Processing device, joining system, server device, program, and generation method

Abstract

To improve convenience regarding joining.SOLUTION: A processing device 50 comprises: a joining image acquisition unit 52E that acquires a joining image of a joining location of workpieces W to be joined by a joining system 10; and a quality identification unit 52F that identifies joining quality of the joining location shown in the joining image on the basis of the acquired joining image. The processing device 50 further includes a joining condition correction unit 52G that corrects joining conditions when the identified joining quality is not good, thereby increasing joining current and joining voltage for a new join. The quality identification unit 52F inputs the joining image into a machine-trained quality determination model 103 and identifies the joining quality from a joining quality determination result output by the quality determination model 103.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a processing device, a joining system, a server device, a program, and a generation method used for various joining processes. [Background technology]

[0002] As disclosed in Patent Document 1, a technique is known in which joining such as resistance welding is performed according to joining conditions (welding conditions) including joining current (welding current) and the like. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6810820 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional joining techniques, for example, still require a lot of effort from the user, and the convenience of joining is low.

[0005] An object of the present invention is to improve convenience regarding joining. [Means for solving the problem]

[0006] In order to solve the above problem, the processing device of the present invention includes a joining image acquisition unit that acquires a joining image of the joining point of the workpieces to be joined by the joining system, a quality identification unit that identifies the joining quality of the joining point shown in the joining image based on the acquired joining image, and a joining condition correction unit that corrects the joining conditions when the identified joining quality is not good.

[0007] The processing device according to the present invention includes a joining condition correction unit that corrects the joining conditions for joining using a joining tool of a joining system, and a notification processing unit that performs notification processing to prompt a user to dress or replace the joining tool when the resistance value of the joining tool drops to a predetermined value.

[0008] The processing device according to the present invention includes a joining condition correction unit that corrects the joining conditions for joining using a joining tool of a joining system, and an order processing unit that processes an order for a new joining tool to replace the joining tool when the resistance value of the joining tool drops to a predetermined value.

[0009] The processing device according to the present invention is equipped with a trial shot control unit that sets the joining conditions for the trial shot of a new joining tool of the joining system to joining conditions in which the power consumption per trial shot is greater than the power consumption per joining shot after the trial shot.

[0010] The processing device according to the present invention comprises a tool image acquisition unit that acquires a tool image of a joining tool used for joining in a joining system, and a throw-away shot control unit that determines whether the throw-away shot can be completed based on the acquired tool image, and performs processing to terminate the throw-away shot when it determines that the shot can be completed.

[0011] A joining system according to the present invention includes the processing device described above and a joining tool used for joining the workpieces.

[0012] The server device of the present invention includes a collection unit that collects data related to connections obtained at the time of connection from each of a plurality of connection systems used by a plurality of companies, and a machine learning unit that performs machine learning to generate a model to be used in the connection system based on the data collected by the collection unit.

[0013] The program according to the present invention causes a computer to function as the processing device or the server device.

[0014] The generation method of the present invention includes a collection step of collecting data related to connections obtained at the time of connection from each of a plurality of connection systems used by a plurality of companies, and a machine learning step of performing machine learning to generate a model to be used in the connection system based on the data collected in the collection step. [Effects of the Invention]

[0015] According to the present invention, the convenience of joining is improved. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram showing a connection environment of a joint system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a schematic diagram of the joining system. [Figure 3] FIG. 3 is a block diagram of the bonding system. [Figure 4] FIG. 4 is a graph showing the change in the combined resistance value of the welding tool and the workpiece when welding is performed many times. [Figure 5] FIG. 5 is a diagram showing an example of the configuration of a trial joining condition database. [Figure 6] FIG. 6 is a flowchart of the trial shot process. [Figure 7] FIG. 7 is a flowchart of the joining process. [Figure 8] FIG. 8 is a graph showing the distribution of junction current when multiple junctions are made. [Figure 9] FIG. 9 is a graph showing the change in junction current when bonding is performed multiple times. [Figure 10] FIG. 10 is a diagram showing an example of the configuration of an integrated power consumption database according to a modified example. [Figure 11] FIG. 11 is a diagram illustrating the configuration of the processing server. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0018] As shown in Fig. 1, the connection systems 10A to 10C according to this embodiment are communicatively connected to a processing server 100 and a sales server via a communication network such as the Internet. The connection systems 10A to 10C are used by users A to C at different locations (for example, different companies), respectively. The number of connection systems and users is arbitrary. The connection systems 10A to 10C are also collectively referred to as connection system 10.

[0019] The processing server 100 and the sales server 200 are server devices, more specifically server computers, and are operated directly or indirectly by the manufacturer X of the junction system 10. Indirect operation includes operation by a company commissioned by the manufacturer X. The processing server 100 provides the junction system 10 with various discrimination functions using a discrimination model described below, and various information registered in a database described below. The sales server 200 delivers a junction tool 21 described below to a user of the junction system 10 in response to an order from the junction system 10, and performs payment processing for the payment.

[0020] The configuration of the joining system 10 is shown in Figure 2. The joining system 10 is installed on a predetermined production line and continuously joins multiple workpieces W. One workpiece W may be joined once or multiple times. The workpiece W includes two members W1 and W2 to be joined by welding.

[0021] The bonding system 10 includes a pair of electrodes 21A and 21B, a drive mechanism 22, a bonding power supply 30, a camera 41, a camera 42, an ammeter 43, a voltmeter 44, and a processing device 50. The pair of electrodes 21A and 21B constitute a bonding tool 21. The bonding tool 21 and the drive mechanism 22 constitute a bonding device 20.

[0022] The joining device 20 is configured to join workpieces W. A pair of electrodes 21A and 21B constituting the joining tool 21 heats the workpieces W by passing an electric current through the workpieces W to generate heat. In other words, the joining tool 21 is configured as a heating tool that heats the workpieces W. The joining tool 21 deteriorates with increased use in joining, so it is mounted in the joining device 20 so that it can be replaced with a new one. Furthermore, a user can use any of multiple types of joining tools 21 with the joining device 20 depending on the material of the workpieces W, etc. The drive mechanism 22 moves one of the electrodes 21A and 21B relative to the other, thereby sandwiching and pressurizing the workpieces W between the electrodes 21A and 21B during joining. By heating the workpieces W while applying pressure, the two members W1 and W2 of the workpieces W are joined by resistance welding. The drive mechanism 22 also includes a mechanism for moving the workpiece W. For example, the drive mechanism 22 is configured to move the workpiece W in the preceding stage to a joining position by the joining device 20, and to transport the joined workpiece W to the succeeding stage. Furthermore, when there are multiple joining locations on the workpiece W, the drive mechanism 22 may be configured to sequentially shift the position of the workpiece W each time a joining is performed.

[0023] The bonding power supply 30 is a circuit that applies a current to the workpiece W via the electrodes 21A and 21B. As the bonding power supply 30, for example, an inverter type power supply including a switching element, a transformer, a rectifier circuit, etc. is adopted.

[0024] The camera 41 is disposed at a position where it captures an image of the welding tool 21 (for example, the electrode 21A). The camera 42 is disposed at a position where it captures an image of the welding point of the workpiece W. The ammeter 43 and the voltmeter 44 detect a welding current, which is a current flowing through the workpiece W, and a welding voltage, which is a voltage applied to both ends of the welding tool 21 (the electrodes 21A and 21B), respectively.

[0025] The processing device 50 is configured to control the overall operation of the bonding system 10. The processing device 50 includes a non-volatile storage device 51 that stores various data, programs, etc., including bonding conditions described below, and a processor 52 that executes the programs and performs the processing described below using the various data. The processor 52 includes a CPU (Central Processing Unit) and the like. The processing device 50 further includes a main memory 53 that stores data used by the processor 52, and an I / O (Input / Output) 54 that relays data exchanged between the processor 52 and the outside of the processing device 50. The I / O (Input / Output) 54 may include an analog-to-digital conversion circuit that converts analog signals input to the processing device 50 (e.g., signals indicating current and voltage detected by the ammeter 43 or voltmeter 44) into digital signals and supplies the digital signals to the processor 52. The processing device 50 also includes an interface 55. The interface 55 includes a communication interface, such as a network card connected to a communication network N, and a user interface, such as a display device and an operating device.

[0026] By executing the programs stored in the memory device 51, the processor 52 operates as a joining execution unit 52A, a joining monitoring unit 52B, a trial shot control unit 52C, a tool image acquisition unit 52D, a joining image acquisition unit 52E, a quality identification unit 52F, a joining condition correction unit 52G, a notification processing unit 52H, and an order processing unit 52I, as shown in Figure 3.

[0027] The joining execution unit 52A controls the joining device 20 and the joining power supply 30 according to the joining conditions prepared in the storage device 51 or the joining conditions for trial welding described below, and performs joining of the workpieces W. Specifically, the joining execution unit 52A moves the workpieces W to a predetermined position using the drive mechanism 22, and applies pressure by sandwiching the portions of the workpieces W to be joined between the electrodes 21A and 21B of the joining tool 21. Thereafter, the joining execution unit 52A controls the joining power supply 30 to join the workpieces W (here, resistance joining) by passing a current through the joining tool 21. The joining conditions include, for example, a joining voltage (in the case of voltage control described below) or a joining current (in the case of current control described below) for passing a current through the workpieces W, a current-flow time for passing the current through the workpieces W, and a pressure applied to the workpieces W. The joining conditions may be automatically set according to the type of the joining tool 21, or may be arbitrarily set by the user via the interface 55.

[0028] The bonding execution unit 52A controls the bonding power supply 30 so as to apply a pulse voltage having a value of the bonding voltage included in the bonding conditions to both ends of the bonding tool 21 sandwiching the workpiece W. For example, the bonding execution unit 52A performs feedback control (voltage control) using the voltage value detected by the voltmeter 44 as a feedback value and the bonding voltage value included in the bonding conditions as a target voltage. The bonding execution unit 52A may also control the bonding power supply 30 so as to pass a pulse current having a value of the bonding current included in the bonding conditions through the workpiece W. In this case, the bonding execution unit 52A performs feedback control (current control) using the current value detected by the ammeter 43 as a feedback value and the bonding current value included in the bonding conditions as a target current. The following describes a case where voltage control is adopted, but the voltage control can be changed to current control. When changing, the current is appropriately replaced with the voltage, and the voltage is appropriately replaced with the current.

[0029] The joining execution unit 52A periodically executes joining in sequence in accordance with the joining conditions. Between joining operations, the joining execution unit 52A may move the workpiece W by the drive mechanism 22. The movement of the workpiece W before and after joining may also be performed manually.

[0030] The bonding monitoring unit 52B monitors the bonding, derives monitor values ​​indicating the state of the bonding, and stores the monitor values ​​in the storage device 51 or the main memory 53. The monitor values ​​include, for example, the bonding current (e.g., its peak value) detected by the ammeter 43. The monitor values ​​may also include, for example, the bonding voltage (e.g., its peak value) detected by the voltmeter 44. The monitor values ​​may also include statistical values ​​of multiple current values ​​and / or multiple voltage values ​​periodically detected by the ammeter 43 and / or the voltmeter 44 during bonding. Furthermore, the monitor values ​​may also include the power consumption (bonding power consumption) of the combination of the bonding tool 21 and the workpiece W during bonding, and / or the combined resistance value of the combination of the bonding tool 21 and the workpiece W, which are derived based on the bonding current and the bonding voltage. The monitor values ​​are sequentially stored in the storage device 51 or the main memory 53 in chronological order for one or more bonding operations.

[0031] The welding of the workpieces W performed by the welding execution unit 52A includes so-called trial welding. The trial welding is performed when the welding tool 21 is new, that is, when the welding system 10 is newly installed or when the welding tool 21 is replaced with a new one. As shown in FIG. 4, the resistance value of a new welding tool 21 and the welding strength of the welding performed by this welding tool 21 are insufficient at the beginning of use. FIG. 4 shows the changes in the resistance value and the welding strength when welding is performed sequentially under constant welding conditions. As shown in FIG. 4, the resistance value and the welding strength increase and stabilize as the number of welding operations increases (the welding tool 21 becomes accustomed). Therefore, multiple welding operations (e.g., 50 times) are performed on a trial workpiece Wd until the resistance value and the welding strength increase and stabilize at a relatively high value. This welding is called a "trial welding." Here, the workpiece Wd is assumed to be of a type (material and dimensions) determined for each type of welding tool 21, but it may be the same as the workpiece W after trial welding.

[0032] The trial shots can be terminated when the integrated value of the power consumption of the welding tool 21 in each trial shot performed sequentially reaches a predetermined value. This is because the resistance value and welding strength increase and stabilize according to the integrated value of the power consumption. The predetermined value differs depending on the type of the welding tool 21 (particularly, the material and shape). In this embodiment, the welding conditions are set so that the integrated value reaches the predetermined value with a small number of welding runs. Specifically, the welding voltage (target value for voltage control) of the welding conditions is set so that the power consumption of the welding tool 21 per trial shot is greater than the power consumption of the welding tool 21 per welding run after the trial shots are completed. In the following explanation, the power consumption of the welding tool 21 is considered to be the welding power consumption, which is the power consumption of the combination of the welding tool 21 and the workpiece W or Wd.

[0033] The test shot is performed by the joining execution unit 52A, the test shot control unit 52C, and the tool image acquisition unit 52D in cooperation with each other. In the test shot, the test shot control unit 52C communicates with the processing server 100 and uses the test shot joining condition database 101 and the test shot discrimination model 102 provided in the processing server 100.

[0034] As shown in FIG. 5 , the trial joining condition database 101 stores multiple records, each of which includes tool identification information for identifying the type of the joining tool 21 and trial joining conditions for the joining tool 21 associated with the tool identification information. The tool identification information includes, for example, the manufacturer and model number of the joining tool 21. When the type of workpiece Wd varies depending on the user, i.e., the joining system 10, the tool identification information may also include the type of workpiece Wd (e.g., material and shape (various dimensions)). This is because the resistance value of the workpiece Wd varies depending on its specifications, and therefore, even if the joining voltage is the same, the voltage applied to the joining tool 21 may differ. The tool identification information of a joining tool 21 whose manufacturer and model number cannot be identified may also include other information for identifying the type of the joining tool 21, such as material and shape (various dimensions). As can be seen from the joining condition database 101, trial joining conditions are prepared for each type of joining tool 21. The joining conditions are derived and registered, for example, by simulation, experiment, or machine learning. The trial shot conditions are determined so that the power consumption of the welding tool 21 per trial shot is greater than the power consumption of the welding tool 21 per welding shot after the trial shot is completed, so that the integrated value of the power consumption, which is accumulated each time a trial shot is performed, quickly reaches a predetermined value required for completing the trial shot. The welding conditions include, for example, a bonding voltage (target voltage in voltage control) higher than the bonding voltage included in the joining conditions after the trial shot stored in the storage device 51. Additionally or alternatively, the welding conditions may include, for example, a longer current application time compared to the joining conditions after the trial shot.

[0035] The re-fire determination model 102 in FIG. 3 is generated by machine learning as a model for determining when re-fire firing should be completed. In this embodiment, multiple re-fire determination models 102 are generated for each type of joining tool 21, i.e., for each piece of tool identification information. As re-fire firing is performed, the appearance (e.g., color) of the joining tool 21 changes, and when the re-fire firing can be completed, the appearance of the joining tool 21 takes on a specific appearance. In other words, the appearance of the joining tool 21 can determine whether re-fire firing can be completed. The re-fire determination model 102 is an AI model that realizes this appearance-based determination. It receives a tool image of the joining tool 21 as input and outputs a determination result as to whether re-fire firing can be completed. An example of an AI model is a neural network. Machine learning may be any learning method, including deep learning. For example, the re-fire determination model 102 is generated by learning with a correct answer based on training data including a captured image of the joining tool 21 and correct answer information indicating that the joining tool 21 is the joining tool 21 before or after the re-fire firing has been completed. The training data is collected in advance by any method.

[0036] The welding execution unit 52A, the trial shot control unit 52C, and the tool image acquisition unit 52D perform the trial shot processing of Fig. 6. The trial shot processing is started, for example, when the user inputs a trial shot start instruction to start the trial shot into the processing device 50 by operating the interface 55. The trial shot start instruction includes tool identification information that identifies the type of the welding tool 21.

[0037] 6, first, the throw-away shot control unit 52C communicates with the processing server 100 and acquires the throw-away shot joining conditions corresponding to the type of joining tool 21 from the joining condition database 101 (step S11). The throw-away shot control unit 52C accesses the processing server 100 using the tool identification information included in the throw-away shot start instruction as a key, and acquires the throw-away shot joining conditions associated with this tool identification information from the joining condition database 101.

[0038] The throw-away shot control unit 52C sets the acquired joining conditions for the throw-away shot in the main memory 53 as joining conditions (hereinafter also referred to as throw-away shot conditions) that the joining execution unit 52A will use for the throw-away shot from now on (step S12).

[0039] Thereafter, the welding execution unit 52A controls the welding device 20 and the welding power source 30 in accordance with the dummy welding conditions set in step S12, thereby performing dummy welding on the workpiece Wd (step S13).

[0040] Thereafter, the tool image acquiring unit 52D controls the camera 41 to capture an image of the welding tool 21, and acquires a tool image in which the welding tool 21 is captured (step S14).

[0041] Thereafter, the rerun shot control unit 52C determines whether the rerun shot can be ended based on the tool image acquired in step S14 (step S15). The rerun shot control unit 52C communicates with the processing server 100 via the interface 55 and sends the processing server 100 a request to determine whether the rerun shot should be ended. This determination request includes the tool identification information included in the rerun shot start instruction and the tool image acquired in step S14. Upon receiving the determination request, the processing server 100 inputs the tool image included in the determination request into the rerun shot determination model 102 corresponding to the tool identification information included in the determination request, and causes the determination model 102 to output the determination result, which is the output of the determination result, as a response to the determination request. The rerun shot control unit 52C determines whether the rerun shot can be ended based on the returned determination result.

[0042] When the trial shot control unit 52C determines that the trial shot can be ended based on the above-mentioned determination result (step S15; Yes), it outputs to the interface 55 the fact that the trial shot can be ended, and notifies the user that the trial shot can be ended (step S16). Upon receiving this notification, the user inputs the start of joining of the workpieces W (also referred to as main joining) via the interface 55.

[0043] When the discard shot control unit 52C determines that the discard shot cannot be ended based on the above-mentioned determination result (step S15; No), it performs the process from step S13 again.

[0044] Through the above series of processes, trial shots are sequentially performed until the trial shot end timing is reached. Note that since the process for making a determination using the trial shot determination model takes time, steps S14 to S16 may be performed, for example, when the number of trial shots reaches a predetermined number set by the trial shot conditions. Alternatively, steps S14 to S16 may be performed only for a predetermined number of trial shots, such as every two or three shots. In such a case, the sequential trial shots and the processes of steps S14 to S16 may be performed in parallel. Trial shot conditions may be prepared for each type of welding tool 21 and each type of welding system 10. In such a case, the trial shot control unit 52C acquires the welding conditions for trial shots corresponding to the type of welding tool 21 and the type of welding system 10 (such as the manufacturer and model number) from the welding condition database 101. In such a case, the trial shot control unit 52C accesses the processing server 100 using the tool identification information as well as the system identification information of the joining system 10 (for example, stored in the memory device 51 as default information) as a key, and obtains the trial shot joining conditions corresponding to these keys.

[0045] After the trial shot, the actual joining is performed with the workpiece W as the joining target. This joining is performed in cooperation with the joining execution unit 52A, the joining monitoring unit 52B, the joining image acquisition unit 52E, the quality identification unit 52F, the joining condition correction unit 52G, the notification processing unit 52H, and the order processing unit 52I shown in FIG. 3. In the actual joining, the quality identification unit 52F communicates with the processing server 100 and uses the quality determination model 103 provided in the processing server 100.

[0046] The quality determination model 103 is generated by machine learning as a model for determining when to replace or dress the welding tool 21 after a trial run. In this embodiment, the quality determination model 103 is prepared for each type of welding system 10 (e.g., manufacturer and model number), type of welding tool 21, and type of workpiece W, i.e., for each of the system identification information, the welding tool identification information, and the workpiece identification information described below. As welding is performed, the welding tool 21 deteriorates and its resistance value increases. The degree of this deterioration can be determined based on the welding quality of the welding point. In other words, the quality determination model 103 is an AI model that realizes this welding quality determination, and is an AI model that inputs an image of the welding point and outputs the welding quality as a determination result. An example of an AI model is a neural network. The machine learning may be any learning method, including deep learning. For example, the quality determination model 103 is generated by learning with a correct answer based on training data including an image of the welding point and the welding quality of the welding point. The training data is collected in advance by any method. The joining quality may be limited to "good" and "bad," or may include "good," "bad," and "warning" between these two. The warning indicates that the joining quality is close to being bad, but not defective. When this warning is detected, a defective joining point will likely occur in the near future during continuous joining. The quality determination model 103 is generated by machine learning using training data of good and defective products. The model outputs a combination of both good products and their likelihood and defective products and their likelihood as a joining quality determination result. Depending on the level of each likelihood, the determination result may be treated as a good product, a warning, or a defective product. In particular, a determination result in which the likelihood of a good product is not as low as that of a defective product but is below a certain value may be treated as a warning. The quality determination model 103 may also output a good product, a warning, or a defective product as a determination result depending on the likelihood.

[0047] The joining execution unit 52A, the joining monitoring unit 52B, the joining image acquisition unit 52E, the quality identification unit 52F, the joining condition correction unit 52G, the notification processing unit 52H, and the order processing unit 52I cooperate to perform the joining process shown in Fig. 7. The joining process is started, for example, when a user operates the interface 55 to input an instruction to start main joining, which instructs joining of the workpieces W after the trial cutting is completed, into the processing device 50. This start instruction includes workpiece identification information that identifies the type of workpiece W. The type of workpiece W identified by the workpiece identification information may include the material and dimensions (particularly thickness).

[0048] 7, the joining execution unit 52A reads out the joining conditions stored in the storage device 51 into the main memory 53 and sets the joining conditions as the joining conditions to be used this time (step S32). The joining execution unit 52A then executes joining of the workpieces W by voltage control with the joining voltage of the set joining conditions as the target voltage (step S32).

[0049] Thereafter, the junction monitoring unit 52B stores monitor values ​​including the junction current (e.g., peak value), the junction voltage (e.g., peak value), and the junction power consumption in the storage device 51 or the main memory 53 (step S33). As a result, the monitor values ​​are stored in the storage device 51 or the main memory 53 in chronological order every time a bonding is performed.

[0050] Thereafter, the joining execution unit 52A increments the number of joining times N, which is the total number of joining times after the trial joining, by 1 (step S34), and determines whether the number of joining times N has reached a predetermined number Nth, for example, 100 (step S35). Steps S32 to S34 are repeated until the number of joining times N reaches the predetermined number Nth (step S35; No).

[0051] When the number of bonding operations N reaches a predetermined number Nth (step S35; Yes), the bonding monitoring unit 52B derives a decrease D of the junction current (peak value) per bonding operation (step S36). The decrease D is derived, for example, from the slope of an approximation line L ( FIG. 8 ) of the change in junction current stored in chronological order in the storage device 51 or the main memory 53 by the bonding monitoring unit 52B. FIG. 8 is a graph showing the distribution of the junction current (peak value) detected in multiple bonding operations. In FIG. 8 , the approximation line L extends to nearly 1000 bonding operations, but the approximation line L is derived, for example, based on the junction current up to 100 bonding operations. Since the decrease degree of the junction current remains constant even when the number of bonding operations increases, a fairly accurate decrease degree can be obtained from the initial predetermined number of bonding operations. The approximation line can be derived using any statistical method or the like.

[0052] Thereafter, the joining execution unit 52A starts sequentially joining the workpieces W in accordance with the current joining conditions stored in the storage device 51 (which may be corrected in subsequent processing, as will be described later) (step S37).

[0053] Furthermore, in parallel with the joining of the workpieces W, the joining image acquiring unit 52E performs the process of step S38. When the joining after step S37 has been performed M times (e.g., 10 times), the joining image acquiring unit 52E controls the camera 42 to capture an image of the joining portion of the workpieces W and acquires a joining image showing this joining portion (step S38). Furthermore, the joining monitoring unit 52B stores monitor values ​​including the joining current (e.g., peak value), the joining voltage (e.g., peak value), and the joining power consumption in the storage device 51 or the main memory 53 (step S39).

[0054] As will be described later, step S38 is repeatedly performed. In the second and subsequent steps of step S38, the joined image acquisition unit 52E acquires a joined image when the joining after step S38 has been performed M times (for example, three times).

[0055] The quality identification unit 52F then identifies the joining quality of the joined portion based on the joining image acquired in step S38 (step S40). The quality identification unit 52F communicates with the processing server 100 via the interface 55 and transmits a request for identifying the joining quality of the joined portion to the processing server 100. The request for identification includes system identification information, which is identification information for the joining system 10 pre-set in the joining system 10, the tool identification information and workpiece identification information acquired above, and the joining image acquired in step S37. The system identification information, workpiece identification information, etc. are input at any time. Upon receiving the request for identification, the processing server 100 inputs the joining image included in the request for identification into the quality identification model 103 corresponding to each piece of identification information included in the request for identification, and outputs the joining quality of the joined portion shown in the joining image. The processing server 100 returns the output identification result to the quality identification unit 52F as a response to the request for identification. The quality identification unit 52F identifies the joining quality of the joined portion based on the returned identification result.

[0056] If the bonding quality identified by the quality identifying unit 52F is good (step S41; Yes), the processes from step S38 onwards are performed again. As a result, the processes of steps S38 to S39 are performed every M bonding operations, that is, M, 2M, 3M, etc., from step S37. As a result, the acquisition of bonding images, storage of monitor values, and identification of the bonding quality of the bonding portion are performed periodically.

[0057] M in step S38 is preset to a value such that "the period T1 required for M bonding operations" > "the processing period T2 from acquisition of one bonded image to identification of the bonded quality." The latter processing period T2 is generally longer than the period required for one bonding operation. By capturing an image of the bonded portion every M bonding operations, the inconvenience of the processing of the bonded image not keeping up with the bonding is eliminated. Furthermore, the processing load is reduced compared to when steps S38 to S40 are performed for each bonding operation.

[0058] If the bonding quality is not good (step S41; No), that is, if the bonding quality is a warning, the degree of decrease in the bonding current in the voltage control is large, so the bonding condition corrector 52G corrects the bonding conditions so as to restore the bonding current (step S42). Here, a correction is made to increase the bonding voltage (i.e., the target voltage value of the voltage control).

[0059] Details of step S42 will be described with reference to FIG. 9. In FIG. 9, the change in the bonding current is represented by an approximation line L (the actual bonding current is a discrete value for each bonding). In FIG. 9, a bonding image is acquired for the p*Mth bonding (for example, if p=500 and M=5, this is the 2500th bonding). After a processing period T2 has elapsed since this time, a time Tc arrives when the bonding quality of the bonding point is determined to be a warning. This time Tc is defined as the present. During the processing period T2, multiple bondings are performed. However, deterioration of the bonding tool 21 (an increase in resistance and a decrease in bonding current) occurs each time a bonding is performed. Therefore, there is a difference between the bonding current at the time of the p*Mth bonding and the bonding current at the time of bonding after time Tc (the p*M+4th bonding time immediately after the bonding conditions are corrected). In step S42, this difference is taken into account and the bonding voltage V of the bonding conditions is corrected.

[0060] Specifically, the bonding condition corrector 52G multiplies the decrease D of the approximation line L derived in step S36 by the number of bondings performed during the processing period T2 plus one more. The number of bondings may be derived experimentally or set in advance, or the number of bondings actually performed during the period may be counted. The value obtained in this way indicates the decrease D1 in the bonding current at the next bonding (p*M+4th bonding) relative to the bonding current at the most recent bonding (p*Mth bonding) identified as having poor bonding quality in step S40.

[0061] Furthermore, the bonding condition corrector 52G subtracts the value of the bonding current at the most recent bonding identified in step S40 as having poor bonding quality from the ideal bonding current value set as the detection value for the ideal bonding. The value obtained thereby indicates a decrease D2 in the junction current at the most recent bonding identified as having poor bonding quality from the ideal junction current. The decrease D1+D2 is used to derive the difference between the ideal junction current and the most recent junction current. The bonding condition corrector 52G corrects the bonding conditions (particularly the junction voltage V) to increase the junction current by the decrease D1+D2. The ideal junction current may be specified as the junction current in the bonding conditions, or may be calculated from the average value of multiple monitored junction currents at an early stage after trial implantation.

[0062] The joining condition corrector 52G then derives a current value Ia by subtracting the decrease D1 from the value of the joining current during the joining identified as having poor joining quality. The current value Ia corresponds to the joining current during the next joining, i.e., the (p*M+4)th joining. The joining condition corrector 52G then divides the joining voltage V=Va (target value for voltage control) under the current joining conditions by the current value Ia thus derived to predict the resistance value R (i.e., the combined resistance value of the combination of the joining tool 21 and the workpiece W) during the next joining (p*M+4)th joining. The joining condition corrector 52G then multiplies the decrease D1+D2 by the resistance value R and adds the value ΔV obtained by the multiplication to the joining voltage V under the current joining conditions. If the joining voltage before correction is V=Va, the joining voltage V after correction becomes V=Va+ΔV=Vb. As a result, the target voltage value for voltage control increases from Va to Vb. This increases the junction current of the next junction, approaching the ideal junction current.

[0063] 3 and 7, the notification processing unit 52H and the order processing unit 52I determine whether the resistance value of the welding tool 21 has decreased to a predetermined resistance value (step S43). This determination is made, for example, based on whether the combined resistance value R calculated in step S42 exceeds a predetermined threshold value, or whether the welding voltage of the corrected welding conditions exceeds a predetermined threshold value. Because the resistance value of the welding tool 21 is proportional to the combined resistance value and the welding voltage, it is possible to determine the magnitude of the resistance value of the welding tool 21 from these values.

[0064] If the resistance value of the welding tool 21 increases to a predetermined resistance value, the welding tool 21 needs to be dressed or replaced at present or in the near future. Therefore, if the resistance value of the welding tool 21 increases to the predetermined resistance value (step S43; Yes), the notification processing unit 52H notifies the user of the need to dress or replace the welding tool 21 by externally outputting information via the interface 55 indicating that the welding tool 21 should be dressed or replaced (step S44). The order processing unit 52I also transmits an order for a new welding tool 21 (including, for example, tool identification information, a user ID, etc.) to the sales server 200 via the interface 55 (step S45). The order from the order processing unit 52I includes the tool identification information, etc. Upon receiving the order, the sales server 200 performs payment processing and processing for shipping the welding tool 21 using a predetermined method, such as instructing a shipping worker to ship the welding tool 21. As a result, a new welding tool 21 is provided to the user. Before transmitting an order for a new welding tool 21, the order processing unit 52I may confirm the order with the user via the interface 55. At this time, input such as the number of welding tools 21 may be received from the user. Note that if the resistance value drops to a second resistance value lower than the predetermined resistance value, the power required for welding is deemed excessive, and the operation of the entire welding system 10 may be terminated. At this time, a notification to that effect may be sent via the interface 55.

[0065] As described above, the processing device 50 includes a joining image acquisition unit 52E that acquires a joining image of the joining point of the workpiece W to be joined by the joining system 10, and a quality identification unit 52F that identifies the joining quality of the joining point shown in the joining image based on the acquired joining image. Furthermore, the processing device 50 further includes a joining condition correction unit 52G that corrects the joining conditions when the identified joining quality is not satisfactory. This configuration corrects the joining conditions based on the joining image, enabling correction of the joining conditions based on the appearance of the joining point. This ensures that the correction is performed properly, reducing the user's burden of monitoring the joining point and the effort required to correct the joining conditions. Therefore, this configuration improves the convenience of joining.

[0066] In this embodiment, the processing device 50 is configured to also operate as a controller of the bonding system 10, but the processing device 50 may be configured separately from the controller of the bonding system 10. In this case, the processing device 50 is, for example, externally attached to the controller.

[0067] The joining system 10 may be configured as a joining system in which the joining tool 21 is a heater tool. In such a case, the joining voltage is a voltage applied to both ends of the heater tool, and the joining current is a current that flows through the heater tool but does not flow through the workpiece W. As such a joining system, a pulse heat joining system that thermocompression bonds, solders, or brazes the members W1 and W2 of the workpiece W may be adopted. The joining system 10 may also be of another type.

[0068] The bonding system 10 may be configured as an ultrasonic bonding system. In this case, the bonding tool 21 includes a vibrator. In such a case, the bonding conditions include the frequency of the ultrasonic waves output by the vibrator, as well as the bonding voltage (in the case of voltage control) applied to the vibrator or the bonding current (in the case of current control) passed through the vibrator. Since the vibrator deteriorates with use and its resistance value increases, the bonding condition corrector 52G increases the bonding voltage or bonding current to obtain a desired amplitude.

[0069] In this embodiment, the bonding condition corrector 52G corrects the bonding conditions in response to deterioration of the bonding tool 21, thereby increasing the bonding current and bonding voltage in a new bonding. However, it may also be possible to determine that the bonding current and bonding voltage are excessive based on the bonding quality of the bonding image, and correct the bonding conditions at that time to reduce the bonding current and bonding voltage in a new bonding. In such a case, the determination results of "warning" and "defective" are further subdivided into a warning or defective product due to insufficient bonding power and a warning or defective product due to excessive bonding power.

[0070] In this embodiment, the above-mentioned bonding image is an image of the bonding portion of the workpieces W after bonding, but as another example, it may be an image of the bonding portion of the workpieces W during bonding. In such a case, the bonding image may be captured as a thermal image. The bonding quality can be determined by the level of temperature shown in the thermal image.

[0071] Furthermore, the quality identification unit 52F inputs the bonded image into the machine-learned quality determination model 103 and identifies the bonded quality from the bonded quality determination result output by the quality determination model 103. Such a configuration improves the accuracy of identifying the bonded quality and improves the convenience of bonding. In this embodiment, the quality determination model 103 is provided in the processing server 100, but it may also be provided in the processing device 50. In this case, the processor 52 may operate as the quality determination model 103 by executing or using a program and / or data realizing the quality determination model 103 stored in the storage device 51 to actually perform the processing performed by the quality determination model 103. Furthermore, the quality identification unit 52F may identify the bonded quality using a non-machine learning algorithm prepared in advance.

[0072] Furthermore, the quality specifying unit 52F inputs the joining image to one of the multiple quality determination models, a quality determination model 103, corresponding to the type of workpiece W. In this way, by preparing multiple quality determination models 103 and using the quality determination model 103 corresponding to the type of workpiece W, the accuracy of quality determination is improved, and convenience regarding joining is improved.

[0073] The multiple quality determination models 103 may be provided in a server device such as a processing server 100 that can communicate with the multiple joining systems 10A to 10C (multiple processing devices 50). This allows the multiple quality determination models 103 to be shared by the multiple joining systems 10A to 10C (each processing device 50), making it easier to update the multiple quality determination models 103 because it is no longer necessary to update each of the multiple processing devices separately.

[0074] The processing device 50 further includes a bonding monitoring unit 52B that derives a decrease D in the detected value per bonding operation based on the detected value of the bonding current or bonding voltage detected in each of multiple bonding operations periodically performed by the bonding system 10. The bonding condition correcting unit 52G corrects the bonding conditions to increase the bonding current or bonding voltage by the sum of a value D1 obtained by multiplying the decrease D by the number of bonding operations (+1) performed from the acquisition of the bonding image to the determination of the bonding quality, and a difference D2 obtained by subtracting the detected value for a bonding operation determined to be non-conforming from the value set as the detected value for an ideal bonding operation. This configuration allows the bonding conditions to be accurately corrected even when it takes a long time from the acquisition of the bonding image to the determination of the bonding quality, thereby improving the convenience of bonding.

[0075] As a modified example, the quality specifying unit 52F may specify the bond quality for a bond performed after a bonding in which the value of the bonding current or bonding voltage detected during bonding by the bonding system 10 falls below a predetermined threshold. For example, the bond monitoring unit 52B monitors the value of the bonding current or bonding voltage detected in successive bonding and detects that the value falls below a threshold that defines a warning range before the bond quality becomes defective. This detection may trigger the quality specifying unit 52F to start specifying the bond quality based on the bond image for the subsequent bonding. In this case, the bond image acquiring unit 52E may also start acquiring the bond image for the subsequent bonding. This configuration prevents the bond quality from being specified based on the bond image every time a bonding is performed, thereby reducing the processing load for bonding.

[0076] In this embodiment, the bonding quality includes at least a pass / fail product, a failure product, and a warning between the pass / fail product and the failure product, and the bonding condition corrector 52G corrects the bonding conditions when the bonding quality is a warning. This allows the bonding conditions to be corrected before the bonding quality becomes a failure product, thereby preventing unintended interruptions of the operation of the bonding system 10. As a result, the convenience of bonding is improved.

[0077] The processing device 50 further includes a notification processor 52H that performs notification processing to prompt the user to dress or replace the welding tool 21 when the resistance value of the welding tool 21 of the welding system 10 increases to a first predetermined value. This notification processing includes not only information display processing on the display device of the interface 55 but also audio output processing, turning on a predetermined alarm lamp, and transmitting information to a terminal carried by the user. This configuration allows the user to be prompted to dress or replace the welding tool 21, reducing the user's burden of monitoring the welding tool 21 and improving welding convenience. Note that, in light of this feature, it is not necessary to determine the welding quality based on the welding image. For example, the processing device 50 may include (1) a joining condition corrector 52G that corrects the joining conditions for the welding performed by the welding tool 21 of the welding system 10, and (2) a notification processor 52H that performs notification processing to prompt the user to dress or replace the welding tool 21 when the resistance value of the welding tool 21 increases to the first predetermined value. The first predetermined value may be set in advance as a value that is reached after the correction of the joining conditions is performed one or more times, for example. This prompts dressing or replacement of the joining tool 21 when the joining current, etc. required for joining cannot be secured or will not be secured in the near future due to the correction of the joining conditions.

[0078] The processing device 50 further includes an order processing unit 52I that processes an order for a new welding tool 21 to be replaced with the welding tool 21 of the bonding system 10 when the resistance value of the welding tool 21 increases to a second predetermined value. This order processing includes sending order data to a device, such as the sales server 200, that processes sales of the welding tool 21 outside the bonding system 10. This configuration facilitates ordering a new welding tool 21, improving welding convenience. This feature eliminates the need to determine the welding quality based on the bonding image. For example, the processing device 50 may include: (1) a bonding condition correction unit 52G that corrects the bonding conditions for the welding using the welding tool 21 of the bonding system 10 to increase the bonding current and bonding voltage of the new welding using the bonding system 10; and (2) an order processing unit 52I that processes an order for a new welding tool 21 to be replaced with the welding tool 21 when the resistance value of the welding tool 21 decreases to a predetermined value. The second predetermined value may be set in advance as a value that is reached after the joining conditions have been corrected one or more times. This allows the joining tool 21 to be ordered when the joining current required for joining cannot be secured or will not be secured in the near future due to the correction of the joining conditions. The second predetermined value may be greater than the first predetermined value. In this case, the order can be placed at an appropriate time when dressing is no longer sufficient.

[0079] Furthermore, the processing device 50 further includes a trial shot control unit 52C that sets the joining conditions for trial shots of a new welding tool 21 of the joining system 10 to joining conditions in which the power consumption per trial shot is greater than the power consumption per welding shot after the trial shot. This reduces the number of trial shots, thereby reducing the labor and waiting time required for trial shots and improving the convenience of welding. Note that, in light of this feature, it is not necessary to determine the welding quality based on the above-mentioned joining image. For example, the processing device 50 may include a trial shot control unit 52C that sets the joining conditions for trial shots of a new welding tool 21 of the joining system 10 to joining conditions in which the power consumption per trial shot is greater than the power consumption per welding shot after the trial shot. The processing device 50 may also include a joining execution unit 52A that performs trial shot welding on a workpiece for trial shots under the set joining conditions.

[0080] The waste shot control unit 52C may set the welding conditions for the waste shot based on the accumulated power required until the end of the waste shot. In this case, instead of the waste shot welding condition database 101 shown in FIG. 5, an accumulated power database 104 shown in FIG. 10, which associates tool identification information with the accumulated power required until the end of the waste shot, may be referenced. The accumulated power is derived and registered, for example, through simulation, experiment, or machine learning. In this case, the waste shot control unit 52C communicates with the processing server 100 to obtain the accumulated power corresponding to the tool identification information from the accumulated power database 104. The waste shot control unit 52C then divides the obtained accumulated power by the desired number of welding operations in advance and sets the target power consumption per welding operation as the waste shot condition. The welding execution unit 52A performs waste shot welding on the workpiece Wd by controlling the welding device 20 and the welding power supply 30 so that the power consumption derived from the current and voltage measured by the ammeter 43 and the voltmeter 44 becomes the target power consumption. This reduces the number of trial shots, thereby reducing the time and effort required for trial shots and waiting time, improving the convenience of joining.

[0081] In this embodiment, the processing device 50 further includes a tool image acquisition unit 52D that acquires a tool image of the welding tool 21. The waste shot control unit 52C determines whether the waste shot can be completed based on the acquired tool image and performs processing to terminate the waste shot when it is determined that the waste shot can be completed. The above example illustrates the process for terminating the waste shot by outputting to the interface 55 a message indicating that the waste shot can be completed and informing the user that the waste shot can be completed. However, this process may be any process, such as automatically terminating the waste shot. This allows the user to easily recognize when the waste shot should be completed, improving welding convenience. Note that, given this feature, it is not necessary to determine the welding quality based on the welding image. For example, the processing device 50 may include (1) a tool image acquisition unit 52D that acquires a tool image of the welding tool 21 used in welding in the welding system 10, and (2) a waste shot control unit 52C that determines whether the waste shot can be completed based on the acquired tool image and performs processing to terminate the waste shot when it is determined that the waste shot can be completed.

[0082] In this embodiment, the throwaway shot control unit 52C inputs a tool image into the throwaway shot discrimination model 102, which has been trained by machine learning, and determines whether the throwaway shot can be terminated based on the determination result output by the throwaway shot discrimination model 102 indicating whether the throwaway shot can be terminated. This configuration improves the accuracy of determining whether the throwaway shot can be terminated, thereby improving convenience regarding joining. In this embodiment, the throwaway shot discrimination model 102 is provided in the processing server 100, but it may also be provided in the processing device 50. In this case, the processor 52 may operate as the throwaway shot discrimination model 102 by executing or using a program and / or data that realizes the throwaway shot discrimination model 102, which is stored in the storage device 51, to actually perform the processing performed by the throwaway shot discrimination model 102. Alternatively, the throwaway shot control unit 52C may determine whether the throwaway shot can be terminated using a non-machine learning algorithm prepared in advance.

[0083] In this embodiment, the throwaway shot control unit 52C inputs a tool image to one of the multiple throwaway shot discrimination models 102 that corresponds to the type of joining tool 21. In this way, multiple throwaway shot discrimination models 102 are prepared, and the throwaway shot discrimination model 102 that corresponds to the type of workpiece W is used, thereby improving the accuracy of quality discrimination and improving joining convenience. The multiple throwaway shot discrimination models 102 are preferably provided in a server device such as a processing server 100 that can communicate with multiple processing devices 50. This allows the multiple throwaway shot discrimination models 102 to be shared by multiple processing devices 50, which makes it easier because it is no longer necessary to update the multiple throwaway shot discrimination models 102 for each processing device.

[0084] The connection conditions and the discrimination model may be generated by a processing server (server device) 100 through machine learning. As shown in FIG. 11 , the processing server 100 includes a non-volatile storage device (storage) 110 that stores various data, including connection conditions, databases, and data for implementing the discrimination model, programs, and the like, and a processor 120 such as a CPU that executes the programs and operates as a collection unit 121 and a machine learning unit 122 using the various data. The processor 120 can also operate as models 102, 103, and 105 generated through machine learning. Various parameters used in the models are optimized through machine learning and stored in the storage device 110 for use. The processing server 100 further includes a main memory 140 that stores data used by the processor 120. The processing server 100 also includes an interface 55. The interface 55 includes a communication interface, such as a network card connected to a communication network N, and a user interface, including a display device and an operation device.

[0085] The collection unit 121 collects, as connection data, data related to connection obtained at the time of connection from each of the multiple connection systems 10A to 10C used at multiple bases (companies, etc.) A to C, respectively, via the interface 55. The connection data includes the following information (1) to (3). (1) Monitor values ​​such as junction current and junction voltage acquired by the junction monitoring unit 52B of the junction system 10 (2) Various images acquired by the tool image acquisition unit 52D or the joint image acquisition unit 52E (3) Type of joining system (manufacturer, model number, etc.), type of joining tool (manufacturer, model number, material, shape, etc.), type of workpiece (material, shape, etc.)

[0086] Each piece of joining data is acquired by the joining system 10 every time a joining is performed, accumulated in chronological order, and transmitted to the collection unit 121 at an arbitrary timing. Attributes (whether or not the trial shot can be completed, the joining quality, etc.) may be added to the various images in (2) among the joining data by an arbitrary method (such as designation by the user). The collection unit 121 may analyze the joining data arranged in chronological order, and may add an attribute indicating that the trial shot cannot be completed to a tool image taken before the welding current stabilizes, for example, as a tool image taken during the trial shot period. The collection unit 121 may also analyze the joining data arranged in chronological order, and may add an attribute indicating that the trial shot cannot be completed to a welding image taken during a period when the welding current is stable and high, for example, as an image of a welded portion with good welding quality.

[0087] The machine learning unit 122 performs machine learning to generate a model used in the joining system 10, based on the welding data collected by the collection unit 121. The machine learning may be any method including deep learning.

[0088] For example, the machine learning unit 122 may read all collected data, input the content of (3) above, and generate a joining condition model 105 by machine learning that outputs optimal joining conditions (e.g., joining current and / or joining voltage) for the content. In such a case, the joining execution unit 52A or the trial shot control unit 52C of each joining system 10 may input the information of (3) above to the joining condition model 105 and set the joining conditions output from the joining condition model 105 as joining conditions (including trial shot conditions) to be used in subsequent joining. Furthermore, the trial shot control unit 52C may input at least the type of joining system or joining device and the type of joining tool from the information of (3) above to the joining condition model 105 and set the joining conditions output from the joining condition model 105 as trial shot conditions to be used in subsequent joining.

[0089] For example, the machine learning unit 122 reads the information in (2) above for each of the contents in (3) above, and generates the disposable discrimination model 102 and the quality discrimination model 103 by machine learning.

[0090] The generation of the above model improves the convenience of welding.

[0091] The configuration of the above embodiment can be modified as appropriate. For example, at least a portion of the processing device 50 and the processing server 100 may be configured using an ASIC (Application Specific Integrated Circuit) and / or an FPGA (Field Programmable Gate Array).

[0092] (Addendum) The following additional configurations are examples of the above-described embodiments and modifications. The additional configurations can be combined with each other or omitted. (Appendix 1) a joining image acquisition unit that acquires a joining image of a joining portion of the workpieces to be joined by the joining system; a quality specifying unit that specifies a joining quality of the joining portion shown in the joining image based on the acquired joining image; a bonding condition correcting unit that corrects bonding conditions when the specified bonding quality is not good; A processing device comprising: (Appendix 2) the quality identification unit inputs the bonded image into a quality determination model trained by machine learning, and identifies the bonded quality from a determination result of the bonded quality output by the quality determination model. 10. The processing device of claim 1. (Appendix 3) the quality specifying unit inputs the joining image into a quality determination model corresponding to the type of the workpiece from among a plurality of quality determination models; 3. The processing device of claim 2. (Appendix 4) a bonding monitoring unit that derives a decrease rate of a detection value of a bonding current or a bonding voltage per one bonding operation based on a detection value of the bonding current or the bonding voltage detected in each of a plurality of bonding operations periodically performed by the bonding system; the bonding condition correction unit corrects the bonding conditions so as to increase the bonding current or the bonding voltage by the sum of a value obtained by multiplying the number of bondings performed from acquisition of the bonding image to identification of the bonding quality + 1 by the degree of decrease, and a difference obtained by subtracting the detection value in a bonding when the bonding quality is determined to be poor from a value set as the detection value in an ideal bonding. The processing device according to any one of Supplementary Notes 1 to 3. (Appendix 5) the quality specifying unit specifies the bonding quality for bonding performed after a bonding in which a value of a bonding current or a bonding voltage detected during bonding by the bonding system becomes equal to or less than a predetermined threshold value. The processing device according to any one of Supplementary Notes 1 to 4. (Appendix 6) The joining quality includes at least a good product, a bad product, and a warning between the good product and the bad product; the joining condition correction unit corrects the joining conditions when the joining quality is in the warning state. The processing device according to any one of Supplementary Notes 1 to 4. (Appendix 7) a notification processing unit that performs notification processing to prompt a user to dress or replace the welding tool when a resistance value of the welding tool used to join the workpieces by the welding system increases to a first predetermined value. The processing device according to any one of Supplementary Notes 1 to 6. (Appendix 8) a joining condition correction unit that corrects joining conditions of joining by a joining tool of the joining system; a notification processing unit that performs notification processing to prompt a user to dress or replace the welding tool when the resistance value of the welding tool increases to a first predetermined value; A processing device comprising: (Appendix 9) an order processing unit that performs order processing for a new joining tool to be replaced with the joining tool when a resistance value of the joining tool used to join the workpieces by the joining system increases to a second predetermined value; A processing device according to any one of Supplementary Notes 1 to 8. (Appendix 10) a joining condition correction unit that corrects joining conditions of joining by a joining tool of the joining system; an order processing unit that performs order processing for a new welding tool to be replaced with the welding tool when the resistance value of the welding tool increases to a second predetermined value; A processing device comprising: (Appendix 11) The joining system further includes a trial shot control unit that sets the joining conditions for trial shots of a new joining tool to joining conditions in which power consumption per trial shot is greater than power consumption per joining shot after the trial shot. The processing device according to any one of Supplementary Notes 1 to 10. (Appendix 12) A trial shot control unit is provided for setting the joining conditions for trial shots of a new joining tool of the joining system to joining conditions in which the power consumption per trial shot is greater than the power consumption per joining shot after the trial shot. Processing equipment. (Appendix 13) The trial shot control unit sets the joining conditions for the trial shot based on the integrated power required until the trial shot is completed. 13. The processing device according to claim 11 or 12. (Appendix 14) a tool image acquisition unit that acquires a tool image obtained by capturing an image of the welding tool; the waste shot control unit determines whether the waste shot can be completed based on the acquired tool image, and performs processing to terminate the waste shot when it determines that the waste shot can be completed. The processing device according to any one of Supplementary Notes 11 to 13. (Appendix 15) a tool image acquisition unit that acquires a tool image obtained by capturing an image of a joining tool used in joining in the joining system; a discard shot control unit that determines whether the discard shot can be completed based on the acquired tool image, and performs processing to terminate the discard shot when it is determined that the discard shot can be completed; A processing device comprising: (Appendix 16) The reclaim-shot control unit inputs the tool image into a reclaim-shot discrimination model that has been machine-learned, and determines whether the reclaim-shot can be terminated based on a discrimination result indicating whether the reclaim-shot can be terminated, output by the reclaim-shot discrimination model. The processing device according to any one of Supplementary Notes 11 to 15. (Appendix 17) the throw-away shot control unit inputs the tool image into the throw-away shot discrimination model corresponding to the type of the joining tool among a plurality of throw-away shot discrimination models. 17. The processing device of claim 16. (Appendix 18) A processing device according to any one of Supplementary Notes 1 to 17; a joining tool used for joining the workpiece; A joining system comprising: (Appendix 19) a collection unit that collects data related to bonding obtained at the time of bonding from each of a plurality of bonding systems used at a plurality of locations; a machine learning unit that performs machine learning to generate a model to be used in the joint system based on the data collected by the collection unit; A server device comprising: (Appendix 20) The model is a quality discrimination model that inputs an image of a joining point of a workpiece to be joined and outputs the joining quality of the joining point. 20. The server device of claim 19. (Appendix 21) The model is a test-fire discrimination model that receives an input of a tool image obtained by capturing an image of a welding tool used in welding in a welding system, and outputs a discrimination result of whether test-fire of a new welding tool can be completed. 20. The server device of claim 19. (Appendix 22) The model is a joining condition model that receives information including a type of joining system, a type of joining tool of the joining system, and a type of workpiece as input, and outputs joining conditions to be set in the joining system. 20. The server device of claim 19. (Appendix 23) A program that causes a computer to function as a processing device according to any one of appendixes 1, 8, 10, 12, or 15, or as a server device according to appendix 19. (Appendix 24) a collecting step of collecting data related to bonding obtained at the time of bonding from each of a plurality of bonding systems used at a plurality of locations; a machine learning step of performing machine learning to generate a model to be used in the joint system based on the data collected by the collection unit; A generating method comprising:

[0093] (Scope of the present invention) Although the present invention has been described above with reference to the embodiments and modifications, the present invention is not limited to the above embodiments and modifications. For example, the present invention includes various modifications to the above embodiments and modifications that can be understood by a person skilled in the art within the scope of the technical concept of the present invention. The configurations listed in the above embodiments and modifications can be combined as appropriate within a range that does not contradict. Furthermore, the omission of each configuration is optional. [Explanation of symbols]

[0094] 10, 10A to 10C... Bonding system, 20... Bonding device, 21... Bonding tool, 21A... Electrode, 21B... Electrode, 22... Driving mechanism, 30... Bonding power supply, 41, 42... Camera, 43... Ammeter, 44... Voltmeter, 50... Processing device, 51... Storage device, 52... Processor, 52A... Bonding execution unit, 52B... Bonding monitoring unit, 52C... Control unit, 52D... Tool image acquisition unit, 52E... Bonding image acquisition unit, 52F... Quality identification unit, 52G... Bonding condition correction unit, 52H... Notification processing unit, 5 2I...order processing unit, 53...main memory, 55...interface, 100...processing server, 101...joining condition database, 102...trash shot discrimination model, 103...quality discrimination model, 104...accumulated power database, 105...joining condition model, 110...storage device, 120...processor, 121...collection unit, 122...machine learning unit, 200...sales server, A...company, L...approximation line, N...communication network, W...work, W1, W2...components, X...manufacturer.

Claims

1. a joining image acquisition unit that acquires a joining image of a joining portion of the workpieces to be joined by the joining system; a quality specifying unit that specifies a joining quality of the joining portion shown in the joining image based on the acquired joining image; a bonding condition correcting unit that corrects bonding conditions when the specified bonding quality is not good; A processing device comprising:

2. the quality identification unit inputs the bonded image into a quality determination model trained by machine learning, and identifies the bonded quality from a determination result of the bonded quality output by the quality determination model. The processing device of claim 1 .

3. the quality specifying unit inputs the joining image into a quality determination model corresponding to the type of the workpiece from among a plurality of quality determination models; The processing device of claim 2 .

4. a bonding monitoring unit that derives a decrease rate of a detection value of a bonding current or a bonding voltage per one bonding operation based on a detection value of the bonding current or the bonding voltage detected in each of a plurality of bonding operations periodically performed by the bonding system; the bonding condition correcting unit corrects the bonding conditions so as to increase the bonding current or the bonding voltage by the sum of a value obtained by multiplying the number of bondings performed from acquisition of the bonding image to identification of the bonding quality plus one time by the degree of decrease, and a difference obtained by subtracting the detection value in a bonding when the bonding quality is determined to be unacceptable from a value set as the detection value in an ideal bonding. The processing device of claim 1 .

5. the quality specifying unit specifies the bonding quality for bonding performed after a bonding in which a value of a bonding current or a bonding voltage detected during bonding by the bonding system becomes equal to or less than a predetermined threshold value. The processing device of claim 1 .

6. The joining quality includes at least a good product, a bad product, and a warning between the good product and the bad product; the joining condition correction unit corrects the joining conditions when the joining quality is in the warning state. The processing device of claim 1 .

7. a notification processing unit that performs notification processing to prompt a user to dress or replace the welding tool when a resistance value of the welding tool used to join the workpieces by the welding system increases to a first predetermined value. The processing device of claim 1 .

8. a joining condition correction unit that corrects joining conditions of joining by a joining tool of the joining system; a notification processing unit that performs notification processing to prompt a user to dress or replace the welding tool when the resistance value of the welding tool increases to a first predetermined value; A processing device comprising:

9. an order processing unit that processes an order for a new welding tool to be replaced with a welding tool used to join the workpieces by the welding system when the resistance value of the welding tool increases to a second predetermined value; The processing device of claim 1 .

10. a joining condition correction unit that corrects joining conditions of joining by a joining tool of the joining system; an order processing unit that processes an order for a new welding tool to be replaced with the welding tool when the resistance value of the welding tool increases to a second predetermined value; A processing device comprising:

11. The joining system further includes a trial shot control unit that sets the joining conditions for trial shots of a new joining tool to joining conditions in which power consumption per trial shot is greater than power consumption per joining shot after the trial shot. The processing device of claim 1 .

12. a trial shot control unit that sets the joining conditions for trial shots of a new joining tool of the joining system to joining conditions in which power consumption per trial shot is greater than power consumption per joining shot after the trial shot; Processing equipment.

13. The trial shot control unit sets the joining conditions for the trial shot based on the integrated power required until the trial shot is completed. The processing device according to claim 11 or 12.

14. a tool image acquisition unit that acquires a tool image obtained by capturing an image of the welding tool; the waste shot control unit determines whether the waste shot can be completed based on the acquired tool image, and performs processing to terminate the waste shot when it determines that the waste shot can be completed. The processing device of claim 11 .

15. a tool image acquisition unit that acquires a tool image obtained by capturing an image of a joining tool used in joining in the joining system; a discard shot control unit that determines whether the discard shot can be completed based on the acquired tool image, and performs processing to terminate the discard shot when it is determined that the discard shot can be completed; A processing device comprising:

16. The reclaim-shot control unit inputs the tool image into a reclaim-shot discrimination model that has been machine-learned, and determines whether the reclaim-shot can be terminated based on a discrimination result indicating whether the reclaim-shot can be terminated, output by the reclaim-shot discrimination model.

16. The processing device according to claim 14 or 15.

17. the throw-away shot control unit inputs the tool image into the throw-away shot discrimination model corresponding to the type of the joining tool, among a plurality of throw-away shot discrimination models; The processing device of claim 16.

18. a processing device according to claim 1, 8, 10, 12 or 15; a joining tool used for joining the workpiece; A joining system comprising:

19. a collection unit that collects data related to bonding obtained at the time of bonding from each of a plurality of bonding systems used at a plurality of locations; a machine learning unit that performs machine learning to generate a model to be used in the joint system based on the data collected by the collection unit; A server device comprising:

20. The model is a quality discrimination model that inputs an image of a joining point of a workpiece to be joined and outputs the joining quality of the joining point.

20. The server device according to claim 19.

21. The model is a test-fire discrimination model that receives an input of a tool image obtained by capturing an image of a welding tool used in welding in a welding system, and outputs a discrimination result of whether test-fire of a new welding tool can be completed.

20. The server device according to claim 19.

22. The model is a joining condition model that receives information including a type of joining system, a type of joining tool of the joining system, and a type of workpiece as input, and outputs joining conditions to be set in the joining system.

20. The server device according to claim 19.

23. A program that causes a computer to function as the processing device of any one of claims 1, 8, 10, 12, and 15, or the server device of claim 19.

24. a collecting step of collecting data related to bonding obtained at the time of bonding from each of a plurality of bonding systems used at a plurality of locations; a machine learning step of performing machine learning to generate a model to be used in the joint system based on the data collected in the collection step; A method for generating the same.

Images