltage, offset, velocity, attitude, method, etc. may be included.

The program calling unit 23b calls various programs stored in the ROM included in the memory 22, the program storage unit 23c, and the like. The program calling unit 23b may call a program on the side of the inspection/repair welding robot MC23. Further, the program calling unit 23b can select and call an appropriate program from among a plurality of programs according to the inspection result (judgment result) of the inspection device 3. FIG. That is, the program calling unit 23b can change the program according to the inspection result (judgment result) by the inspection device 3. FIG.

The program storage unit 23c stores various programs that the robot control device 2 uses. For example, the repair welding basic program described above, the repair welding program edited by the program editing unit 23a, and the like may be stored in the program storage unit 23c.

The calculation unit 24 is a functional block that performs various calculations. For example, calculations for controlling the manipulator 200 and the wire feeding device 300 controlled by the robot control unit 26 are performed based on the repair welding program. Alternatively, the calculation unit 24 may calculate an offset amount required for repair welding to the defective portion based on the position of the defective portion.

The inspection device control section 25 generates a control signal for controlling the inspection device 3 . This control signal is transmitted to the inspection device 3 via the communication unit 20 . Conversely, the inspection device control unit 25 receives various information from the inspection device 3 via the communication unit 20, and edits, for example, a repair welding program based on the information (program editing unit 23a). Send to , and perform various other processing.

The robot control unit 26 controls the manipulator 200 and the wire feeding device 300 based on the repair welding program called by the program calling unit 23b or stored in the program storage unit 23c and the calculation result from the calculation unit 24. drive the Welding power supply control unit 27 drives welding power supply 4 based on the repair welding program called by program calling unit 23b or stored in program storage unit 23c and the calculation results from calculation unit 24.

In the case where the inspection robot MC2 and the repair welding robot MC3 are separate units, the information about the defect location is sent from the inspection device 3 connected to the inspection robot MC2 via the host device 1 to the repair welding robot. It may be sent to the robot controller 2 connected to MC3. The program editing unit 23a of the robot control device 2 connected to the repair welding robot MC3 is based on the information (for example, the result of determination by the inspection device 3 described later) regarding the defective portion received from the host device 1 via the communication unit 20. to edit a program (control signal) for performing repair welding.

Further, in the above configuration example, the mode in which the program editing unit 23a and the program calling unit 23b are provided on the robot control device 2 side has been described. However, the program editing section and the program calling section may be provided on the inspection device 3 side. In this case, the inspection device 3 may call the above program and edit the repair welding program. The calling source of the program is not limited to the inspection device 3, and the program may be called from the robot control device 2, the inspection/repair welding robot MC23 connected to the robot control device 2, or the like. The called program is edited by the program editor, and the edited program is transmitted from the inspection device 3 to the robot control device 2 as a repair welding program. Welding can be performed.

(Inspection device 3)
Next, the inspection device 3 will be described. The inspection device 3 inspects (determines) the welded portion of the work Wk based on the shape data of the weld bead for each welded portion acquired by the shape detection section 500 . This determination may include determination of whether the welding result of the welded portion is good or bad, determination of whether the defective welding portion can be corrected, determination of whether automatic repair welding can be performed, and the like. These various determinations will be described later.

The inspection device 3 includes a communication section 30 , a processor 31 , a memory 32 , a shape detection control section 34 , a data processing section 35 , a determination threshold storage section 36 and a determination section 37 .

The communication unit 30 is communicably connected to the robot control device 2 . Note that the communication unit 30 may be directly connected to the host device 1 so as to be communicable. The communication unit 30 receives information about welding points from the host device 1 or the robot control device 2 . The information about the weld location may include, for example, the work type, work S/N, welding line ID, and the like.

The inspection device 3 also transmits the inspection result of the welded portion to the host device 1 or the robot control device 2 via the communication unit 30 .

The processor 31 is configured using a CPU or FPGA, for example, and cooperates with the memory 32 to perform various types of processing and control. Specifically, the processor 31 refers to the programs and data held in the memory 32 and executes the programs to realize the functions of the respective units. Each section includes a shape detection control section 34 , a data processing section 35 , a determination threshold storage section 36 and a determination section 37 . The function of each part is, for example, the function of controlling the shape detection unit 500 based on the control signal related to the inspection corresponding to the welding point received from the robot control device 2, the shape data of the weld bead received from the shape detection unit 500, and the a function of generating image data based on the image data; and a function of inspecting the weld location based on the generated image data.

Note that when performing machine learning, which will be described later, the processor 31 may be configured to include, for example, a plurality of GPUs for calculation. In this case, the processor 31 may be used together with the above-described CPU or the like.

The memory 32 has, for example, a RAM as a work memory used when executing each process of the processor 31 and a ROM for storing programs and data defining the operation of the processor 31 . Data or information generated or obtained by the processor 31 is temporarily stored in the RAM. A program that defines the operation of the processor 31 is written in the ROM. The memory 32 may also include, for example, a hard disk (HDD: Hard Disk Drive), a solid state drive (SSD: Solid State Drive), and the like.

The shape detection control unit 34 controls the shape detection unit based on the shape data of the weld bead at the welding location received from the shape detection unit 500 and the control signal related to the inspection corresponding to the welding location received from the robot control device 2. 500 is controlled. When the shape detection unit 500 is positioned at a position where the shape detection unit 500 can capture an image of the welded portion (shape detection is possible), the shape detection control unit 34 irradiates a laser beam to acquire the shape data of the weld bead at the welded portion. Upon receiving the shape data acquired by the shape detection unit 500 , the shape detection control unit 34 outputs this shape data to the data processing unit 35 .

The data processing unit 35 converts the shape data of the weld bead at the welding location input from the shape detection control unit 34 into image data. The shape data is, for example, point cloud data of a shape line consisting of a reflection trajectory of a laser beam irradiated on the surface of the weld bead. The data processing unit 35 performs statistical processing on the input shape data, and generates image data regarding the shape of the weld bead at the weld location. In order to emphasize the position and shape of the weld bead, the data processing unit 35 may perform edge enhancement correction that emphasizes the peripheral portion of the weld bead.

The determination threshold value storage unit 36 stores each threshold value set according to the welding point in order to execute the determination described later according to the welding point. Each threshold is, for example, an allowable range (threshold) regarding the displacement of the welded portion, a threshold regarding the height of the weld bead, a threshold regarding the width of the weld bead, and the like. In addition, the determination threshold storage unit 36 stores, as each threshold after repair welding, an allowable range (for example, a minimum allowable value, a maximum allowable value, etc. regarding the height of the weld bead) that satisfies the quality requested by the customer. .

Further, the determination threshold value storage unit 36 may store the upper limit of the number of inspections for each welding location. As a result, the inspection device 3 determines that it is difficult or impossible to repair the defective portion by automatic repair welding for the defective portion exceeding a predetermined number of times when the defective portion is repaired by the automatic repair welding system 1000a. A decrease in operating rate can be suppressed.

The determination unit 37 refers to the threshold value stored in the determination threshold storage unit 36 to determine the weld location based on the shape data of the weld bead at the weld location. The details of this determination will be described later with reference to FIG. 3 and subsequent figures.

The determination unit 37 measures the position of the defective portion (for example, the start and end positions of the defective portion, the position of the hole in the weld bead, the position of the undercut, etc.), analyzes the content of the defect, and determines the cause of the defect. to estimate The determination unit 37 generates an inspection result (determination result) for the welded portion from the measured position of the defective portion and the estimated defect factor, and transmits the generated inspection result to the host device 1 via the robot control device 2. Send to

When determining that there is no defective portion, the determination unit 37 generates an alert notifying that there is no defective portion, and transmits the generated alert to the host device 1 via the robot control device 2. do. The alert sent to the host device 1 is sent to and displayed on the monitor MN1.

In addition, the data processing unit 35 counts the number of inspections for each welding location, and if the welding inspection result does not improve even when the number of inspections exceeds the number of times stored in the determination threshold storage unit 36, the defective location is automatically repaired by welding. is difficult or impossible to correct. In this case, the determination unit 37 generates an alert including the position of the defective portion and the cause of the defect, and transmits the generated alert to the host device 1 via the robot control device 2 . The alert sent to the host device 1 is sent to and displayed on the monitor MN1.

Note that the inspection device 3 may generate an alert with content other than the above. This alert is also sent to the host device 1 via the robot control device 2 . The alert sent to the host device 1 is sent to and displayed on the monitor MN1.

FIG. 3 is a flow chart showing a first operation procedure example of automatic repair welding by the automatic repair welding system 1000a according to the first embodiment. This flowchart is based on the system configuration shown in FIG. 2, and shows an example in which inspection and repair welding are performed on the workpiece Wk after final welding has been performed.

First, a visual inspection is performed on the workpiece Wk for which final welding has been performed (step St101). This appearance inspection is performed using the shape data obtained by the shape detection unit 500 . Here, one work Wk may have a plurality of welding points (welding lines). Therefore, in this example, a visual inspection is performed for each welding location (welding line) (loop from steps St101 to St103).

Here, an example of the state of data inside the processing will be given. The weld bead shape data detected by the shape detector 500 may include a plurality of weld points (weld lines). The data processing unit 35 detects each welding point (welding line) from this shape data. This detection algorithm may use conventional techniques. In addition, the inspection device 3 may acquire the work type, work serial number (work S/N), welding line ID, etc. of the work Wk in which the weld bead exists from the host device 1 or the robot control device 2. . The inspection device 3 can manage the aforementioned shape data by associating the workpiece type, workpiece S/N, welding line ID, and the like. The inspection device 3 may store these data in the memory 32 or the like.

The data processing unit 35 of the inspection device 3 determines the type of welding failure based on the shape data of the weld bead obtained as a result of the visual inspection for the welding point (weld line) currently being processed in the loop (steps St101 to St103). Alternatively, characteristic data, which will be described later, is generated (step St102).

Reference is now made to FIG. FIG. 10 is a conceptual diagram showing a comparison between master data and weld bead shape data to be inspected. The left side of FIG. 10 shows master data stored in the memory 32 or the like. The master data may be image data obtained by detecting (a welding portion of) a workpiece (a welded portion of the workpiece) when the welding is performed satisfactorily, and converting the image into an image. The right side of FIG. 10 shows shape data of the weld bead in the workpiece Wk to be inspected.

For example, when arc welding is performed on the welded portion of the work Wk, various shape defects may occur at the welded portion. For example, holes or undercuts may occur due to burn-down of a portion of the weld. In addition, the length of the weld bead along the weld line, the width of the weld bead in the direction perpendicular to the weld line, the height of the weld bead, etc. may deviate beyond the allowable range from the reference values. These "holes", "undercuts", "bead length", "bead width", "bead height", etc. correspond to the welding failure types (step St102). However, failure types other than those described above may be used.

The data processing unit 35 may calculate the characteristic data for each of these defect types based on the master data and shape data of the weld bead in the workpiece Wk to be inspected.

The characteristic data is data for identifying good or bad welding for each of the types of failure described above. For example, FIG. 10 illustrates a broken bead 601 and a hole 602 . A broken bead 601 has a shorter weld bead length along the weld line than the master data. That is, the start position and end position of the weld bead deviate from the start position and end position of the weld bead in the master data. The data processing unit 35 calculates the amount of this deviation as characteristic data. Note that the data processing unit 35 may calculate the length of the weld bead itself as the characteristic data.

Similarly, a pierce 602 is a hole in the weld bead. The data processing unit 35 calculates, for example, the diameter of this hole as characteristic data.

In addition, the data processing unit 35 can calculate the bead width, the size of the undercut, etc. as characteristic data. The types of characteristic data are not limited to those described above.

Then, the determination unit 37 compares the characteristic data calculated as described above with the threshold value stored in the determination threshold storage unit 36 . By comparing the characteristic data with the threshold value, determinations (steps St104, St105, and St107) to be described later can be performed. Note that the same threshold value or different threshold values may be used between determinations in different determination steps.

The data processing unit 35 may also calculate the inspection score by comprehensively using the plurality of characteristic data. For example, let ΔL be the difference between the length of the weld bead in the shape data of the workpiece Wk to be inspected and the length of the weld bead in the master data. Let ΔW be the difference between the width of the weld bead in the shape data of the workpiece Wk to be inspected and the length of the weld bead in the master data. Let ΔH be the difference between the height of the weld bead in the shape data of the workpiece Wk to be inspected and the height of the weld bead in the master data. If a hole is detected in the shape data of the workpiece Wk to be inspected, the diameter of the hole is set to r. At this time, for example, the inspection score S may be calculated as follows.

Inspection score S=w1×ΔL+w2×ΔW+w3×ΔW+w4×r

In the above formula, w1 to w4 are coefficients indicating weighting (importance of the characteristic data).

The data processing unit 35 may calculate, for example, the inspection score as described above as the characteristic data (step St102). It should be noted that the above formula is an example, and the data processing unit 35 may calculate the test score using a formula other than the above. Also, the test score need not be a single value. For example, an inspection score for weld bead dimensions (length, width, height, etc.) and an inspection score for perforations, undercuts, etc. are separately calculated, and these are combined and used as a group of inspection scores. may

The data processing unit 35 may calculate characteristic data other than the above. For example, the shape data of a plurality of works already inspected by the inspection device 3 and the characteristic data calculated from the shape data are accumulated in the memory 32 or the like, and the standard deviation value and the variance value for this accumulated data are stored. etc. may be calculated by the data processing unit 35 as characteristic data.

The inspection device 3 can further associate the information indicating the defect type and the characteristic data generated in step St102 with the aforementioned shape data and manage them. That is, at the time of step St102, the inspection apparatus 3 can manage the shape data of the workpiece Wk to be inspected by linking the workpiece type, workpiece S/N, welding line ID, defect type, and characteristic data. . The inspection device 3 may store these data in the memory 32 or the like.

Next, the inspection device 3 determines whether or not the visual inspection of all welding points (all welding lines) has been completed for the work Wk currently being inspected (step St103). If the appearance inspection for all welding points (all welding lines) has been completed (step St103, Y), the process proceeds to step St104. If the visual inspection of all welding points (all welding lines) has not been completed (step ST103, N), the process returns to step St101 to perform the visual inspection of the next welding point (welding line) on the workpiece Wk. I do.

Next, step St104 will be described. The judging unit 37 of the inspection device 3 judges whether the welding inspection result is good or bad for the work Wk for which the visual inspection of all welding points (all welding lines) has been completed. This welding result quality determination may be a determination based on the aforementioned characteristic data and a threshold value, or may be a determination based on an inspection score, and machine learning is performed as will be described later based on FIGS. The judgment may be made by a judgment model (AI learning model).

If the welding inspection result is good (step St104, Y), repair welding for this work Wk is not necessary in the first place, so the processing for this work Wk ends. At this time, an alert to the effect that the welding inspection result for the workpiece Wk is good may be issued as described above.

On the other hand, if the welding inspection result for the work Wk is unsatisfactory (step St104, N), step St105 is executed to perform repair welding if the defective portion can be corrected.

The determination unit 37 of the inspection device 3 determines whether or not the defective welding portion included in the welding portion (weld line) of the work Wk can be corrected by repair welding (step St105). This correction propriety determination may be a determination based on the characteristic data and the threshold value described above, may be a determination based on the inspection score, or may be a determination by a machine-learned determination model (AI learning model). good too. Also, the criterion in step St105 may be different from the criterion in step St104. For example, a value different from the threshold value used in step St104 may be used as the threshold value in step St105.

If it is determined that the defective portion can be corrected (step St105, Y), step St106, which will be described later, is executed. If it is determined that the defective portion cannot be corrected (step St105, N), the defective portion cannot be corrected even if repair welding is attempted, so the processing for this workpiece Wk is terminated. At this time, an alert to the effect that the welding inspection result for the workpiece Wk is unsatisfactory and cannot be corrected may be issued as described above.

The inspection device 3 can manage the aforementioned shape data in association with the information indicating whether or not the correction is possible, which is generated in step St105. That is, at the time of step St105, the inspection device 3 associates the shape data of the workpiece Wk to be inspected with the workpiece type, workpiece S/N, welding line ID, defect type, characteristic data, and data indicating whether or not correction is possible. can be managed The inspection device 3 may store these data in the memory 32 or the like.

Next, the data processing unit 35 generates correction types and parameters for correction (step St106). The type of correction may be an appropriate correction method (the same welding method as the main welding, or various welding methods different from the main welding) for the defective portion of the welding. Parameters for correction (correction parameters) include the current and voltage of the welding power supply 4, the position information of the defective portion, the offset amount of the repair welding start position or the repair end position, the welding speed, the attitude of the welding torch 400, weaving. It may be the presence or absence of, etc. However, it is not limited to these.

Generation of correction types and correction parameters by the data processing unit 35 can be realized by, for example, storing table data for program generation (program generation logic table) in the memory 32 or the like. Reference is now made to FIG. FIG. 11 is a diagram showing a program generation logic table, showing (a) an example of using defect type and characteristic data as keys, and (b) an example of using defect type and inspection score as keys.

As shown in FIG. 11A, in the program generation logic table, correction types and correction parameters that can be used at that time are registered according to the above-described defect types and characteristic data value ranges. It's okay. The data processing unit 35 extracts a set of correction type and correction parameter from the program generation logic table based on the defect type and characteristic data generated in step St102. For example, if the defect type is "hole" and the value of the characteristic data (diameter, etc.) is 2 to 4, the correction type is "repair welding twice" and the correction parameter is "the first repair welding is Data set data6 is used, and data set data7 is used for the second repair welding." Although it is only an example, A, V, and S included in the data set may be "current," "voltage," and "welding speed," respectively.

Further, as shown in FIG. 11(b), in the program generation logic table, correction types and correction parameters may be registered according to the above-described defect types and inspection scores. For example, if the defect type is "perforated" and the inspection score value is 83 to 94, the correction type is "perform repair welding twice" and the correction parameter is "the first repair welding data set data6. data set data7 is used for the second repair welding", and a set of correction type and correction parameter can be extracted.

The program generation logic table described above may hold the data itself. Alternatively, the data may be stored in another storage area or the like, and the program generation logic table may hold reference information (link information) to the data.

It should be noted that the type of modification and the algorithm for generating parameters for modification by the data processing unit 35 are not limited to the above program generation logic table. For example, the welding method (correction type) is the same as the welding method in the main welding, and the correction parameters are the parameters used at the time of the main welding, and the coefficient based on the inspection score value is added. Multiplied values may also be used. To give a more specific example, when the inspection score is below a certain level, a current 0.6 times (coefficient 0.6) the current value in the main welding is used without changing the welding method. For example, perform repair welding.

Next, the determination unit 37 of the inspection device 3 determines whether automatic welding is possible for the defective welding portion included in the welding portion (weld line) of the work Wk (step St107). This determination may be made using the program generation logic table described above. That is, if even one set of correction type and correction parameter can be extracted from the program generation logic table, it can be determined that automatic correction is possible (step St107, Y). Conversely, if even one set of correction type and correction parameter cannot be extracted from the program generation logic table, it can be determined that automatic correction is impossible (step St107, N).

The determination unit 37 of the inspection device 3 can also determine whether or not automatic welding can be performed by another method. For example, the determination unit 37 may make a determination by comparing the characteristic data generated in step St102 with a threshold value for determining whether automatic welding is possible, which is stored in the determination threshold storage unit 36 . Further, even if the determination unit 37 makes a determination by comparing the inspection score (or group of inspection scores) generated in step St102 with the threshold value for determining whether or not automatic welding is possible, which is stored in the determination threshold storage unit 36. good. Furthermore, as will be described later with reference to FIGS. 8 and 9, determination may be made using a determination model (AI learning model) subjected to machine learning.

When it is determined that automatic welding is possible for the defective welding portion included in the welding portion (weld line) of the work Wk (step St107, Y), the conditions for repair welding are determined based on this determination result. After setting (step St109), repair welding is executed (step St110). For example, using the type of repair and the parameters for performing the repair generated in step St106 as setting conditions (step St109), the program editing unit in the robot control device 2 or the inspection device 3 executes the repair program. To edit. Alternatively, the program calling unit in the robot control device 2 or the inspection device 3 invokes the repair program using the type of repair and the parameters for performing the repair generated in step St106 as setting conditions (step St109). change. Using this edited or changed repair program, the robot controller 2 controls the inspection/repair welding robot MC23 to perform repair welding (step St110).

If it is determined that automatic welding is impossible for the defective welding portion included in the welding portion (weld line) of the work Wk (step St107, N), the defective portion itself can be corrected (step St105, Y) Therefore, manual welding by the operator (step St108) is performed instead of automatic welding. Therefore, the inspection device 3 alerts the operator that welding is possible, as described above.

Note that the inspection device 3 can manage the information indicating whether or not automatic welding is possible, which is generated in step St107, by linking it to the aforementioned shape data. That is, at the time of step St107, the inspection device 3 adds the work type, work S/N, welding line ID, defect type, characteristic data, data indicating whether correction is possible, and automatic Data indicating whether correction is possible can be linked and managed. The inspection device 3 may store these data in the memory 32 or the like.

In addition, whether the repair welding is performed automatically (step St110) or manually performed by the operator (step St108), the process may shift to step St101 again after the repair welding. This is because the weld bead after repair welding is inspected again to see if it is good. The inspection device 3 associates the characteristic data and the inspection score (step St102) calculated in this reinspection with the shape data as data items different from the characteristic data and the inspection score in the previous inspection. you can

When a worker performs repair welding (step St108), subsequent handling of the work Wk can be entrusted to the worker. In this case, the process does not return to step St101, and the process for this workpiece Wk can be terminated (step St108, dashed line).

In addition, the type of correction generated in step St106, the parameters for performing the correction, and the above-described various data linked to the shape data can be useful reference information for a welder who manually performs repair welding. Therefore, when repair welding is performed by the operator instead of automatic repair welding (step St108), the inspection device 3 may alert the operator of the various data described above. This alert may be given by displaying data on the monitor MN1 under the control of the host device 1, as described above.

As described above, the appearance inspection and repair welding of the work Wk can be performed. It should be noted that an alert may be sent to the higher-level device 1 or the like regarding the results of the welding inspection result pass/fail determination in step St104, the correction feasibility determination in step St105, and the automatic welding feasibility determination in step St107. The host device 1 may display this judgment result on the monitor MN1.

FIG. 4 is a flowchart showing a second operation procedure example of automatic repair welding by the automatic repair welding system 1000a according to the first embodiment. This flow chart is based on the system configuration shown in FIG. 2, and shows an example of performing inspection and repair welding on a workpiece Wk that has undergone final welding.

The operating procedure in the flowchart of FIG. 4 is basically the same as the operating procedure in the flowchart of FIG. Specifically, steps St101, St102, St103, St104, St105, St106, St107, St108, St109, and St110 in FIG. , St209, and St210, respectively. Therefore, the description of the same parts as the operation procedure shown in the flowchart of FIG. 3 will be omitted, and the different contents will be described below.

In the flowchart of FIG. 4, the step St204 for determining whether or not the inspection of all points has been completed comes after the step St203 for determining the quality of the welding inspection result. That is, without waiting for the appearance inspection of all welding points (all welding lines) of the workpiece Wk currently being processed to be completed, every time a welding point with a poor welding inspection result is found, The processing order is such that repair welding (steps St208 and St210) can be performed.

In the case of the processing procedure shown in the flowchart of FIG. 3, repair welding can be collectively performed for a plurality of defective welding points after the visual inspection is completed for all the welding points included in one work Wk. That is, since the number of times of going back and forth between visual inspection and repair welding is reduced, there is an advantage that complexity is eliminated. In particular, when the inspection robot MC2 and the repair welding robot MC3 shown in FIG. 1 are separate robots, the frequency of movement of the work Wk between these robots is reduced. is preferred.

On the other hand, in the case of the processing procedure shown in the flowchart of FIG. 4, if it is determined in step St205 that correction is not possible, the processing itself for the workpiece Wk ends (step St205, N). In other words, when an uncorrectable weld defect is first found, the subsequent processing can be skipped and the inspection process can be terminated early. Therefore, the processing procedure shown in the flowchart of FIG. 4 has the advantage of eliminating unnecessary processing and reducing the processing time.

Although not shown, step St204 may be executed when it is determined in step St205 that correction is not possible. In the case of this partially corrected processing procedure, the loop processing from step St201 to step St204 will continue even after the uncorrectable welding point is found, so the advantage of omitting the above processing cannot be enjoyed. . However, the partially modified procedure has the advantage that more visual inspection results can be obtained for the work Wk.

<Embodiment 2>
FIG. 5 is a diagram showing an internal configuration example of an automatic repair welding system 1000b regarding control of the inspection robot MC2 and the repair welding robot MC3 according to the second embodiment. Unlike Embodiment 1, the inspection robot MC2 and the repair welding robot MC3 are separate bodies. In order to make the explanation easier to understand, configurations related to the monitor MN1, the interface UI1, and the external storage ST are omitted.

In FIG. 5, an inspection robot MC2 and a repair welding robot MC3 are arranged to face each other with the same workpiece Wk sandwiched therebetween. However, the arrangement position is not limited to this. For example, inspection robot MC2 and repair welding robot MC3 may be located in different factories or cells.

As illustrated, a robot controller 2a for controlling the repair welding robot MC3 is connected to the repair welding robot MC3. A robot controller 2b for controlling the inspection robot MC2 and an inspection device 3 are connected to the inspection robot MC2. The robot control device 2a and the robot control device 2b are connected via the host device 1 so as to be communicable.

Further, the inspection device control section 25 is omitted from the robot control device 2a. This is because the inspection device 3 is not connected to the repair welding robot MC3 side, and the shape detection unit 500 is not provided on the repair welding robot MC3 side. Further, the welding power source control section 27 is omitted from the robot control device 2b. This is because the inspection robot MC2 does not have welding torch 400, wire feeder 300, welding wire 301, etc., which are components related to welding.

As shown in FIG. 5, when the inspection robot MC2 and the repair welding robot MC3 are separated, inspection of the workpiece Wk (inspection robot MC2) and repair welding according to the inspection results (repair welding robot MC3) can be performed independently.

FIG. 6 is a flow chart showing an example of an operation procedure of automatic repair welding by the automatic repair welding system 1000b according to the second embodiment. This flowchart is based on the system configuration shown in FIG. 5, and shows an example in which inspection and repair welding are performed on the workpiece Wk after the main welding has been performed. The operating procedure in the flowchart shown in FIG. 6 is basically the same as the operating procedure in the flowchart shown in FIG. Specifically, steps St101 to St107 in FIG. 3 correspond to steps St301 to St307 in FIG. 6, respectively. Therefore, the description of the parts similar to the operation procedure shown in FIG. 6 will be omitted, and the different contents will be described below.

After step St307 in FIG. 6, notification/operator welding (step St108), condition setting (step St109), and repair welding (St110) do not follow. This is because the inspection robot MC2 and the repair welding robot MC3 are separated from each other as shown in FIG. That is, in the processing procedure shown in FIG. 6, after the inspection device 3 connected to the inspection robot MC2 performs the automatic welding propriety determination (step St307), repair welding of a separate body, not the inspection robot MC2, is performed. Robot MC3 performs repair welding as needed.

Therefore, in step St308, the inspection device 3 issues a notification (alert). This notification may include information indicating the determination results of the determinations in steps St304, St305, and St307, and when repair welding is performed, information indicating conditions for repair welding (step St306, correction type and parameters) may be included.

Upon receiving the above notification from the inspection device 3, the host device 1 may transmit a control signal for performing repair welding to the robot control device 2a based on the above information included in the notification. Upon receiving this control signal, the robot control device 2a drives the repair welding robot MC3 under the control of the robot control unit 26 to perform repair welding on the workpiece Wk.

Editing of the program for repair welding may be performed by the program editing unit 23a provided in the robot control device 2a, or may be performed by the program editing unit 23a provided in the robot control device 2b. Although not shown, the inspection apparatus 3 may include a program calling section and a program editing section, and the program editing section provided in the inspection apparatus 3 may edit the repair program. The robot control device 2a receives the repair welding program edited by the robot control device 2b or the inspection device 3 via the communication unit 10 of the host device 1, and performs repair based on the edited repair welding program. The welding robot MC3 may be driven to perform repair welding on the workpiece Wk.

After the repair welding by the robot control device 2a is completed, under the control of the host device 1, a welding inspection (reinspection) may be performed again with the inspection device 3 as the processing subject. This re-inspection may be performed according to the procedure shown in FIG. 6 above.

<Modification 1>
Next, Modification 1 to Embodiments 1 and 2 will be described. In Embodiments 1 and 2, the welding inspection result pass/fail determination (steps St104, St204, St304), correction feasibility determination (St105, St205, St305), and automatic welding feasibility determination (St107, St207, St307) are performed. were judged separately. On the other hand, in Modification 1, these three types of determination are performed in one determination. In addition, although this modified example 1 will be described based on the system configuration shown in FIG. 2, other system configurations may be used.

FIG. 7 is a flowchart showing an example of an operation procedure of automatic repair welding by the automatic repair welding system 1000a according to Modification 1. As shown in FIG. Note that Steps St401 and St403 are the same as Steps St101 and St103 in FIG. 3, so description thereof will be omitted.

Step St402 will be described. For example, using the above inspection score as characteristic data, if a plurality of thresholds are set for this inspection score, "welding inspection result is good", "correctable by automatic repair welding", "correctable by repair welding by operator", " The determination unit 37 can determine the four states of "cannot be corrected" at once.

In order to make this determination, the data processing unit 35 generates inspection scores in step St402. The generated inspection score is denoted as inspection score S here. At least the first threshold T1, the second threshold T2, and the third threshold T3 are stored in the determination threshold storage unit 36. FIG. In this example, it is assumed that the larger the value of the inspection score, the better the quality of the weld bead, and the first threshold T1<the second threshold T2<the third threshold T3.

After the inspection of all locations is completed (step St403, Y), the determination unit 37 determines whether the inspection score S is less than the first threshold value T1, is greater than or equal to the first threshold value T1 and less than the second threshold value T2, determines whether the second is greater than or equal to the threshold T2 and less than the third threshold T3, or is greater than or equal to the third threshold T3 (step St404). If the inspection score S is less than the first threshold value T1, it is treated as "impossible to correct" (step St405). If the inspection score S is greater than or equal to the first threshold value T1 and less than the second threshold value T2, it is treated as "correctable by repair welding by the operator" (step St406). If the inspection score S is equal to or greater than the second threshold T2 and less than the third threshold T3, it is treated as "correctable by automatic repair welding" (step St407). If the inspection score S is equal to or greater than the third threshold value T3, it is treated as "good welding inspection result" (step St410).

Note that the determination unit 37 determines whether the test score S is equal to or less than the first threshold, exceeds the first threshold and is equal to or less than the second threshold, exceeds the second threshold and is equal to or less than the third threshold, It may be determined whether the threshold value of 3 is exceeded (step St404). In this case, if the inspection score S is equal to or less than the first threshold, it is treated as "impossible to correct" (step St405). If the inspection score S exceeds the first threshold and is equal to or less than the second threshold, it is treated as "correctable by repair welding by the operator" (step St406). If the inspection score S exceeds the second threshold and is equal to or less than the third threshold, it is treated as "correctable by automatic repair welding" (step St407). If the inspection score S exceeds the third threshold, it is treated as "good welding inspection result" (step St410).

In addition, after calculating the inspection score such that the smaller the inspection score value, the better the quality of the weld bead, by comparing this inspection score with the threshold value, "welding inspection result is good" and "correction by automatic repair welding" is performed. The determination unit 37 may simultaneously determine the four cases of "possible", "correctable by repair welding by the operator", and "impossible to correct".

Then, when the determination result is "correctable by automatic repair welding", repair welding can be performed. The processing contents of steps St407, St408, and St409, which are controls for repair welding at this time, may be the same as the processing contents of steps St106, St109, and St110 in FIG. 3, respectively. In steps St407 and St408, the data processing unit 35 may acquire (extract) the correction type and correction parameters by referring to the above-described program generation logic table.

Even if characteristic data other than the inspection score is generated in step St402, the determination unit 37 compares the characteristic data with a plurality of threshold values to determine whether "welding inspection result is good" or "corrected by automatic repair welding". The four cases of "possible", "correctable by repair welding by the operator", and "impossible to correct" may be determined at once.

<Modification 2>
Next, Modification 2 of Embodiments 1 and 2 will be described. In Embodiments 1 and 2, the welding inspection result pass/fail determination (steps St104, St204, St304), correction feasibility determination (St105, St205, St305), and automatic welding feasibility determination (St107, St207, St307) are performed. , characteristic data and thresholds, inspection scores, program-generated tables, etc. On the other hand, in Modification 2, these determinations are performed using a determination model that has been learned by machine learning. Further, in Modification 2, by receiving feedback from the host device 1 and the robot control device 2, the accuracy of the judgment model is improved.

FIG. 8 is a diagram showing an internal configuration example of an automatic repair welding system 1000c regarding control of the inspection/repair welding robot MC23 according to Modification 2. As shown in FIG. FIG. 9 is a diagram showing a configuration example of the data processing unit 35 in the internal configuration example of FIG. Modification 2 will be described below with reference to FIGS. 8 and 9. FIG. The inspection/repair welding robot MC23 shown in FIG. 8 is a robot in which the inspection robot MC2 and the repair welding robot MC3 shown in FIG. 1 are integrated. In order to make the explanation easier to understand, configurations related to the monitor MN1, the interface UI1, and the external storage ST are omitted.

An automatic repair welding system 1000c shown in FIG. 8 has substantially the same configuration as the automatic repair welding system 1000a shown in FIG. Therefore, the description of the parts similar to those of the automatic repair welding system 1000a shown in FIG. 2 will be omitted, and the different contents will be described below.

In the automatic repair welding system 1000c shown in FIG. 8, the judgment threshold storage section 36 and the judgment section 37 are not provided. Note that the determination process is performed by the data processing unit 35 instead of the determination unit 37 . Also, the automatic repair welding system 1000c shown in FIG. This feedback receiver 38 will be described later.

In addition to the data processing unit 35, FIG. 9 shows the feedback receiving unit 38 and the shape detecting unit 500 in the inspection/repair welding robot MC23.

The data processing section 35 includes two decision blocks. The first determination block is composed of a first storage unit 352a, a first learning data generation unit 353a, a first determination model generation unit 354a, and a first determination unit 355a. The second judgment block is composed of a second storage section 352b, a second learning data generation section 353b, a second judgment model generation section 354b, and a second judgment section 355b.

The first determination block performs determinations corresponding to the acceptance/rejection determination of the welding inspection result (step St104) and the determination of whether correction is possible (step St105) in the first embodiment shown in FIG. That is, the first determination block can output three types of determination results, ie, "welding inspection result is good", "correctable", and "correctable".

The second determination block makes a determination corresponding to the automatic welding propriety determination (step St107) in the first embodiment shown in FIG. That is, on the premise that the first determination block determines that "correction is possible" (step St105, Y), the second determination block determines "correctable by automatic repair welding" and "correctable by operator repair welding". It is possible to output two types of judgment results.

Since there are two types of determination blocks as described above, the feedback receiving section 38 that plays a role of updating data for use in machine learning, which will be described later, also has two types of first feedback receiving section 38a and second feedback receiving section 38b. may be provided.

However, the feedback receiving section 38 may be one system. In this case, the feedback receiving section 38 may update different data for each of the first decision block and the second decision block.

Data for generating learning data is accumulated in the first storage unit 352a and the second storage unit 352b. Typically, the weld bead shape data acquired by the shape detection unit 500 is accumulated. In addition, the first storage unit 352a and the second storage unit 352b store work type, work S/N, welding line ID, defect type, characteristic data, data indicating whether correction is possible, and automatic Data indicating whether correction is possible or not may be stored together. These data linked to the shape data may be used as label data when performing machine learning. For example, supervised learning is performed using a CNN (Convolutional Neural Network) and shape data labeled with data indicating whether automatic correction is possible. However, the types of machine learning are not limited to those mentioned above.

The amount of data accumulated in the first storage unit 352a and the second storage unit 352b can be increased by adding data. First, as shown in FIG. 9, data addition from the feedback receiver 38 is possible. Alternatively, the shape data input from the shape detection unit 500 may be processed by the shape processing unit 351 and then added to the first storage unit 352a and the second storage unit 352b.

In order to obtain learning data for machine learning, it is necessary to shape the data for efficient learning. For example, it is possible to standardize the format of the shape data, delete unnecessary data items, and delete data itself that can be statistically noise in unsupervised learning. Therefore, the first learning data generation unit 353a and the second learning data generation unit 353b perform data shaping based on the accumulated data stored in the first storage unit 352a and the second storage unit 352b, respectively, to obtain the learning data. to generate

Each of the first judgment model generation unit 354a and the second judgment model generation unit 354b performs machine learning using the learning data generated as described above to generate a judgment model. The generated determination model is implemented as the first determination unit 355a and the second determination unit 355b. The decision model may be a type of decision model that generates a score based on the input data, or it may be a classifier type of decision model that classifies the input data into classes.

Note that the determination model generation timing may be when the learning data is updated (the first learning data generation unit 353a and the second learning data generation unit 353b), and the number of determinations made using the determination model reaches a certain number of times. It may be time or when data from the shape detection unit 500 or the feedback reception unit 38 is added.

By inputting shape data processed by the shape processing unit 351 to the generated first determination unit 355a and second determination unit 355b, determination results are output. Note that the output result of the first determination unit 355a may be input to the second determination unit 355b.

The setting unit 356 sets the output content according to the output determination result. For example, the setting unit 356 provides alert information of judgment results (good/automatic repair/human repair/impossible) to the monitor MN1 connected to the host device 1, and setting conditions for repair welding to the robot control unit 26. Set information indicating Information set by the setting unit 356 is output to the outside of the inspection device 3 via the communication unit 30 .

When using machine learning, it is preferable if the determination by the determination model (the first determination unit 355a and the second determination unit 355b) can be updated so as to be more accurate according to the environment. Therefore, the inspection device 3 of the present disclosure includes a feedback receiving section 38 (see FIGS. 8 and 9).

The feedback receiving unit 38 receives feedback information from the robot control device 2, the host device 1, and the like. Examples of this feedback information include the following information. However, it is not limited to these.
・Information indicating the required specifications regarding welding (main welding, repair welding) determined by each factory or work user.
・Shape data indicating the weld location after main welding or repair welding (both data for good welding and data for poor welding).
Information indicating work conditions (including correction parameters) when the worker performs repair welding.
・Information related to correction/updating of judgment results linked to welding line ID (obtained by manual input through interface UI1, etc.)
- Update information of the base data itself stored in the first storage unit 352a and the second storage unit 352b accompanying the software update.
・Feedback information after repair welding performed at other factories.

For example, when the feedback receiving unit 38 receives feedback information such as that described above, the information is added (or or updated). Based on this feedback information, the first learning data generator 353a and the second learning data generator 353b update (generate) the learning data. The determination model (first determination unit 355a, second determination unit 355b) is updated by performing machine learning (first determination model generation unit 354a, second determination model generation unit 354b) with the updated (generated) learning data. and the accuracy of judgment is improved.

In addition, the criteria for judging the above-mentioned "welding inspection results are good", "correctable by automatic repair welding", "correctable by repair welding by workers", "cannot be corrected", etc. are the actual welding (or repair welding) It may differ depending on the factory etc. Moreover, this criterion may differ depending on the user who uses the workpiece after welding. Therefore, as described above, the feedback information is received by the feedback receiving unit 38, and the information is reflected in the learning data as described above, thereby generating learning data suitable for the desired environment, A suitable automatic judgment can be made.

Further, in the above example, the data processing unit 35 stores data related to machine learning, generates learning data, generates a judgment model, and the like. However, some of these processes may be performed by an external system (especially a processing device having a large number of GPUs placed on the cloud). For example, only the trained determination models (first determination unit 355a and second determination unit 355b) are left inside the data processing unit 35, and data accumulation, learning data generation, and determination model generation are performed by an external system. good too. Further, the destination to which the feedback information is added (updated) by the feedback receiving unit 38 may be the external system.

As described above, the automatic repair welding system includes an inspection device 3 that visually inspects the welding location of the work Wk, and a robot control device 2 that controls the robot that welds the work Wk. state determination is performed, and the state determination includes at least determining whether automatic repair welding is possible for the welding location of the work Wk, or determining whether repair welding by the operator is possible, automatic repair welding availability Determination is included, and the robot control device 2 controls the robot to perform automatic repair welding on the welding location of the workpiece Wk according to the result of the state determination. As a result, automatic repair welding that improves and stabilizes welding quality can be performed.

Further, the state determination further includes determination of the quality of the welding result at the welded portion of the work Wk and determination of whether or not the welded portion of the work Wk can be corrected. As a result, appropriate processing can be performed according to the quality of the welding result and whether or not the repair is possible, including the case where automatic repair welding is not performed.

The inspection device 3 also calculates characteristic data based on the shape data of the weld bead at the welded portion of the workpiece Wk, and performs state determination by comparing the characteristic data with a predetermined threshold value. This makes it possible to determine the state of the welded portion of the workpiece Wk based on a predetermined threshold value.

Further, the characteristic data may be an inspection score based on the master data and shape data of the weld bead at the welded portion of the workpiece Wk. As a result, the state of the welded portion of the work Wk can be determined based on the inspection score that comprehensively considers a plurality of characteristics.

In addition, the inspection device 3 includes a determination model, and the determination model is a determination model in which machine learning is performed using shape data of the weld bead at the welded portion of the work Wk, and the inspection device 3 uses the determination model to: Make a status judgment. As a result, it is possible to determine the state of the welded portion of the workpiece Wk using the determination model subjected to machine learning.

In addition, the determination model includes a first determination model for determining whether the welding result at the welded portion of the work Wk is good or bad, and a correction propriety determination at the welded portion of the work Wk, and a second determination model for determining whether automatic repair welding is possible. 2 decision models; As a result, it is possible to use different determination models to determine the quality of the welding result at the welded portion of the work Wk, determine whether the welded portion of the work Wk can be repaired, and determine whether automatic repair welding can be performed. That is, each can be determined based on different criteria.

Also, the inspection device 3 receives the feedback information and updates the judgment model using the feedback information. As a result, the determination by the determination model can be updated so as to have higher accuracy according to the environment.

The inspection device 3 also outputs the result of the state determination. As a result, a subject other than the inspection device 3 can recognize the determination result, and for example, the robot control device 2 can receive this output to perform repair welding, or an operator can perform manual welding. .

Also, the inspection device 3 or the robot control device 2 edits or changes the program used for automatic repair welding according to the result of the state determination. As a result, appropriate repair welding can be automatically performed according to the determination result.

In addition, the inspection device 3 determines the quality of the welding result at the welded portion of the work Wk, and when it is determined to be defective in the quality determination, determines whether or not the welded portion of the work Wk can be corrected, and corrects the welded portion. If it is determined that the repair is possible in the approval/disapproval determination, automatic repair welding approval/disapproval determination is performed. As a result, the state of the welded portion can be appropriately classified into four states: "good welding inspection result", "correctable by automatic repair welding", "correctable by repair welding by the operator", and "not correctable". .

Further, when it is determined that the welding portion of the work Wk cannot be automatically repair-welded in the automatic repair-weldability determination, the state determination for the work Wk may be terminated. As a result, when it is determined that automatic repair welding is not possible, it is possible to entrust the processing of the work Wk to the operator, and to quickly start visual inspection of the next work Wk.

In addition, the state determination includes determination of whether or not the welded portion of the workpiece Wk can be corrected. As a result, when it is determined that the work Wk cannot be corrected, the subsequent processing for that work Wk can be omitted, and the appearance inspection for the next work Wk can be started early.

Further, in the automatic repair welding method in the system provided with the inspection device 3 for visually inspecting the welding portion of the work Wk and the robot control device 2 for controlling the robot for welding the work Wk, the inspection device 3 inspects the work Wk. The state of the welding location is determined, and the state determination includes at least determining whether automatic repair welding is possible for the welding location of the work Wk, or determining whether repair welding by the operator is possible. The robot controller 2 controls the robot so as to perform automatic repair welding on the welding portion of the workpiece Wk according to the result of the state determination. As a result, automatic repair welding that improves and stabilizes welding quality can be performed.

Various embodiments have been described above with reference to the drawings, but it goes without saying that the present disclosure is not limited to such examples. It is obvious that a person skilled in the art can conceive of various modifications, modifications, substitutions, additions, deletions, and equivalents within the scope of the claims. Naturally, it is understood that it belongs to the technical scope of the present disclosure. In addition, the constituent elements of the various embodiments described above may be combined arbitrarily without departing from the gist of the invention.

INDUSTRIAL APPLICABILITY The present disclosure is useful as a repair welding system and a repair welding method capable of performing repair welding.

1 host device 2 robot control device 2a robot control device 2b robot control device 3 inspection device 4 welding power supply device 10 communication unit 11 processor 12 memory 13 cell control unit 20 communication unit 21 processor 22 memory 23a program editing unit 23b program calling unit 23c program Storage unit 24 Calculation unit 25 Inspection device control unit 26 Robot control unit 27 Welding power supply control unit 30 Communication unit 31 Processor 32 Memory 34 Shape detection control unit 35 Data processing unit 36 Judgment threshold storage unit 37 Judgment unit 38 Feedback reception unit 38a First Feedback receiving section 38b Second feedback receiving section 50 Communication section 200 Manipulator 300 Wire feeder 301 Welding wire 351 Shape processing section 352a First storage section 352b Second storage section 353a First learning data generation section 353b Second learning data generation section 354a First judgment model generation unit 354b Second judgment model generation unit 355a First judgment unit 355b Second judgment unit 356 Setting unit 400 Welding torch 500 Shape detection unit 1000 Automatic repair welding system 1000a Automatic repair welding system 1000b Automatic repair welding system 1000c Automatic Repair Welding System MC1 Main Welding Robot MC2 Inspection Robot MC3 Repair Welding Robot MC23 Inspection/Repair Welding Robot MN1 Monitor UI1 Interface Wk Work

Claims (13)

Equipped with an inspection device that visually inspects the welding location of the work and a robot control device that controls the robot that welds the work,
The inspection device determines the state of the welded portion of the work, and if it is determined that a defective portion exists in the welded portion of the work and that the defective portion can be corrected as the state determination, the welded portion Based on the state determination result, generating a type indicating a repair welding correction method for correcting the defective portion and a parameter used for repairing the defective portion ,
The inspection device holds a program generation logic table that associates the state determination result with the type and parameter of the repair welding, and based on the state determination result and the program generation logic table, the repair welding In addition to generating the type and the parameter, determining whether the repair welding can be performed by the robot by automatic correction,
When the inspection device determines that the repair welding can be automatically performed by the robot, the robot control device performs welding of the work according to the type and the parameters of the repair welding . controlling the robot to automatically modify the repair weld to a location;
repair welding system. The state determination further includes determination of the quality of the welding result at the welded portion of the work and determination of whether or not the welded portion of the work can be corrected.
The repair welding system according to claim 1. The inspection device calculates characteristic data based on the shape data of the weld bead at the welded portion of the workpiece, and compares the characteristic data with a predetermined threshold value to perform the state determination.
The repair welding system according to claim 1 or 2 . The characteristic data is an inspection score based on master data and shape data of the weld bead at the welded portion of the workpiece,
The repair welding system according to claim 3 . The inspection device comprises a judgment model,
The judgment model is a judgment model in which machine learning is performed using shape data of a weld bead at a welded portion of the workpiece,
The inspection device performs the state determination using the determination model.
The repair welding system according to claim 1. The decision model is
a first judgment model for judging the quality of the welding result at the welded portion of the work and determining whether or not the welded portion of the work can be corrected;
a second determination model that determines whether the automatic correction of the repair welding is possible;
The repair welding system according to claim 5 . The inspection device receives feedback information and uses the feedback information to update the decision model.
The repair welding system according to claim 6 . The inspection device outputs a result of the state determination,
Repair welding system according to any one of claims 1 to 7 . The inspection device or the robot control device,
Editing or changing the program used for the automatic correction of the repair welding according to the result of the state determination;
Repair welding system according to any one of claims 1 to 8 . The inspection device is
Determining the quality of the welding result at the welding location of the work,
If it is determined to be defective in the quality determination, determine whether or not the welded portion of the work can be corrected,
If it is determined that the repairability can be corrected in the repairability determination at the welding location, determine whether the repair welding can be automatically corrected .
Repair welding system according to any one of claims 1 to 9 . When it is determined that the welding location of the work cannot be automatically repaired in the determination of whether or not the automatic repair is possible for the repair welding, the state determination is terminated for the work.
Repair welding system according to any one of claims 1 to 10 . The state determination includes a determination of whether or not the welded portion of the workpiece can be corrected,
terminating the state determination when it is determined that the modification is not modifiable in the modifiability determination;
Repair welding system according to any one of claims 1 to 11 . In a repair welding system equipped with an inspection device that visually inspects a welded portion of a work and a robot control device that controls a robot that welds the work,
The inspection device determines the state of the welded portion of the work, and if it is determined that a defective portion exists in the welded portion of the work and that the defective portion can be corrected as the state determination, the welded portion Based on the state determination result, generating a type indicating a repair welding correction method for correcting the defective portion and a parameter used for correcting the defective portion ,
The inspection device holds a program generation logic table that associates the state determination result with the type and parameter of the repair welding, and based on the state determination result and the program generation logic table, the repair welding In addition to generating the type and the parameter, determining whether the repair welding can be performed by the robot by automatic correction,
When the inspection device determines that the repair welding can be automatically performed by the robot, the robot control device performs welding of the work according to the type and the parameters of the repair welding . controlling the robot to automatically modify the repair weld to a location;
Repair welding method.

Images