Bead appearance inspection device and bead appearance inspection system

Abstract

This bead appearance inspection device is provided with: an acquisition unit which acquires input data pertaining to a weld bead; a storage unit which stores a first determination reference and a second determination reference to be used for inspecting for a failure in the weld bead; a first determination unit which executes a first inspection determination for the weld bead; and k (k is an integer equal to or greater than 1) second determination units which execute a second inspection determination for the weld bead, wherein, an appearance inspection result for the weld bead is created and output by using a determination result of whether a first inspection result acquired by the first inspection determination meets the first determination reference and a determination result of whether a second inspection result meets the second determination reference.

Description

[Technical field]

[0001] The present disclosure relates to a bead appearance inspection device and a bead appearance inspection system. [Background technology]

[0002] Patent Document 1 discloses a shape inspection device that projects a slit light onto a weld bead, captures shape lines formed sequentially on the weld bead by scanning the slit light, and acquires the three-dimensional shape of the weld bead as point cloud data based on the captured image data of each shape line formed sequentially. This shape inspection device sets an arbitrary cutting line different from the shape line formed by scanning the slit light in response to an input on the weld bead displayed based on the point cloud data, and calculates the cross-sectional shape of the weld bead at the cutting line based on the point cloud data corresponding to the cutting line. In addition, the shape inspection device compares various feature amounts calculated according to the calculated cross-sectional shape with the allowable ranges of various feature amounts registered in advance, and judges whether the feature amounts are good or bad. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2012-37487 Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure provides a bead appearance inspection device and a bead appearance inspection system that make it more efficient for a user to perform setting operations regarding bead appearance inspection of workpieces produced by main welding. [Means for solving the problem]

[0005] The present disclosure provides an acquisition unit that acquires input data related to a weld bead, and a first judgment criterion and a second judgment criterion that are set by a user and used for inspecting defects in the weld bead. Obtained, andThe present invention provides a bead visual inspection device comprising a first judgment unit that performs a first inspection judgment on the weld bead, and k (k: an integer greater than or equal to 1) second judgment units that perform a second inspection judgment on the weld bead, wherein the k second judgment units output second inspection results obtained by the second inspection judgment to the first judgment unit, and the first judgment unit creates and outputs visual inspection results of the weld bead using a judgment result of whether the first inspection result obtained by the first inspection judgment satisfies the first judgment criterion and a judgment result of whether the input second inspection result satisfies the second judgment criterion.

[0006] The present disclosure also provides a method for inspecting defects in a weld bead by using an acquisition unit that acquires input data related to the weld bead and a first judgment criterion and a second judgment criterion that are set by a user. Obtained, and The present invention provides a bead visual inspection system comprising a first judgment unit that performs a first inspection judgment on the weld bead, and k (k: an integer greater than or equal to 1) second judgment units that perform a second inspection judgment on the weld bead, wherein the k second judgment units output second inspection results obtained by the second inspection judgment to the first judgment unit, and the first judgment unit creates and outputs visual inspection results of the weld bead using a judgment result of whether the first inspection result obtained by the first inspection judgment satisfies the first judgment criterion and a judgment result of whether the input second inspection result satisfies the second judgment criterion. provide. Effect of the Invention

[0007] According to the present disclosure, the user's setting operations for bead appearance inspection of workpieces produced by main welding can be made more efficient. [Brief description of the drawings]

[0008] [Figure 1] Schematic diagram showing an example of the system configuration of a welding system [Diagram 2] FIG. 1 is a diagram showing an example of the internal configuration of an inspection control device, a robot control device, and a higher-level device according to a first embodiment; [Diagram 3]A sequence diagram showing an example of a series of processing procedures including main welding, bead appearance inspection, and repair welding by the welding system according to the first embodiment. [Figure 4] A table showing examples of first and second test judgments for each of a plurality of test items [Diagram 5] FIG. 1 is a sequence diagram showing an example of a processing procedure of a first inspection judgment (point group comparison) and a second inspection judgment to an Nth inspection judgment (AI judgment) according to the first embodiment; [Figure 6] FIG. 13 is a diagram showing an example of the internal configuration of an inspection control device, a robot control device, and a higher-level device according to a second embodiment. [Figure 7] A sequence diagram showing an example of a series of processing procedures including main welding, bead appearance inspection, and repair welding by the welding system according to the second embodiment. [Figure 8] FIG. 11 is a sequence diagram showing an example of a processing procedure of the first inspection judgment (point group comparison) and the second inspection judgment to the Nth inspection judgment (AI judgment) according to the second embodiment; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] (Background to this disclosure) Conventionally, there have been known device configurations that automatically inspect the external shape of a weld bead, such as determining that a workpiece produced by main welding is non-defective when calculated values ​​of features (e.g., bead width, bead height, etc.) related to the shape of the weld bead are within an allowable range, as in Patent Document 1. However, at actual welding sites, workers often visually inspect the external shape of the weld bead to determine whether the main welding of the workpiece was successful.

[0010] In the visual inspection of weld beads, in addition to the above-mentioned feature quantities related to the shape of the weld bead, there may be a wide range of inspection items, such as the displacement of the weld bead, the presence or absence of holes, and the presence or absence of welding defects such as spatters, and the criteria for determining whether or not a product is good are often not uniform depending on the user. For this reason, in the visual inspection of weld beads, not only are the inspection items different for each user, but the quality of the finished workpiece also differs for each user, so it is considered that there is room for improvement over the conventional technology in that greater usability of the visual inspection is required along with customizability that allows the inspection items to be arbitrarily adjusted.

[0011] Therefore, in the following embodiments, examples of a bead appearance inspection device and a bead appearance inspection system that make it more efficient for a user to perform setting operations regarding bead appearance inspection of workpieces produced by main welding will be described.

[0012] Hereinafter, with reference to the drawings as appropriate, a detailed description will be given of an embodiment specifically disclosing a bead appearance inspection device and a bead appearance inspection system according to the present disclosure. However, more detailed description than necessary may be omitted. For example, detailed description of already well-known matters and duplicated description of substantially the same configuration may be omitted. This is to avoid the following description becoming unnecessarily redundant and to facilitate understanding by those skilled in the art. Note that the attached drawings and the following description are provided to enable those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter described in the claims.

[0013] (Embodiment 1) The bead appearance inspection device according to the first embodiment inputs input data related to the weld bead of the workpiece produced by main welding, and performs a first inspection judgment on the shape of the weld bead based on a comparison between the input data and the master data using the input data and master data of a non-defective workpiece, and is equipped with k (k is an integer of 1 or more) types of artificial intelligence and performs a second inspection judgment on welding defects of the weld bead based on processing of the k types of artificial intelligence targeting the input data. Welding defects of the weld bead include, for example, holes, pits, undercuts, spatters, and protrusions. Note that the welding defects are not limited to those mentioned above. The bead appearance inspection device outputs the results of the appearance inspection of the weld bead to an output device based on the judgment results of the first appearance inspection and the k second inspection judgments.

[0014] Hereinafter, the object to be welded (e.g., metal) is defined as the "original workpiece," the object produced (manufactured) by this welding is defined as the "workpiece," and the object in which defective welding detected during visual inspection of the "workpiece" has been repair welded is defined as the "repair workpiece."

[0015] The process in which an original workpiece is joined to another original workpiece by a welding robot to produce a workpiece is defined as "main welding," and the process in which a welding robot repairs or corrects defective parts of the workpiece is defined as "repair welding."

[0016] It should be noted that the "work" or "repair work" is not limited to a work produced by a single main welding operation, but may be a composite work produced by two or more main welding operations.

[0017] (Welding system configuration) FIG. 1 is a schematic diagram showing an example of the system configuration of a welding system 100. The welding system 100 includes a higher-level device 1 connected to an external storage ST, an input interface UI1, and a monitor MN1, a robot control device 2, an inspection control device 3, a sensor 4, a main welding robot MC1a, and a repair welding robot MC1b. The main welding robot MC1a and the repair welding robot MC1b may be configured as separate robots, or may be configured as the same welding robot MC1. To make the following description easier to understand, the description will be given assuming that the welding robot MC1 executes the main welding and repair welding processes. Note that FIG. 1 shows only one pair of one robot control device 2, a main welding robot MC1a, and a repair welding robot MC1b, but a plurality of such pairs may be provided. In FIG. 1, the sensor 4 is shown as a separate body from the welding robot MC1, but may be provided integrally with the welding robot MC1 (see FIG. 2).

[0018] The higher-level device 1 controls the start and completion of the main welding performed by the welding robot MC1 through the robot controller 2. For example, the higher-level device 1 reads out welding-related information input or set in advance by a user (e.g., a welding operator or a system administrator; the same applies below) from the external storage ST, and generates a main welding execution command including the contents of the welding-related information using the welding-related information, and sends it to the corresponding robot controller 2. When the main welding by the welding robot MC1 is completed, the higher-level device 1 receives a main welding completion report indicating that the main welding by the welding robot MC1 has been completed from the robot controller 2, updates the status to indicate that the corresponding main welding has been completed, and records it in the external storage ST. Note that the execution command for the main welding described above is not limited to being generated by the higher-level device 1, and may be generated by, for example, an operation panel (e.g., PLC: Programmable Logic Controller) of equipment in a factory or the like where the main welding is performed, or an operation panel (e.g., TP: Teach Pendant) of the robot controller 2. Note that the teach pendant (TP) is a device for operating the welding robot MC1 connected to the robot controller 2.

[0019] Furthermore, the higher-level device 1 comprehensively controls the start and completion of bead appearance inspection using the robot control device 2, the inspection control device 3, and the sensor 4. For example, when the higher-level device 1 receives a report of the completion of main welding from the robot control device 2, it generates an execution command for a bead appearance inspection of the workpiece produced by the welding robot MC1 and sends it to each of the robot control device 2 and the inspection control device 3. When the bead appearance inspection is completed, the higher-level device 1 receives an appearance inspection report indicating that the bead appearance inspection has been completed from the inspection control device 3, updates the status to indicate that the corresponding bead appearance inspection has been completed, and records the updated status in the external storage ST.

[0020] Furthermore, the host device 1 centrally controls the start and completion of repair welding performed by the welding robot MC1 via the robot control device 2. For example, when the host device 1 receives an appearance inspection report from the inspection control device 3, it generates an execution command for repair welding of the workpiece produced by the welding robot MC1 and sends it to the robot control device 2. When the repair welding is completed, the host device 1 receives a repair welding completion report from the robot control device 2 indicating that the repair welding has been completed, updates the status to indicate that the corresponding repair welding has been completed, and records the report in the external storage ST.

[0021] Here, the welding-related information is information indicating the contents of the main welding performed by the welding robot MC1, and is created in advance for each process of the main welding and registered in the external storage ST. The welding-related information includes, for example, the number of original workpieces used in the main welding, workpiece information including the ID, name, and welding location of the original workpieces used in the main welding, the scheduled execution date of the main welding, the production number of workpieces to be welded, and various welding conditions during the main welding. The welding-related information does not need to be limited to the data of the above items. Based on an execution command for the main welding sent from the upper device 1, the robot control device 2 causes the welding robot MC1 to start the execution of the main welding using the original workpieces specified in the execution command. The above-mentioned welding-related information is not limited to being managed by the upper device 1 with reference to the external storage ST, and may be managed, for example, in the robot control device 2. In this case, since the robot control device 2 can grasp the state in which the main welding is completed, the actual execution date may be managed instead of the scheduled execution date for the welding process among the welding-related information. In this specification, the type of welding is not limited, but for ease of understanding, a process of joining a plurality of original workpieces to produce one workpiece will be described as an example.

[0022] The higher-level device 1 is connected to the monitor MN1, the input interface UI1, and the external storage ST so as to enable input and output of data therebetween, and is further connected to the robot control device 2 so as to enable data communication therebetween. The higher-level device 1 may be a terminal device P1 that integrally includes the monitor MN1 and the input interface UI1, and may further integrally include the external storage ST. In this case, the terminal device P1 is a PC (Personal Computer) that is used by a user prior to the execution of the actual welding. The terminal device P1 is not limited to the above-mentioned PC, and may be a computer device having a communication function, such as a smartphone or a tablet terminal.

[0023] The monitor MN1 may be configured using a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electroluminescence). The monitor MN1 may display a screen showing a notification that main welding has been completed, a notification that bead appearance inspection has been completed, or a notification that repair welding has been completed, which are output from the higher-level device 1. Also, instead of or together with the monitor MN1, a speaker (not shown) may be connected to the higher-level device 1, and the higher-level device 1 may output a voice message via the speaker indicating the notification that main welding has been completed, the notification that bead appearance inspection has been completed, or the notification that repair welding has been completed.

[0024] The input interface UI1 is a user interface that detects input operations by the user and outputs the same to the higher-level device 1, and may be configured using, for example, a mouse, a keyboard, a touch panel, etc. The input interface UI1 accepts input operations, for example, when the user creates welding-related information, and when the user sends an execution command for main welding to the robot control device 2.

[0025] The external storage ST is configured using, for example, a hard disk drive or a solid state drive. The external storage ST stores, for example, data of welding-related information created for each main welding, the status (production status) of the work produced by the main welding or the repair work repaired by repair welding, and work information of the work or the repair work (see above).

[0026] A robot control device 2 as an example of a bead appearance inspection device is connected to a higher-level device 1 so as to be able to communicate data with the robot control device 2, and is also connected to a welding robot MC1 so as to be able to communicate data with the robot control device 2. When the robot control device 2 receives an execution command for main welding sent from the higher-level device 1, the robot control device 2 controls the corresponding welding robot MC1 based on the execution command to execute the main welding. When the robot control device 2 detects the completion of the main welding, the robot control device 2 generates a main welding completion report indicating that the main welding has been completed, and notifies the higher-level device 1 of this. This allows the higher-level device 1 to properly detect the completion of the main welding by the robot control device 2. The method of detecting the completion of the main welding by the robot control device 2 may be, for example, a method of determining based on a signal indicating the completion of the main welding from a sensor (not shown) provided in the wire feeder 300, or may be a known method, and the content of the method of detecting the completion of the main welding may not be limited.

[0027] Furthermore, when the robot control device 2 receives an execution command for a bead visual inspection sent from the higher-level device 1, it controls a welding robot MC1 (see FIG. 2) equipped with a sensor 4 in accordance with an appearance inspection program created or prepared in advance by the robot control device 2, and executes a bead visual inspection of the corresponding workpiece based on the execution command. An appearance inspection report indicating that the bead visual inspection has been completed is sent from the inspection control device 3 to the higher-level device 1, but it may also be sent to the higher-level device 1 by the robot control device 2 itself, or from the robot control device 2 receiving an instruction from the inspection control device 3. This allows the higher-level device 1 to properly detect the completion of the bead visual inspection.

[0028] Furthermore, when the robot control device 2 receives an execution command for repair welding sent from the higher-level device 1, it controls the corresponding welding robot MC1 based on the execution command in accordance with the repair welding program created by the inspection control device 3 to execute repair welding. When the robot control device 2 detects the completion of repair welding, it generates a repair welding completion report indicating that repair welding has been completed and notifies the higher-level device 1. This allows the higher-level device 1 to properly detect the completion of repair welding based on the robot control device 2. Note that the method of detecting the completion of repair welding by the robot control device 2 may be, for example, a method of determining based on a signal indicating the completion of repair welding from a sensor (not shown) provided in the wire feeder 300, or a known method, and the content of the method of detecting the completion of repair welding need not be limited.

[0029] The welding robot MC1 is connected to a robot controller 2 so as to be able to communicate data with the robot controller 2. The welding robot MC1 performs main welding or repair welding as instructed by the higher-level device 1 under the control of the corresponding robot controller 2. As described above, the welding robot MC1 may be composed of a main welding robot MC1a provided for main welding and a repair welding robot MC1b provided for repair welding. In addition, when the sensor 4 is integrally attached to the welding robot MC1, the welding robot MC1 drives the sensor 4 according to an appearance inspection program to assist in the execution of the bead appearance inspection as instructed by the higher-level device 1.

[0030] The inspection control device 3, which is an example of a bead appearance inspection device, is connected to the upper device 1, the robot control device 2, and the sensor 4 so that data can be communicated between them. When the inspection control device 3 receives an execution command for a bead appearance inspection sent from the upper device 1, it executes a bead appearance inspection (for example, an inspection of whether or not the weld bead formed on the workpiece satisfies a predefined welding standard) of the welded portion of the workpiece produced by the welding robot MC1 together with the sensor 4. Details of the bead appearance inspection will be described later with reference to Figs. 4 and 5. For example, the inspection control device 3 performs a bead appearance inspection based on the welded portion information of the workpiece included in the execution command for the bead appearance inspection, using input data on the shape of the weld bead acquired by the sensor 4 (for example, point cloud data that can identify the three-dimensional shape of the weld bead) based on a comparison with master data of a good workpiece predefined for each workpiece. Hereinafter, such a bead appearance inspection is defined as a "first inspection judgment". The inspection control device 3 is also equipped with k (k: integer equal to or greater than 1) types of artificial intelligence (AI), forms neural networks using the AI, and performs a bead appearance inspection to determine the presence or absence of welding defects based on the AI ​​targeting the above-mentioned input data. Hereinafter, such a bead appearance inspection is defined as a "second inspection judgment." In the first embodiment, the inspection control device 3 is capable of executing the above-mentioned first inspection judgment and second inspection judgment. The inspection control device 3 performs a comprehensive judgment of the bead appearance inspection using the results of executing the first inspection judgment and the second inspection judgment, and generates an appearance inspection report including this comprehensive judgment result and a notification that the bead appearance inspection has been completed, sends it to the higher-level device 1, and outputs it to the monitor MN2.

[0031] Furthermore, when the inspection control device 3 determines that a welding defect has been detected in the bead visual inspection of the workpiece, it creates a repair welding program for making corrections such as repairs to the welding defect using the visual inspection results including the position information of the welding defect points (so-called detection points). The inspection control device 3 sends this repair welding program and the visual inspection results to the robot control device 2 in association with each other.

[0032] The sensor 4 is connected to the inspection control device 3 so as to be able to communicate data with the inspection control device 3. When the sensor 4 is attached to the welding robot MC1 (see FIG. 2), the sensor 4 is operable to perform a three-dimensional scan of a stage on which a workpiece Wk is placed, in response to the driving of the manipulator 200 under the control of the robot control device 2. The sensor 4 acquires data (e.g., point cloud data OD1 described later) capable of identifying the three-dimensional shape of the workpiece placed on the stage (see FIG. 2) in response to the driving of the manipulator 200 under the control of the robot control device 2, and sends the data to the inspection control device 3.

[0033] The monitor MN2, which is an example of an output device, may be configured using a display device such as an LCD or an organic EL. The monitor MN2 displays a screen showing, for example, a notification that the bead visual inspection has been completed, or the notification and the result of the bead visual inspection (for example, the result of the above-mentioned overall judgment), output from the inspection control device 3. Also, instead of the monitor MN2, or together with the monitor MN2, a speaker (not shown) may be connected to the inspection control device 3, and the inspection control device 3 may output, via the speaker, a sound showing the notification that the bead visual inspection has been completed, or the notification and the result of the bead visual inspection (for example, the result of the above-mentioned overall judgment).

[0034] Fig. 2 is a diagram showing an example of the internal configuration of the inspection control device 3, the robot control device 2, and the higher-level device 1 according to the first embodiment. To make the explanation easier to understand, the monitors MN1, MN2, and the input interface UI1 are omitted from Fig. 2. Note that the workpiece Wk shown in Fig. 2 may be an original workpiece placed before main welding is performed, a workpiece to be subjected to bead appearance inspection (i.e., a workpiece produced by main welding), or a workpiece to be subjected to repair welding.

[0035] Welding robot MC1 executes various processes, such as main welding instructed by upper device 1, movement of sensor 4 during bead appearance inspection, and repair welding, under the control of robot control device 2. Welding robot MC1 performs, for example, arc welding in the main welding or repair welding process. However, welding robot MC1 may perform other welding methods (for example, laser welding, gas welding) other than arc welding. In this case, although not shown in the figures, a laser head may be connected to a laser oscillator via an optical fiber instead of welding torch 400. Welding robot MC1 is configured to include at least manipulator 200, wire feeder 300, welding wire 301, and welding torch 400.

[0036] Manipulator 200 has an articulated arm, and moves each arm based on a control signal from robot control unit 25 of robot control device 2. In this way, manipulator 200 can change the positional relationship between workpiece Wk and welding torch 400 (for example, the angle of welding torch 400 with respect to workpiece Wk) by driving the arm.

[0037] Wire feeder 300 controls the feeding speed of welding wire 301 based on a control signal from robot controller 2. Wire feeder 300 may include a sensor (not shown) capable of detecting the remaining amount of welding wire 301. Robot controller 2 can detect the completion of the main welding or repair welding process based on the output of this sensor.

[0038] Welding wire 301 is held by welding torch 400. When power is supplied from power supply device 500 to welding torch 400, an arc is generated between the tip of welding wire 301 and workpiece Wk, and arc welding is performed. For convenience of explanation, illustration and explanation of the configuration for supplying shielding gas to welding torch 400 are omitted.

[0039] The higher-level device 1 uses welding-related information input or set in advance by a user to generate execution commands for various processes, such as main welding, bead appearance inspection, and repair welding, and sends the commands to the robot control device 2. As described above, when the sensor 4 is integrally attached to the welding robot MC1, the execution command for the bead appearance inspection is sent to both the robot control device 2 and the inspection control device 3. The higher-level device 1 is configured to include at least a communication unit 10, a processor 11, and a memory 12.

[0040] The communication unit 10 is connected to the robot control device 2 and the external storage ST so as to be able to communicate data with each other. The communication unit 10 sends execution commands for various processes of main welding, bead visual inspection, or repair welding, which are generated by the processor 11, to the robot control device 2. The communication unit 10 receives a main welding completion report, visual inspection report, or repair welding completion report sent from the robot control device 2, and outputs them to the processor 11. Note that the execution command for main welding or repair welding may include, for example, control signals for controlling each of the manipulator 200, wire feeder 300, and power supply device 500 provided in the welding robot MC1.

[0041] The processor 11 is configured using, for example, a CPU (Central Processing Unit) or an FPGA (Field Programmable Gate Array), and performs various processes and controls in cooperation with the memory 12. Specifically, the processor 11 refers to a program stored in the memory 12 and executes the program to functionally realize the cell control unit 13.

[0042] The memory 12 has, for example, a RAM (Random Access Memory) as a work memory used when executing the processing of the processor 11, and a ROM (Read Only Memory) that stores a program that defines the processing of the processor 11. Data generated or acquired by the processor 11 is temporarily stored in the RAM. The program that defines the processing of the processor 11 is written in the ROM. The memory 12 also stores data on welding-related information read from the external storage ST, the status of the work or repair work, and data on work information (see above) of the work or repair work sent from the robot control device 2.

[0043] The cell control unit 13 generates an execution command for performing main welding, bead visual inspection of the workpiece, or repair welding based on the welding-related information stored in the external storage ST. Also, the cell control unit 13 generates an execution command for an appearance inspection program for driving the welding robot MC1 during bead visual inspection of the workpiece Wk (e.g., workpiece) after main welding, and further, an execution command for an appearance inspection program including this appearance inspection program based on the welding-related information stored in the external storage ST. Note that this appearance inspection program may be created in advance and stored in the external storage ST. In this case, the cell control unit 13 simply reads and acquires the appearance inspection program from the external storage ST. The cell control unit 13 may generate different execution commands for each of the various processes of main welding or repair welding performed by the welding robot MC1. The execution command for main welding or repair welding, or the execution command for the appearance inspection program including the appearance inspection program generated by the cell control unit 13 is sent to the corresponding robot control device 2, or to the robot control device 2 and the inspection control device 3, via the communication unit 10.

[0044] The robot control device 2 controls the processing of the corresponding welding robot MC1 (e.g., the sensor 4, the manipulator 200, the wire feeder 300, and the power supply device 500) based on an execution command for main welding, bead appearance inspection, or repair welding sent from the higher-level device 1. The robot control device 2 is configured to include at least a communication unit 20, a processor 21, and a memory 22.

[0045] The communication unit 20 is connected to enable data communication between the host device 1, the inspection control device 3, and the welding robot MC1. Although the illustration is simplified in Fig. 2, data is transmitted and received between the robot control unit 25 and the manipulator 200, between the robot control unit 25 and the wire feeder 300, and between the power supply control unit 26 and the power supply device 500 via the communication unit 20. The communication unit 20 receives a command to perform main welding, bead appearance inspection, or repair welding sent from the host device 1. The communication unit 20 sends work information of the work produced by main welding or the repair work produced by correction by repair welding to the host device 1.

[0046] Here, the work information includes not only the ID of the work or the repair work, but also at least the ID, name, and welding location of the original work used for the main welding, the welding conditions when the main welding is performed, and the welding conditions when the repair welding is performed. Furthermore, the work information may include information (e.g., coordinates) indicating the position of a detection point indicating a defective part of the work. Furthermore, the welding conditions or the repair welding conditions are, for example, the material and thickness of the original work, the material and wire diameter of the welding wire 301, the type of shielding gas, the flow rate of the shielding gas, the set average value of the welding current, the set average value of the welding voltage, the feed speed and feed amount of the welding wire 301, the number of welding times, the welding time, and the like. Furthermore, in addition to these, for example, information indicating the type of main welding or repair welding (e.g., TIG welding, MAG welding, pulse welding), and the moving speed and moving time of the manipulator 200 may be included.

[0047] Processor 21 is configured using, for example, a CPU or FPGA, and performs various processes and controls in cooperation with memory 22. Specifically, processor 21 refers to a program stored in memory 22 and executes the program to functionally realize a welding program creation unit 23, a calculation unit 24, a robot control unit 25, and a power supply control unit 26.

[0048] The memory 22 has, for example, a RAM as a work memory used when executing the processing of the processor 21, and a ROM that stores a program that defines the processing of the processor 21. Data generated or acquired by the processor 21 is temporarily stored in the RAM. The program that defines the processing of the processor 21 is written in the ROM. The memory 22 also stores data of execution commands for main welding, bead appearance inspection, or repair welding sent from the higher-level device 1, and data of work information of a workpiece produced by main welding or a repair workpiece produced by repair welding. The memory 22 also stores a main welding program for main welding executed by the welding robot MC1. The main welding program is a program that defines a specific procedure (step) of main welding for joining multiple original workpieces using welding conditions in main welding.

[0049] Based on an execution command for main welding sent from higher-level device 1 via communication unit 20, main welding program creation unit 23 generates a main welding program for main welding to be performed by welding robot MC1 using work information (e.g., ID, name, and welding location of the main work) of each of a plurality of original workpieces included in the execution command. The main welding program may include various parameters such as welding current, welding voltage, offset amount, welding speed, and attitude of welding torch 400 for controlling power supply device 500, manipulator 200, wire feeder 300, welding torch 400, etc. during the execution of main welding. The main welding program may be stored in processor 21 or in RAM in memory 22.

[0050] Calculation unit 24 performs various calculations. For example, calculation unit 24 performs calculations of parameters for controlling welding robot MC1 (specifically, manipulator 200, wire feeder 300, and power supply 500) controlled by robot control unit 25 based on the main welding program generated by main welding program creation unit 23.

[0051] Robot control unit 25 generates a control signal for driving welding robot MC1 (specifically, each of manipulator 200, wire feeder 300, and power supply unit 500) based on the actual welding program generated by actual welding program creation unit 23. Robot control unit 25 sends the generated control signal to welding robot MC1. Also, based on the appearance inspection program sent from higher-level device 1, robot control unit 25 drives manipulator 200 of welding robot MC1 during bead appearance inspection so as to target the operating range of welding robot MC1 defined in the actual welding program. This allows sensor 4 (see FIG. 1) attached to welding robot MC1 to move with the operation of welding robot MC1 and obtain input data related to the shape of the weld bead of workpiece Wk (for example, point cloud data capable of identifying the three-dimensional shape of the weld bead).

[0052] Power supply control unit 26 drives power supply device 500 based on the main welding program generated by main welding program creation unit 23 and the calculation result of calculation unit 24.

[0053] The inspection control device 3 controls the bead appearance inspection process of the workpiece or repair workpiece produced by main welding by the welding robot MC1 based on an execution command of the appearance inspection sent from the upper device 1. The bead appearance inspection is, for example, an inspection of whether or not the weld bead formed on the workpiece or repair workpiece satisfies a predetermined welding standard (for example, a quality standard), and is composed of the first inspection judgment and the second inspection judgment described above. To simplify the following description, the inspection control device 3 judges whether or not the weld bead formed on the workpiece Wk (for example, a workpiece or repair workpiece) satisfies a predetermined welding standard by a comprehensive judgment based on the results of the first inspection judgment and the second inspection judgment described above, based on input data regarding the shape of the weld bead acquired by the sensor 4 (for example, point cloud data that can identify the three-dimensional shape of the weld bead). The inspection control device 3 is configured to include at least a communication unit 30, a processor 31, a memory 32, and an inspection result storage unit 33.

[0054] The communication unit 30 is connected to enable data communication between the higher-level device 1, the robot control device 2, and the sensor 4. Although the illustration is simplified in FIG. 2, data is sent and received between the shape detection control unit 35 and the sensor 4 via the communication unit 30. The communication unit 30 receives a command to execute a bead appearance inspection sent from the higher-level device 1. The communication unit 30 sends a comprehensive judgment result of the bead appearance inspection using the sensor 4 (for example, the presence or absence of welding defects in the weld beads of the work or repair work, and the type and location of the welding defects) to the higher-level device 1.

[0055] Processor 31 is configured using, for example, a CPU or FPGA, and performs various processes and controls in cooperation with memory 32. Specifically, processor 31 refers to a program stored in memory 32 and executes the program to functionally realize a judgment threshold storage unit 34, a shape detection control unit 35, a data processing unit 36, an inspection result judgment unit 37, and a repair welding program creation unit 38.

[0056] The memory 32 as an example of the storage unit includes, for example, a RAM as a work memory used when executing the processing of the processor 31, and a ROM that stores a program that defines the processing of the processor 31. The RAM temporarily stores data generated or acquired by the processor 31. The ROM has a program that defines the processing of the processor 31 written therein. The memory 32 also stores data of a command to perform bead appearance inspection of the workpiece sent from the upper device 1, and data of workpiece information of the workpiece generated by main welding or the repair workpiece generated by repair welding. The memory 32 also stores data of a repair welding program created by a repair welding program creation unit 38. The repair welding program is a program that defines a specific procedure (step) of repair welding for correcting a defective welded portion of a weld bead by using the welding conditions in the repair welding and the position information of a corresponding portion (corresponding point) on the movement trajectory of the welding robot MC1 that is closest to the detection point (see above). This program is created by the repair welding program creation unit 38 and sent from the inspection control device 3 to the robot control device 2.

[0057] The inspection result storage unit 33 is configured using, for example, a hard disk or a solid state drive. The inspection result storage unit 33 stores data indicating the results of a bead appearance inspection of a welded portion in the workpiece Wk (for example, a workpiece or a repair workpiece) as an example of data generated or acquired by the processor 31. The data indicating the results of this bead appearance inspection is generated, for example, by the inspection result determination unit 37 (specifically, any one of the first inspection determination unit 371, the second inspection determination unit 372, ..., the Nth inspection determination unit 37N included in the inspection result determination unit 37).

[0058] The judgment threshold storage unit 34 is configured, for example, by a cache memory provided in the processor 31, and stores thresholds (for example, each threshold set for each type of welding defect) for all bead appearance inspection processes of the first inspection judgment unit 371, ..., Nth inspection judgment unit 37N included in the inspection result judgment unit 37, which are set in advance by a user operation and correspond to the welding location. The respective thresholds are, for example, an allowable range for the positional deviation of the weld bead, respective thresholds for the length, height, and width of the weld bead, and respective thresholds for holes, pits, undercuts, and spatters. The judgment threshold storage unit 34 may store an allowable range (for example, a minimum allowable value, a maximum allowable value, etc.) that satisfies a minimum welding standard (quality) required by a customer or the like as each threshold for the bead appearance inspection after repair welding. These thresholds are executed by any one of the first inspection judgment units 371, ..., Nth inspection judgment units 37N included in the inspection result judgment unit 37, and are used in a process of judging whether or not each inspection result created by each of the first inspection judgment units 371, ..., Nth inspection judgment units 37N passes the bead appearance inspection. Furthermore, the judgment threshold storage unit 34 may store an upper limit value of the number of bead appearance inspections for each welding point. In this way, when the number of times exceeds a predetermined upper limit value when repairing a defective point by repair welding, the inspection control device 3 judges that it is difficult or impossible to repair the defective point by automatic repair welding by the welding robot MC1, and can suppress a decrease in the operation rate of the welding system 100.

[0059] The shape detection control unit 35, which is an example of an acquisition unit, acquires input data (e.g., point cloud data capable of identifying the three-dimensional shape of the weld bead) related to the shape of the weld bead sent from the sensor 4 while the robot control unit 2 operates the welding robot MC1 to which the sensor 4 is attached based on the appearance inspection program in the bead appearance inspection based on an execution command of the bead appearance inspection of the welded part of the workpiece Wk (e.g., workpiece) sent from the upper device 1. When the sensor 4 reaches a position where the sensor 4 can image the weld bead (in other words, the three-dimensional shape of the welded part can be detected) in response to the driving of the manipulator 200 by the robot control unit 2 described above, the shape detection control unit 35 causes the sensor 4 to irradiate, for example, a laser beam to acquire the input data related to the shape of the weld bead (e.g., point cloud data capable of identifying the three-dimensional shape of the weld bead). When the shape detection control unit 35 receives the input data acquired by the sensor 4 (see above), it passes the input data to the data processing unit 36.

[0060] Data processing unit 36, an example of a conversion unit, when it acquires input data (see above) related to the shape of the weld bead from shape detection control unit 35, converts the data into a data format suitable for the first inspection judgment in inspection result judgment unit 37, and also converts the data into a data format suitable for each of the second inspection judgment, ..., Nth inspection judgment in inspection result judgment unit 37. The conversion of the data format may include a correction process for removing unnecessary point cloud data (e.g., noise) included in the input data (i.e., point cloud data) as a so-called preprocessing, and the above-mentioned preprocessing may be omitted for the first inspection judgment. Data processing unit 36 ​​generates image data showing the three-dimensional shape of the weld bead by making the data format suitable for the first inspection judgment and performing, for example, statistical processing on the input shape data. Note that data processing unit 36 ​​may perform edge emphasis correction for emphasizing the peripheral portion of the weld bead in order to emphasize the position and shape of the weld bead, as data for the first inspection judgment. The data processing unit 36 ​​may count the number of times bead visual inspection is performed for each defective weld location, and if the welding inspection result is not good even when the number of bead visual inspections exceeds the number stored in advance in the memory 32, it may determine that it is difficult or impossible to correct the defective weld location by automatic repair welding. In this case, the inspection result determination unit 37 generates an alert screen including the position of the defective weld location and the type of defective weld (e.g., hole, pit, undercut, spatter, protrusion), and sends the generated alert screen to the higher-level device 1 via the communication unit 30. The alert screen sent to the higher-level device 1 is displayed on the monitor MN1. This alert screen may also be displayed on the monitor MN2.

[0061] The inspection result determination unit 37 can perform a total of N (N: an integer of 2 or more) types of bead appearance inspections (for example, each of the above-mentioned first inspection judgment and second inspection judgment). Specifically, the inspection result determination unit 37 has a first inspection judgment unit 371, a second inspection judgment unit 372, ..., an Nth inspection judgment unit 37N. To simplify the explanation of Fig. 2, the explanation will be given assuming N=2, but the same applies when N=an integer of 3 or more.

[0062] The first inspection and judgment unit 371 performs a first inspection and judgment (i.e., a bead appearance inspection based on a comparison between input data on the shape of the weld bead acquired by the sensor 4 and master data of a good workpiece that is predefined for each workpiece) to inspect the shape reliability of the weld bead (for example, whether it follows a straight or curved weld line), bead chipping, and bead position deviation (see FIG. 4). FIG. 4 is a table showing appropriate examples of the first inspection and judgment and the second inspection and judgment for each of a plurality of inspection items. The first inspection and judgment unit 371 performs a comparison (so-called image processing) between the data converted by the data processing unit 36 ​​for the first inspection and judgment (for example, image data generated based on point cloud data) and the master data of a good workpiece. Therefore, as shown in FIG. 4, the first inspection and judgment unit 371 can inspect the shape reliability of the weld bead, bead chipping, and bead position deviation with high accuracy. First inspection and determination unit 371 calculates an inspection score indicating the inspection results of the shape reliability of the weld bead, bead chipping, and bead position deviation, and creates the calculated value of the inspection score as the first inspection result.

[0063] Second inspection and judgment unit 372 to Nth inspection and judgment unit 37N perform second inspection and judgment (i.e., each forms a neural network using k=(N-1) types of artificial intelligence and performs bead appearance inspection based on AI to determine the presence or absence of welding defects on input data related to the shape of the weld bead acquired by sensor 4 or on the input data after the input data has been preprocessed by data processing unit 36) and inspects the weld bead for the presence or absence of holes, pits, undercuts, spatters, and protrusions (see FIG. 4). Holes, pits, undercuts, spatters, and protrusions in the weld bead are merely listed as examples, and the defect types inspected by Nth inspection and judgment unit 37N are not limited to these. When second inspection and judgment unit 372 to Nth inspection and judgment unit 37N each judges that a corresponding type of welding defect has been detected, it identifies the position of the weld bead where the welding defect was detected. Each of the second inspection and judgment unit 372 to the Nth inspection and judgment unit 37N judges the presence or absence of each welding defect using a learning model (AI) obtained in advance by a learning process for each type of welding defect or for each group of types of welding defects. This allows each of the second inspection and judgment unit 372 to the Nth inspection and judgment unit 37N to highly accurately inspect the presence or absence of holes, pits, undercuts, spatters, and protrusions in the weld bead. Note that each of the second inspection and judgment unit 372 to the Nth inspection and judgment unit 37N does not inspect the shape reliability of the weld bead, bead chipping, and bead position deviation performed by the first inspection and judgment unit 371. The second inspection and judgment unit 372 to the Nth inspection and judgment unit 37N calculates the inspection results (in other words, inspection scores indicating the occurrence probability) of holes, pits, undercuts, spatters, and protrusions in the weld bead, and creates the calculated value of the inspection score as the second inspection and judgment result.

[0064] Here, any one of the first inspection and judgment unit 371, ..., Nth inspection and judgment unit 37N has a function as a comprehensive judgment unit. The inspection and judgment unit as a comprehensive judgment unit acquires the inspection results from each of the first inspection and judgment unit 371, ..., Nth inspection and judgment unit 37N, and judges whether each inspection result is pass or not (i.e., whether there is a welding defect in the weld bead or not) based on each acquired inspection result and the threshold value recorded in the judgment threshold storage unit 34. The inspection and judgment unit as a comprehensive judgment unit creates an appearance inspection report including all the inspection results of the first inspection and judgment unit 371, ..., Nth inspection and judgment unit 37N and information on whether each inspection result is pass or not, stores it in the memory 32, and sends it to the higher-level device 1 via the communication unit 30.

[0065] In addition, the inspection judgment unit as a comprehensive judgment unit may judge whether repair welding by the welding robot MC1 is possible (in other words, whether repair welding by the welding robot MC1 or manual repair welding is preferable) based on the inspection results (inspection scores) included in the above-mentioned first inspection result or second inspection result, and may output the judgment result by including it in the above-mentioned visual inspection report.

[0066] 4, the inspection result determination unit 37 performs a comprehensive and highly accurate inspection of the shape reliability of the weld bead, bead chipping, bead position shift, holes, pits, undercuts, spatters, and protrusions by using the first inspection judgment and the second inspection judgment in combination so as to be suitable for the inspection of each type of welding defect. Note that the above description is an example of N=2, but when N=3, the second inspection judgment unit 372 can detect, for example, the presence or absence of holes and pits in the weld bead as the type of welding defect by using AI, and the Nth inspection judgment unit 37N (N=3) can detect, for example, the presence or absence of undercuts, spatters, and protrusions in the weld bead as the type of welding defect by using different AI. In other words, in the second inspection judgment, multiple AIs (learning models) may be prepared so that different AIs can detect each combination of types of welding defects (for example, combinations of (holes and pits), (undercuts, spatters, and protrusions)) that are the inspection items.

[0067] Here, the processing and processing procedure of the bead appearance inspection executed in the inspection result judgment unit 37 will be described with reference to Fig. 5. Fig. 5 is a sequence diagram showing an example of the processing procedure of the first inspection judgment (point cloud comparison) and the second inspection judgment to the Nth inspection judgment (AI judgment) according to the first embodiment. In the example shown in Fig. 5, the first inspection judgment unit 371 has a function as a comprehensive judgment unit. In Fig. 5, N=2 is set for ease of explanation.

[0068] First inspection determination unit 371 acquires a threshold value for the first inspection result (an example of a first determination criterion) and a threshold value for the second inspection result (an example of a second determination criterion) from determination threshold storage unit 34 (St31). Here, the threshold value for the first inspection result is a threshold value for determining whether or not the first inspection result is equal to or greater than the threshold value (i.e., whether or not the first inspection result is pass in the bead appearance inspection). Also, the threshold value for the second inspection result is a threshold value for determining whether or not the second inspection result is equal to or greater than the threshold value (i.e., whether or not the second inspection result is pass in the bead appearance inspection). First inspection determination unit 371 receives input data relating to the shape of the weld bead acquired by sensor 4 (for example, point cloud data capable of identifying the three-dimensional shape of the weld bead) (St32). In addition, the first inspection and judgment unit 371 obtains converted input data from the data processing unit 36 ​​that has been converted for the second inspection and judgment to be performed by the second inspection and judgment unit 372 (St33), and outputs the converted input data to the second inspection and judgment unit 372 (St341).

[0069] First inspection and determination unit 371 executes image processing using the input data, and inspects the shape reliability, bead chipping, and bead misalignment of the weld bead (St351A). First inspection and determination unit 371 calculates an inspection score created for each inspection item (e.g., shape reliability, bead chipping, bead misalignment), and creates the calculated value of this inspection score as the first inspection result (St361A).

[0070] The second inspection and judgment unit 372 executes an appearance inspection to measure the bead appearance using the converted input data acquired from the first inspection and judgment unit 371 (St351B), calculates a probability value of occurrence of defects as an output value of an AI engine (e.g., neural network) for each inspection item included in the appearance inspection (e.g., shape reliability, bead chipping, bead misalignment, etc.), and creates the calculated value of this inspection score as a second inspection result (St361B). The second inspection and judgment unit 372 outputs the created second inspection result to the first inspection and judgment unit 371 (St371).

[0071] The first inspection determination unit 371 determines whether the created first inspection result is equal to or less than the threshold value for the first inspection result, and whether the second inspection result input from the second inspection determination unit 372 is equal to or more than the threshold value for the second inspection result (St381). The first inspection determination unit 371 comprehensively determines whether the first inspection result and the second inspection result are pass or not (St39), creates an appearance inspection report including each inspection result and information on whether each inspection result is pass or not, and stores it in the memory 32. If the first inspection determination unit 371 determines that both the first inspection result and the second inspection result are pass, it transmits the appearance inspection report to the higher-level device 1 via the communication unit 30, and if it determines that either the first inspection result or the second inspection result is fail, it transmits the appearance inspection report to the repair welding program creation unit 38 and to the higher-level device 1 via the communication unit 30.

[0072] The processing procedures of the first inspection judgment (point cloud comparison) and the second inspection judgment to the Nth inspection judgment (AI judgment) shown in Fig. 4 are not limited to the above example. For example, the processing of steps St33 to St341 (input data conversion and output processing) may be executed after the processing of step St361A (calculation of the first inspection result). Also, Fig. 4 shows an example in which the first inspection judgment unit 371 executes a comprehensive judgment of the first inspection result and the second inspection result, but the execution of the comprehensive judgment process of step St39 may be executed by the second inspection judgment unit 372 (that is, it may be executed by any one of the first inspection judgment unit 371, ..., Nth inspection judgment unit 37N).

[0073] Therefore, as shown in FIG. 4, the inspection result judgment unit 37 can comprehensively and highly accurately inspect the shape reliability of the weld bead, bead chipping, bead position deviation, holes, pits, undercuts, spatters, and protrusions by using the first inspection judgment and the second inspection judgment in combination to suit the inspection of each type of welding defect.

[0074] Note that the above description has been given with N=2 as an example, but when N=3, second inspection judgment unit 372 can detect, for example, the presence or absence of holes or pits in the weld bead as types of welding defects using an AI, and Nth inspection judgment unit 37N (N=3) can detect, for example, the presence or absence of undercuts, spatters, and protrusions in the weld bead as types of welding defects using a different AI. In other words, in the second inspection judgment, multiple AIs (learning models) may be prepared arbitrarily so that different AIs can be used to detect each combination of types of welding defects that are inspection items (for example, a combination of (holes and pits), (undercuts, spatters, and protrusions)).

[0075] Repair welding program creation unit 38 creates a repair welding program for workpiece Wk (e.g., workpiece or repair workpiece) to be performed by welding robot MC1, using the appearance inspection report of workpiece Wk (e.g., workpiece or repair workpiece) by inspection result determination unit 37 and workpiece information (e.g., information such as coordinates indicating the position of a detection point of a welding defect in the workpiece or repair workpiece). The repair welding program may include various parameters such as welding current, welding voltage, offset amount, welding speed, and attitude of welding torch 400 for controlling power supply device 500, manipulator 200, wire feeder 300, welding torch 400, etc. during repair welding. The created repair welding program may be stored in processor 31 or in RAM in memory 32.

[0076] The sensor 4 is, for example, a three-dimensional shape sensor, which is attached to the tip of the welding robot MC1 and can acquire a plurality of point cloud data that can specify the shape of the welded part on the workpiece Wk (for example, workpiece), and generates point cloud data that can specify the three-dimensional shape of the welded part based on the point cloud data and sends it to the inspection control device 3. In addition, when the sensor 4 is not attached to the tip of the welding robot MC1 and is arranged separately from the welding robot MC1, it may be composed of a laser light source (not shown) configured to be able to scan the welded part on the workpiece Wk (for example, workpiece or repair workpiece) based on the position information of the welded part sent from the inspection control device 3, and a camera (not shown) arranged to be able to image an imaging area including the periphery of the welded part and to image the reflection trajectory of the laser light reflected from the laser light irradiated to the welded part (i.e., the shape line of the welded part). In this case, the sensor 4 sends the shape data of the welded part based on the laser light imaged by the camera (in other words, image data of the welded bead) to the inspection control device 3. In addition, the above-mentioned camera is configured to have at least a lens (not shown) and an image sensor (not shown). The image sensor is a solid-state imaging element such as a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS), and converts an optical image formed on an imaging surface into an electrical signal.

[0077] (Welding system operation) Next, a series of operational procedures for main welding, bead visual inspection, and repair welding by welding system 100 according to embodiment 1 will be described with reference to Fig. 3. Fig. 3 is a sequence diagram showing an example of a series of processing procedures including main welding, bead visual inspection, and repair welding by welding system 100 according to embodiment 1. In the explanation of Fig. 3, the operational procedures performed between host device 1, robot control device 2, and inspection control device 3 for each process of main welding using multiple original workpieces and repair welding performed when the bead visual inspection of the workpiece is failed (i.e., the overall judgment result indicates that there is a welding defect) will be described by exemplifying them.

[0078] In Fig. 3, the higher-level device 1 acquires work information (e.g., ID, name, and welding location of the original work) of the original work to be subjected to the main welding (St1), and generates an execution command for the main welding including the work information of the original work. The higher-level device 1 sends an execution command for the main welding including the work information of the original work to the robot control device 2 (St2). Note that the robot control device 2 may execute the processes of steps St1 and St2, respectively, without going through the higher-level device 1. In this case, it is preferable that the memory 22 of the robot control device 2 stores the same data as that stored in the external storage ST, or that the robot control device 2 is connected so as to be able to acquire data from the external storage ST.

[0079] When the robot control device 2 receives an execution command for the main welding sent from the upper device 1, the robot control device 2 uses the work information of each of the multiple original workpieces included in the execution command to create a main welding program for the main welding to be performed by the welding robot MC1, and causes the welding robot MC1 to perform the main welding according to the main welding program (St3). When the robot control device 2 determines the completion of the main welding by the welding robot MC1 using various known methods, it generates a main welding completion notice indicating that the main welding has been completed and sends it to the upper device 1 (St4). When the upper device 1 receives the main welding completion notice, it generates an execution command for an appearance inspection program including a work appearance inspection program and sends it to the robot control device 2 (St5), and also generates an execution command for a bead appearance inspection of the workpiece and sends it to the inspection control device 3 (St6). When the bead appearance inspection starts, the robot control device 2 executes the appearance inspection program received from the upper device 1, and moves the sensor 4 attached to the welding robot MC1 along the weld line (St7). The sensor 4 acquires point cloud data capable of identifying the three-dimensional shape of the workpiece while being moved by the robot control device 2 so as to be able to scan the welding spot of the workpiece (St7).

[0080] The inspection control device 3 uses the point cloud data capable of identifying the three-dimensional shape of the weld bead acquired by the sensor 4 as input data, and executes each of the above-mentioned first inspection judgment and second inspection judgment individually (in parallel) (St7). The inspection control device 3 performs an overall judgment of the bead appearance inspection of the weld bead of the workpiece based on the results of each of the individual bead appearance inspections in step St7 (i.e., the first inspection judgment and the second inspection judgment) (St8).

[0081] When the inspection control device 3 determines that the workpiece has a welding defect and therefore needs repair welding (St9) as a result of the overall determination in step St8, the inspection control device 3 obtains the actual welding program from the robot control device 2 and creates a repair welding program by modifying a part of the actual welding program (St9). The modified part is, for example, a content indicating a location (range) where repair welding is performed. Although detailed illustration is omitted in FIG. 3, the inspection control device 3 may request data of the actual welding program from the robot control device 2 in step St9 and obtain data of the actual welding program sent from the robot control device 2 in response to this request, or may obtain data of the actual welding program sent from the robot control device 2 after step St3 in advance. In this way, the inspection control device 3 can efficiently create data of the repair welding program by partially modifying data of the actual welding program obtained from the robot control device 2. The inspection control device 3 generates an appearance inspection report including the result of the overall determination in step St8 and the repair welding program, and sends it to the robot control device 2 (St10). The inspection control device 3 also sends the appearance inspection report generated in the same manner to the upper device 1 (St11).

[0082] Upon receiving the appearance inspection report in step St11, the upper device 1 generates an execution command for repair welding of the workpiece and sends it to the robot control device 2 (St12). When the robot control device 2 receives the execution command for repair welding sent from the upper device 1, it causes the welding robot MC1 to execute repair welding according to the repair welding program (received in step St10) for the workpiece specified in the execution command (St13). When the robot control device 2 determines the completion of repair welding by the welding robot MC1 by various known methods, it sends work information of the repair work (for example, the ID of the repair work, work information including the IDs of each of the multiple original workpieces used in the main welding (for example, the ID, name, and welding location of the original workpiece), and the welding conditions at the time of each of the main welding and repair welding) to the upper device 1 (St14).

[0083] When the upper device 1 receives the work information including the ID of the repair work sent from the robot control device 2, it sets a management ID suitable for the user corresponding to the ID of the repair work, and stores data indicating that the welding of the repair work corresponding to this management ID has been completed in the external storage ST (St15).

[0084] (Embodiment 2) In the first embodiment, both the first inspection judgment and the second inspection judgment are executed in the inspection control device 3. In the second embodiment, an example in which the first inspection judgment and the second inspection judgment are executed in different devices will be described. In the following, it will be described that the first inspection judgment is executed in the inspection control device 3, and the second inspection judgment is executed in the higher-level device 1. However, the second inspection judgment may be executed in a device other than the higher-level device 1.

[0085] (Welding system configuration) Fig. 6 is a diagram showing an example of the internal configuration of an inspection control device 3A, a robot control device 2, and a higher-level device 1A according to the second embodiment. In the description of Fig. 6, the same reference numerals are given to the same components as those in Fig. 2, and the description is simplified or omitted, and only the differences are described. The configuration of a welding system 100A according to the second embodiment is the same as that of welding system 100 according to the first embodiment (see Fig. 1).

[0086] A welding system 100A as an example of a bead appearance inspection system includes an upper device 1A connected to an external storage ST, an input interface UI1, and a monitor MN1, a robot control device 2, an inspection control device 3A, a sensor 4, a main welding robot MC1a, and a repair welding robot MC1b.

[0087] In the inspection control device 3A as an example of a bead appearance inspection device, a processor 31A includes a judgment threshold storage unit 34, a shape detection control unit 35, a data processing unit 36, an inspection result judgment unit 37A, and a repair welding program creation unit 38. The inspection result judgment unit 37A only has a first inspection judgment unit 371. The configuration of the first inspection judgment unit 371 is the same as that of the first embodiment, and therefore a description thereof will be omitted.

[0088] In upper device 1A as an example of a bead appearance inspection device, processor 11A includes cell control unit 13 and second inspection and judgment unit 142 to Nth inspection and judgment unit 14N. Like second inspection and judgment unit 372 to Nth inspection and judgment unit 37N, second inspection and judgment unit 142 to Nth inspection and judgment unit 14N perform second inspection and judgment (i.e., bead appearance inspection that forms a neural network using k=(N-1) types of artificial intelligence and determines the presence or absence of defective welding points based on AI targeting input data related to the shape of the weld bead acquired by sensor 4) and inspects the presence or absence of holes, pits, undercuts, spatters, and protrusions in the weld bead (see FIG. 4).

[0089] (Welding system operation) Next, a series of processing procedures including main welding, bead appearance inspection, and repair welding by welding system 100A according to embodiment 2 will be described with reference to FIG. 7. FIG. 7 is a sequence diagram showing an example of a series of processing procedures including main welding, bead appearance inspection, and repair welding by welding system 100A according to embodiment 2. In the explanation of FIG. 7, the operation procedures performed between upper device 1A, robot control device 2, and inspection control device 3A for each process of main welding using multiple original workpieces and repair welding performed based on the bead appearance inspection of a workpiece being rejected will be exemplified and explained. In the explanation of FIG. 7, the same step numbers are assigned to processes that overlap with the processes of FIG. 3, and explanations are simplified or omitted, and different contents will be explained.

[0090] In Fig. 7, after step St6, the robot control device 2 executes the appearance inspection program received from the higher-level device 1A with the start of the bead appearance inspection, and moves the sensor 4 attached to the welding robot MC1 along the weld line (St7A). While the sensor 4 is moved by the robot control device 2 so as to be able to scan the welding portion of the work, it acquires point cloud data capable of identifying the three-dimensional shape of the work (St7A). The inspection control device 3A uses the point cloud data capable of identifying the three-dimensional shape of the weld bead acquired by the sensor 4 as input data, and executes the first inspection judgment described above (St7A). Also, the inspection control device 3A generates an execution command for the second inspection judgment described above in the processor 31A and sends it to the higher-level device 1A (St21).

[0091] When the higher-level device 1A receives the execution command for the second inspection judgment sent from the inspection control device 3A in step St21, the second inspection judgment is executed in each of the second inspection judgment unit 142 to the Nth inspection judgment unit 14N based on the execution command (St22). The details of the second inspection judgment executed in step St22 are the same as those described in the first embodiment, so the description will be omitted. The higher-level device 1A generates a processing result of the second inspection judgment (i.e., detection of the presence or absence of welding defects for each inspection item by AI processing) and sends it to the inspection control device 3A (St23). The inspection control device 3A performs a comprehensive judgment of the bead appearance inspection of the workpiece based on the results of the first inspection judgment by the inspection control device 3A in step St7A and the second inspection judgment by the higher-level device 1A in step St22 (St8A). The details of the comprehensive judgment executed in step St8A are the same as those described in the first embodiment, so the description will be omitted. The processing after step St8A is the same as that in FIG. 3, so the description will be omitted.

[0092] As described above, welding system 100A as an example of a bead appearance inspection system according to the second embodiment inputs input data (e.g., point cloud data OD1) related to a weld bead of a workpiece produced by welding in inspection control device 3A. Welding system 100A uses the input data and master data MD1 of a non-defective workpiece, performs a first inspection judgment of the weld bead based on a comparison between the input data and master data MD1 in inspection control device 3A, and is equipped with k (k: an integer of 1 or more) types of artificial intelligence, and performs a second inspection judgment of the weld bead in each of second inspection judgment unit 142 to Nth inspection judgment unit 14N of host device 1A based on processing of the k types of artificial intelligence targeting the input data. Inspection control device 3A generates a result of a bead appearance inspection of the weld bead by inspection result judgment unit 37A based on the judgment results of first inspection judgment unit 371 of inspection control device 3A and second inspection judgment unit 142 to Nth inspection judgment unit 14N of host device 1A, and outputs the result to an output device (e.g., monitor MN2).

[0093] The processing procedure of the bead appearance inspection executed in the inspection result determination unit 37A in the inspection control device 3A and in the processor 11A in the higher-level device 1A will be described with reference to Fig. 8. Fig. 8 is a sequence diagram showing an example of the processing procedure of the first inspection judgment (point cloud comparison) and the second inspection judgment to the Nth inspection judgment (AI judgment) according to the second embodiment. In the explanation of Fig. 8, the higher-level device 1A is equipped with k types of artificial intelligence, and the processing procedure of the bead appearance inspection performed between the inspection control device 3A and the higher-level device 1A will be described. Note that in the explanation of Fig. 8, the same step numbers are given to the processes that overlap with the processes in Fig. 5, and the explanation is simplified or omitted, and the different contents will be explained.

[0094] First inspection and judgment unit 371 acquires a threshold value for judging whether or not the first inspection result and the second inspection result are pass (i.e., whether or not the bead appearance inspection is pass) from judgment threshold storage unit 34 (St31). First inspection and judgment unit 371 receives input data related to the shape of the weld bead acquired by sensor 4 (for example, point cloud data capable of identifying the three-dimensional shape of the weld bead) (St32). First inspection and judgment unit 371 also acquires converted input data converted for the second inspection and judgment executed by second inspection and judgment unit 142 from data processing unit 36 ​​(St33), and sends the converted input data to second inspection and judgment unit 142 in higher-level device 1A (St342).

[0095] First inspection and determination unit 371 executes image processing using the input data, and inspects the shape reliability, bead chipping, and bead misalignment of the weld bead (St351A). First inspection and determination unit 371 calculates an inspection score created for each inspection item (e.g., shape reliability, bead chipping, bead misalignment), and creates the calculated value of this inspection score as the first inspection result (St361A).

[0096] The second inspection and judgment unit 142 executes an appearance inspection to measure the bead appearance using the converted input data acquired from the first inspection and judgment unit 371 (St352B), calculates a probability value of occurrence of defects as an output value of an AI engine (e.g., neural network) for each inspection item included in the appearance inspection (e.g., shape reliability, bead chipping, bead misalignment, etc.), and creates the calculated value of the inspection score as the second inspection result (St362B). The second inspection and judgment unit 142 sends the created second inspection result to the first inspection and judgment unit 371 in the inspection control device 3A (St372).

[0097] The first inspection determination unit 371 determines whether the created first inspection result is equal to or less than the threshold value for the first inspection result, and whether the second inspection result input from the second inspection determination unit 142 is equal to or more than the threshold value for the second inspection result (St382). The first inspection determination unit 371 comprehensively determines whether the first inspection result and the second inspection result are pass or not (St39), creates an appearance inspection report including each inspection result and information on whether each inspection result is pass or not, and stores it in the memory 32. If the first inspection determination unit 371 determines that both the first inspection result and the second inspection result are pass, it transmits the appearance inspection report to the higher-level device 1A via the communication unit 30, and if it determines that either the first inspection result or the second inspection result is fail, it transmits the appearance inspection report to the repair welding program creation unit 38 and to the higher-level device 1A via the communication unit 30.

[0098] As described above, welding system 100 according to the first embodiment and welding system 100A according to the second embodiment include shape detection control unit 35 that acquires input data (e.g., point cloud data) related to the weld bead, memory 32 that stores a threshold value for a first inspection result (an example of a first judgment criterion) and a threshold value for a second inspection result (an example of a second judgment criterion) that are set by a user operation and are used to inspect the weld bead for defects, first inspection and judgment unit 371 (an example of a first judgment unit) that performs a first inspection and judgment on the weld bead, and k second inspection and judgment units 372 to Nth inspection and judgment units 37N or k second inspection and judgment units 142 to Nth inspection and judgment units 14N (an example of a second judgment unit) that perform a second inspection and judgment on the weld bead. The k second inspection and judgment units output the second inspection results acquired by the second inspection and judgment to first inspection and judgment unit 371. First inspection judgment unit 371 creates and outputs an appearance inspection result of the weld bead using a judgment result of whether or not the first inspection result acquired by the first inspection judgment satisfies a threshold value for the first inspection result (an example of a first judgment criterion) and a judgment result of whether or not the input second inspection result satisfies a threshold value for the second inspection result (an example of a second judgment criterion), where k=(N-1), and so on.

[0099] This allows the inspection control device 3, 3A to perform the first inspection and judgment on the weld bead in combination with the second inspection and judgment, thereby enabling more efficient visual inspection of the weld bead of the workpiece produced by main welding. Furthermore, the inspection control device 3, 3A stores information on threshold values ​​used in N types of bead visual inspection (i.e., the first inspection and judgment unit 371 to the Nth inspection and judgment unit 37N, or the first inspection and judgment unit 371 to the Nth inspection and judgment unit 14N) in the first inspection and judgment unit 371, so that the threshold value only needs to be set by the user's operation once, thereby making the threshold value setting work more efficient. Therefore, the inspection control device 3, 3A can improve the convenience of the visual inspection of the weld bead for the user.

[0100] Furthermore, first inspection and judgment unit 371 according to the first embodiment performs a first inspection and judgment on the weld bead based on a comparison between the input data and master data MD1. Furthermore, second inspection and judgment unit 372 to Nth inspection and judgment unit 37N according to the first embodiment are equipped with k types of artificial intelligence and perform a second inspection and judgment on the weld bead based on processing by the k types of artificial intelligence using the input data. As a result, inspection control device 3, 3A according to the first embodiment can perform a first inspection and judgment based on a comparison between input data indicating the three-dimensional shape of the weld bead and master data MD1 and a second inspection and judgment that detects the presence or absence of welding defects in the weld bead based on AI processing in combination, so that the appearance inspection of the weld bead of the workpiece produced by main welding can be performed more efficiently.

[0101] The inspection control device 3, 3A further includes a data processing unit 36 ​​that converts the input data into an input data format that allows the k second inspection and judgment units to perform the second inspection and judgment. The k second inspection and judgment units perform the second inspection and judgment using the input data converted by the data processing unit 36. As a result, the inspection control device 3, 3A can improve the accuracy of the AI ​​processing performed by each of the second inspection and judgment units 372 to 37N by converting one piece of input data into a data format required by each of the k second inspection and judgment units, and can improve the accuracy of detecting the presence or absence of defects in the weld beads (for example, holes, pits, undercuts, spatters). Therefore, the inspection control device 3, 3A can comprehensively inspect the appearance inspection items of the weld beads that are detected with high accuracy by the first inspection and judgment and the appearance inspection items of the weld beads that are detected with high accuracy by the second inspection and judgment.

[0102] Furthermore, the weld bead appearance inspection items that are the subject of the first inspection judgment are the weld bead shape, chipping of the weld bead, and displacement of the weld bead. The weld bead appearance inspection items that are the subject of the second inspection judgment are holes, pits, undercuts, spatters, and protrusions in the weld bead. This allows the inspection control device 3, 3A to comprehensively inspect the weld bead appearance inspection items that are detected with high accuracy by the first inspection judgment (e.g., weld bead shape, chipping of the weld bead, and displacement of the weld bead) and the weld bead appearance inspection items that are detected with high accuracy by the second inspection judgment (e.g., holes, pits, undercuts, spatters, and protrusions in the weld bead).

[0103] Although various embodiments have been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It is clear that a person skilled in the art can conceive of various modifications, corrections, substitutions, additions, deletions, and equivalents within the scope of the claims, and it is understood that these also naturally belong to the technical scope of the present disclosure. In addition, the components in the various embodiments described above may be arbitrarily combined within the scope of the invention.

[0104] This application is based on a Japanese patent application (Patent Application No. 2020-038207) filed on March 5, 2020, the contents of which are incorporated by reference into this application. [Industrial Applicability]

[0105] The present disclosure is useful as a bead appearance inspection device and a bead appearance inspection system that make it more efficient for a user to perform setting operations regarding bead appearance inspection of workpieces produced by main welding. [Explanation of symbols]

[0106] 1,1A Upper device 2 Robot control device 3,3A Inspection control device 4 Sensors 10,20,30 Communications Department 11,11A,21,31,31A Processors 12,22,32 memory 13 Cell control section 23 Welding Program Creation Department 24 Arithmetic section 25 Robot control unit 26 Power supply control unit 33 Test result storage unit 34 Judgment threshold memory unit 35 Shape detection control section 36 Data Processing Section 37, 37A Inspection result judgment section 371 First Inspection and Judgment Department 142,372 2nd Inspection and Judgment Department 14N, 37N Nth Inspection and Judgment Department 100,100A welding system 200 Manipulator 300 Wire feeder 301 Welding Wire 400 Welding Torch 500 power supply MC1 Welding Robot MC1a Main welding robot MC1b Repair welding robot MN1,MN2 monitor ST External Storage

Claims

1. An acquisition unit that acquires input data related to a weld bead; a first determination unit that acquires a first determination criterion and a second determination criterion set by a user for use in inspecting the weld bead for defects, and executes a first inspection determination on the weld bead; and k (k: integer equal to or greater than 1) second determination units that perform a second inspection and determination on the weld bead, The k second determination units output second inspection results acquired by the second inspection determination to the first determination unit; the first determination unit creates and outputs an appearance inspection result of the weld bead using a determination result of whether or not the first inspection result acquired by the first inspection determination satisfies the first determination criterion and a determination result of whether or not the input second inspection result satisfies the second determination criterion. Bead appearance inspection device.

2. The first determination unit performs the first inspection determination on the weld bead based on a comparison between the input data and master data, The second determination unit is equipped with k types of artificial intelligence and performs the second inspection determination on the weld bead based on processing by the k types of artificial intelligence using the input data.

2. The bead appearance inspection device according to claim 1.

3. a conversion unit that converts the input data into a format of input data that enables the second inspection judgment by the k second judgment units to be executed, The k second determination units perform the second inspection determination using the converted input data converted by the conversion unit.

2. The bead appearance inspection device according to claim 1.

4. The appearance inspection items related to the shape of the weld bead that are the subject of the first inspection and judgment are the shape of the weld bead, a chipping of the weld bead, and a positional deviation of the weld bead, The appearance inspection items related to defects of the weld bead that are the subject of the second inspection and judgment are holes, pits, undercuts, spatters, and protrusions of the weld bead.

2. The bead appearance inspection device according to claim 1.

5. An acquisition unit that acquires input data related to a weld bead; a first determination unit that acquires a first determination criterion and a second determination criterion set by a user for use in inspecting the weld bead for defects, and executes a first inspection determination on the weld bead; and k (k: integer equal to or greater than 1) second determination units that perform a second inspection and determination on the weld bead, The k second determination units output second inspection results acquired by the second inspection determination to the first determination unit; the first determination unit creates and outputs an appearance inspection result of the weld bead using a determination result of whether or not the first inspection result acquired by the first inspection determination satisfies the first determination criterion and a determination result of whether or not the input second inspection result satisfies the second determination criterion. Bead appearance inspection system.

Images