Production data generation device, production data generation method, and recording medium

Abstract

Provided is a production data generation device capable of setting appropriate action parameters. A production data generation device (100) includes a model selection section (103) that selects at least one action parameter model (Pm) from among a plurality of action parameter models (Pm) that are different from each other; a parameter estimation section (105) that estimates an action parameter (m), which is an action condition of a component mounting device (M4) or (M5) for mounting a component (P) to be mounted to a board (B), based on the at least one selected action parameter model (Pm) and component information (d) related to the component (P) to be mounted to the board (B); and a data generation section (102) that generates production data (Dp) including component data (Dc) having the component information (d) and the action parameter (m).

Description

TECHNICAL FIELD

[0001] The present disclosure relates to an apparatus, a method, a program, and the like that generate production data for producing a mounting board. BACKGROUND

[0002] A component mounting line including at least one component mounting apparatus produces a mounting board by mounting components on a board. At this time, the component mounting line mounts components on a board based on production data. The production data includes identification information of each component mounted on a board and mounting order of these components. In addition, the production data sometimes includes component data of each component mounted on a board. The component data includes information indicating a shape or the like of a mounted component and an action parameter of a component mounting apparatus that processes the component. The action parameter includes, for example, a suction speed or a mounting load of a mounting head possessed by the component mounting apparatus.

[0003] For example, in the mounting board manufacturing system of Patent Literature 1, a control parameter (or a machine parameter) corresponding to the action parameter is corrected based on the result of a component mounting operation. Thereby, correction of component data can be performed appropriately and efficiently.

[0004] PRIOR ART DOCUMENTS

[0005] PATENT LITERATURE

[0006] Patent Literature 1: JP Patent Publication No. 2019-4129 SUMMARY

[0007] PROBLEMS TO BE SOLVED BY THE INVENTION

[0008] However, in the mounting board manufacturing system of Patent Literature 1 described above, there is a problem that it is sometimes difficult to set appropriate action parameters.

[0009] Therefore, the present disclosure provides an apparatus and the like that generate production data capable of setting appropriate action parameters.

[0010] MEANS FOR SOLVING THE PROBLEMS

[0011] An apparatus for generating production data according to an aspect of the present disclosure includes a model selection section that selects at least one learning model from a plurality of learning models that are different from each other and represent a relationship between an action condition of a component mounting apparatus that mounts components on a board and the components; a parameter estimation section that estimates an action condition, i.e., an action parameter, of the component mounting apparatus that mounts a mounting target component on a board, based on the selected at least one learning model and component information related to the mounting target component; and a data generation section that generates production data including component data having the component information and the action parameter.

[0012] In addition, these general or specific aspects can be realized by a system, a method, an integrated circuit, a computer program, or a recording medium such as a CD-ROM that is readable by a computer, and can also be realized by any combination of a system, a method, an integrated circuit, a computer program, and a recording medium. In addition, the recording medium can be a non-transitory recording medium.

[0013] Effects of Invention

[0014] The production data generation device of the present disclosure can set appropriate action parameters.

[0015] In addition, further advantages and effects of the present disclosure in one aspect will be apparent from the description and the drawings. The relevant advantages and / or effects are provided by the features described in the several embodiments and the description and the drawings, but it is not necessary to provide all of them in order to obtain one or more of the same features. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a diagram showing an example of the structure of the production system in Embodiment 1.

[0017] Figure 2 is a diagram showing an example of the structure of the component mounting device in Embodiment 1.

[0018] Figure 3 is a diagram partially showing an example of the A-A cross section in Figure 2

[0019] Figure 4 is a block diagram showing the functional structure of each of the production data generation device and the component mounting line in Embodiment 1.

[0020] Figure 5 is a diagram showing an example of the component library in Embodiment 1.

[0021] Figure 6 is a diagram showing an example of the production data in Embodiment 1.

[0022] Figure 7A is a diagram showing an example of a plurality of learning models managed in units of time, which are held by the learning model holding section in Embodiment 1.

[0023] Figure 7B is a diagram showing another example of a plurality of learning models managed in units of time, which are held by the learning model holding section in Embodiment 1.

[0024] Figure 8 is a diagram showing an example of a plurality of learning models managed in units of production equipment, which are held by the learning model holding section in Embodiment 1.

[0025] ​Figure 9A is a diagram indicating an example of a plurality of learning models managed in units of production types, which are held by the learning model holding section in Embodiment 1.

[0026] Figure 9B is a diagram indicating an example of a plurality of learning models managed in units of production types and production apparatuses, which are held by the learning model holding section in Embodiment 1.

[0027] Figure 10A is a diagram for explaining an outline of the action parameter estimation processing in Embodiment 1.

[0028] Figure 10B is a diagram for explaining an outline of the action parameter model learning processing in Embodiment 1.

[0029] Figure 11 is a diagram indicating an example of the overall processing in Embodiment 1.

[0030] Figure 12 is a diagram indicating another example of the overall processing in Embodiment 1.

[0031] Figure 13 is a flowchart indicating a processing action of the production data generation apparatus in Embodiment 1.

[0032] Figure 14 is a diagram indicating an example of the structure of the production system in Embodiment 2.

[0033] Figure 15 is a block diagram indicating the functional structures of the production management apparatus, the component mounting line, and the processing apparatus, respectively, in Embodiment 2.

[0034] Figure 16 is a diagram indicating an example of the overall processing in Embodiment 2.

[0035] Figure 17A is a diagram indicating an example of the screening information generated based on the inspection result of the mounting substrate in Embodiment 2.

[0036] Figure 17B is a diagram indicating an example of the screening information generated based on the mounting actual performance of the component mounting line in Embodiment 2.

[0037] Figure 18A is a diagram indicating an example of the screening information generated by the selection of the component in Embodiment 2.

[0038] Figure 18B is a diagram indicating an example of the screening information generated by the selection of the substrate in Embodiment 2.

[0039] Figure 19is a flowchart showing a processing action of a production management device in Embodiment 2. DETAILED DESCRIPTION

[0040] To solve the above problem, a production data generation device according to an aspect of the present disclosure includes: a model selection section that selects at least one learning model from a plurality of learning models that respectively represent relationships between action conditions of a component mounting device that mounts components to a board and the components; a parameter estimation section that estimates, based on the selected at least one learning model and component information related to a mounting target component mounted to the board, an action condition, i.e., an action parameter, of the component mounting device for mounting the mounting target component to the board; and a data generation section that generates production data including component data having the component information and the action parameter. For example, the component information can indicate at least one of a size, a shape, an appearance, a category, and a supply method for supplying the component corresponding to the component information. Further, the action parameter can be a parameter related to at least one of transfer, recognition, suction, and assembly of the component by the component mounting device.

[0041] Thus, since at least one learning model is selected from a plurality of learning models that are different from each other for use in estimation of the action parameter, it is possible to improve the likelihood of estimating a suitable action parameter for the mounting target component. Therefore, it is possible to set a suitable action parameter. Further, in a case where component data having such an action parameter and component information is included in production data and the production data is used in mounting of components to a board by a component mounting device, it is possible to produce a mounted board with good quality. That is, it is possible to seek quality improvement of the mounted board.

[0042] In addition, the action parameter can be not only one parameter but also a set of a plurality of parameters. In this case, when a plurality of learning models is selected by the model selection section, the parameter estimation section can estimate a plurality of parameters included in the action parameter respectively in accordance with the selected plurality of learning models.

[0043] Further, the production data generation device can further include: a data acquisition section that acquires actual performance production data including actual performance component data having component information related to a mounted component and an action parameter used in mounting of the mounted component and used by a component mounting device; and a learning section that performs updating of the relationships represented by a learning model corresponding to the acquired actual performance production data from among the plurality of learning models through learning using the actual performance component data as teaching data.

[0044] The action parameter of the actual performance component data included in the actual performance production data is used in mounting of the mounted component, at which time correction or the like is performed. That is, the action parameter is corrected so that a mounted board of better quality is produced. Therefore, by using the actual performance component data having such an action parameter as teaching data in learning of the learning model, further properization of the learning model can be sought. As a result, in a case where the learning model is selected by the model selection section, the estimation accuracy of the action parameter can be improved.

[0045] Further, the plurality of learning models can respectively correspond to mutually different periods, and the learning section can learn the learning model corresponding to the period in which the actual performance production data is acquired.

[0046] For example, one of the plurality of learning models can correspond to the entire period (for example, the entire period from the past to the present), and the remaining at least one learning model can respectively correspond to mutually different eras. The mutually different eras are, for example, the 1990s, the 2000s, the 2010s, and the like. Thereby, the learning model corresponding to the entire period or an arbitrary era is selected from among the learning models and used in estimation of the action parameter. Therefore, a proper action parameter corresponding to the period can be estimated for the component to be mounted.

[0047] Further, the plurality of learning models can respectively correspond to mutually different production equipment, and the learning section can learn the learning model corresponding to the production equipment including the component mounting device in which the actual performance production data is used.

[0048] For example, the production equipment can be a component mounting device, can be one component mounting line including a component mounting device, or can be equipment including a plurality of component mounting lines. In this case, for example, one of the plurality of learning models can correspond to production equipment including all component mounting lines arranged in a factory, and the remaining at least one learning model can respectively correspond to mutually different component mounting lines. Thereby, the learning model corresponding to all component mounting lines or an arbitrary component mounting line is selected from among the learning models and used in estimation of the action parameter. Therefore, a proper action parameter corresponding to the production equipment can be estimated for the component to be mounted.

[0049] Further, the plurality of learning models can respectively correspond to mutually different types of mounted boards, and the learning section can learn the learning model corresponding to the type of the mounted board produced using the actual performance production data.

[0050] For example, the type of mounting substrate is a mass production type or a trial production type. In this case, for example, one of the plurality of learning models corresponds to the mass production type, and the remaining one corresponds to the trial production type. Thereby, the learning model corresponding to the mass production type or the trial production type is selected from among the learning models, and used in the estimation of the action parameter. Therefore, the appropriate action parameter corresponding to the type of mounting substrate can be estimated for the mounting target component.

[0051] Embodiments will be specifically described below with reference to the drawings.

[0052] In addition, each of the embodiments described below indicates a general or specific example. The numerical values, shapes, materials, component configurations, arrangement positions of components, connection methods, steps, order of steps, and the like shown in the following embodiments are examples, and the gist is not intended to be limited to the disclosure. Furthermore, regarding components among the components in the following embodiments that are not described in the components of the independent claims indicating the most general concept, the components are described as arbitrary components.

[0053] In addition, each of the drawings is a schematic view, and the illustration is not necessarily strict. Furthermore, in each of the drawings, the same reference numerals are attached to the same components.

[0054] (Embodiment 1)

[0055] [Production system]

[0056] Figure 1 is a view that indicates an example of the structure of the production system in the present embodiment.

[0057] The production system 1 in the present embodiment is provided with three component mounting lines L1 to L3 and a production data generation device 100.

[0058] The component mounting lines L1 to L3 are each an example of a production apparatus that mounts a substrate, and produces a mounting substrate by performing a solder printing work, a component mounting work, a reflow soldering work, and the like on a substrate that is transported from an upstream side, and transports the produced mounting substrate to a downstream side.

[0059] The production data generation device 100 generates and outputs production data for producing a mounting substrate for each of the component mounting lines L1 to L3. In addition, the production data generation device 100 can communicate with these component mounting lines L1 to L3 via wireless or wired. The wireless can be Wi-Fi (registered trademark), Bluetooth (registered trademark), ZigBee, or a specific low power radio.

[0060] The component mounting line Ll is provided with the line management device 200, the board supply device Ml, the board handover device M2, the solder printing device M3, the component mounting devices M4 and M5, the reflow soldering device M6, and the board recovery device M7. In addition, the devices included in the component mounting line Ll other than the line management device 200 are arranged in the order of the board supply device Ml, the board handover device M2, the solder printing device M3, the component mounting devices M4 and M5, the reflow soldering device M6, and the board recovery device M7 and are connected in series. In addition, these devices other than the line management device 200 are referred to as work devices hereinafter. In addition, the component mounting line Ll can not include all of the above work devices as long as it includes the board supply device Ml, at least one component mounting device, and the board recovery device M7. Further, the component mounting line Ll can include, in addition to the above work devices, a solder coating device that coats a board with solder, a component insertion machine that mounts a radial component or an axial component to a board, and the like.

[0061] The line management device 200 acquires the production data generated by the production data generation device 100, and causes the work devices included in the component mounting line Ll to perform production of mounting boards based on the production data.

[0062] The board supply device Ml supplies a board used in the production of a mounting board in the component mounting line Ll to the solder printing device M3 via the board handover device M2. The solder printing device M3 performs the above-described solder printing work. That is, the solder printing device M3 prints solder on a board handed over from the board handover device M2 in a screen printing manner.

[0063] The component mounting devices M4 and M5 each perform the above-described component mounting work of mounting at least one component to a board. In addition, the component mounting line Ll is provided with two component mounting devices M4 and M5, but the number of devices is not limited to two, and can be one or more than three. Further, it can be said that a mounting board is substantially produced by the component mounting work performed by the component mounting devices M4 and M5.

[0064] The reflow soldering device M6 performs the above-described reflow soldering work. That is, the reflow soldering device M6 heats a board on which a component is mounted, which is carried in from the component mounting devices M4 and M5, to harden solder on the board and join an electrode portion of the board and the component. Specifically, the reflow soldering device M6 melts and solidifies solder for joining a component by performing heating that follows a given heating profile. Thus, the component is joined to the board in a solder joining manner. The board recovery device M7 recovers the board on which the solder joining is performed from the reflow soldering device M6.

[0065] The component mounting lines L2 and L3 also have the same structure as the component mounting line LI. In addition, in the present embodiment, the component mounting lines LI to L3 each have the same structure, but can have mutually different structures. Furthermore, in the present embodiment, the component mounting lines LI to L3 are provided with the line management device 200, but the line management device 200 can be provided independently for each of the component mounting lines LI to L3, and can also be assembled into each of the component mounting lines LI to L3.

[0066] [Component mounting device]

[0067] Figure 2 is a view showing an example of the structure of the component mounting device M4. In the present embodiment, the component mounting device M5 also has the same structure as the component mounting device M4. In addition, in the present embodiment, the direction of conveyance of the substrate B is referred to as the X-axis direction, and the direction perpendicular to the X-axis direction is referred to as the Y-axis direction. The X-axis direction and the Y-axis direction are directions along a horizontal plane. Furthermore, the direction perpendicular to the X-axis direction and the Y-axis direction is referred to as the Z-axis direction. The positive side and the negative side of the X-axis direction are the downstream side and the upstream side, respectively, in the direction of conveyance of the substrate B, and the positive side and the negative side of the Y-axis direction are the rear side (or the depth side) and the front side (or the near front side), respectively, in the front-rear direction. The positive side and the negative side of the Z-axis direction are the upper side and the lower side, respectively, in the up-down direction. In the present embodiment, the direction of conveyance of the substrate B is the X-axis direction, and the direction perpendicular to the X-axis direction is the Y-axis direction. In addition, in the present embodiment, the direction perpendicular to the X-axis direction and the Y-axis direction is the Z-axis direction. Figure 2 The upper surface of the component mounting device M4 is shown in FIG. 1.

[0068] The component mounting device M4 is provided with a base 4, a substrate conveyance mechanism 5, two component supply portions 6, two X-axis beams 9, a Y-axis beam 8, two mounting heads 10, two component recognition cameras 11, and two substrate recognition cameras 12.

[0069] The substrate conveyance mechanism 5 is provided with two guide rails in the X-axis direction, and is disposed in the center of the base 4. The substrate conveyance mechanism 5 conveys the substrate B that is carried in from the upstream side, and positions and holds the substrate B at a position for performing a component mounting operation.

[0070] The two component supply portions 6 are disposed so as to sandwich the substrate conveyance mechanism 5 in the Y-axis direction. A plurality of feeders 7 are disposed in parallel with each other in the X-axis direction in each of the component supply portions 6. The feeder 7 supplies a component to a position at which the component is taken out by the mounting head 10 (hereinafter referred to as a component take-out position) by pitch-feeding a component tape in which components are housed in a tape conveyance direction.

[0071] In addition, a tray feeder, a tube feeder, or a bulk feeder, or the like can be disposed in the component supply portion 6. The tray feeder supplies a component from a tray in which the component is housed. The tube feeder supplies a component from a tubular case in which the component is housed. The bulk feeder supplies a component from a bulk case in which the component is housed.

[0072] The Y-axis beam 8 is provided on the end of the one side of the upper surface of the base 4 in the X-axis direction (the right side in the middle) Figure 2 The two X-axis beams 9 are freely movable in the Y-axis direction in a state of being aligned in the X-axis direction, and are coupled to the Y-axis beam 8.

[0073] The mounting head 10 is freely movable in the X-axis direction with respect to each of the two X-axis beams 9. The mounting head 10 has a plurality of suction assemblies 10a that are capable of ascending and descending while holding a component. A suction nozzle 10b (see FIG. 2) is provided at the front end of each of the suction assemblies 10a. Figure 3 ].

[0074] The two mounting heads 10 are moved in the X-axis direction and the Y-axis direction by driving the Y-axis beam 8 and the X-axis beams 9, respectively. Thus, the two mounting heads 10 respectively suck and take out components from component take-out positions of the feeders 7 provided in the component supply section 6 corresponding to the mounting heads 10, and mount the components to mounting points (or mounting positions) of the substrates B positioned on the substrate conveyance mechanism 5.

[0075] The two component recognition cameras 11 are provided between one of the two component supply sections 6 and the substrate conveyance mechanism 5. The component recognition cameras 11 take an image of a component when the mounting head 10 that has taken out the component from the component supply section 6 moves above the component recognition cameras 11. That is, the component recognition cameras 11 recognize the holding posture of a component by taking an image of the component held in the state of the mounting head 10.

[0076] The substrate recognition camera 12 is mounted on the disc 9a that houses the mounting head 10. Thus, the substrate recognition camera 12 moves integrally with the mounting head 10. Such a substrate recognition camera 12 moves above the substrate B positioned on the substrate conveyance mechanism 5 in conjunction with the movement of the mounting head 10, takes an image of a substrate mark (not shown) provided on the substrate B, and recognizes the position of the substrate B. In the mounting of a component to the substrate B by the mounting head 10, the mounting position is corrected based on the recognition result of the component by the component recognition cameras 11 and the recognition result of the position of the substrate B by the substrate recognition camera 12.

[0077] Figure 3 is a view partially showing an example of the A-A cross section in Figure 2 . The component mounting apparatus M4 has a function of mounting a component P to a substrate B.

[0078] The component supply section 6 has a feeder base 13a, a plurality of feeders 7 mounted on the feeder base 13a, and a trolley 13 that supports the feeder base 13a, as shown in Figure 3 .

[0079] The trolley 13 is configured to freely attach and detach the component mounting devices M4, M5, and further has a cassette rack 15. The cassette rack 15 is configured to be able to hold a plurality of component reels C. The component reels C hold component tapes 14 in a wound state. The plurality of component reels C are respectively held in an upper holding position Hu or a lower holding position Hd of the cassette rack 15. The component tape 14 pulled out from the component reel C held by the cassette rack 15 is fitted to the feeder 7. Alternatively, the feeder 7 can be disposed on a feeder base 13a provided on the base 4 without using the trolley 13. Further, the component reels C can be held by the trolley 13 instead of the cassette rack 15.

[0080] Further, in the present embodiment, as described above, the component mounting devices M4 and M5 each have the same structure, but can have mutually different structures.

[0081] [Function structure of production data generation device and component mounting line]

[0082] Figure 4 is a block diagram showing the function structure of each of the production data generation device 100 and the component mounting lines L1 to L3.

[0083] The production data generation device 100 has a control section 101, a data generation section 102, a model selection section 103, a learning section 104, a parameter estimation section 105, a display section 106, an input / output section 107, a data acquisition section 108, a production data holding section DB1, a learning model holding section DB2, and a component library holding section DB3.

[0084] The model selection section 103 selects at least one learning model from among a plurality of learning models held in the learning model holding section DB2.

[0085] The learning model holding section DB2 holds the plurality of learning models described above. These plurality of learning models are mutually different models, and represent the relationship between the action condition of the component mounting device M4 or M5 for mounting a component P to a board B and the component P.

[0086] The parameter estimation section 105 estimates the action condition, i.e., the action parameter, of the component mounting device M4 or M5. That is, the parameter estimation section 105 estimates the action parameter for mounting the mounting target component P to the board B, based on at least one learning model selected by the model selection section 103 and component information related to the mounting target component P mounted to the board B.

[0087] The data generation section 102 generates production data containing component data having the above-described component information and operation parameters. Here, the production data indicates, for example, the mounting order of at least one component P mounted on the substrate B and the positions at which these components P are mounted to the substrate B (i.e., the above-described mounting positions), and contains the component data of each of the at least one component P. Further, the component data of each component P is held in the component library holding section DB3. That is, the component library holding section DB3 holds a component library containing the component data of each of a plurality of components P. Therefore, if the component data of the component P mounted on the substrate B is contained in the component library, the data generation section 102 selects the component data of the component P from the component library, and generates production data containing the selected component data. On the other hand, if the component data of the component P mounted on the substrate B is not contained in the component library, the data generation section 102 generates production data containing component data having the component information of the component P and the operation parameters estimated as described above.

[0088] The data generation section 102 generates and outputs such production data for each of the component mounting lines L1 to L3, and stores the production data in the production data holding section DB1.

[0089] The data acquisition section 108 acquires actual performance production data containing actual performance component data having component information related to a mounted component P and operation parameters used in the mounting of the mounted component P, and used by the component mounting apparatus M4 or M5.

[0090] The actual performance production data is, for example, data in which the production data generated by the data generation section 102 is corrected or adjusted, and the like. That is, the component mounting apparatus M4 or M5 included in each of the component mounting lines L1 to L3 mounts a component P to a substrate B based on the production data, but sometimes produces a defective mounting substrate through the mounting. In such a case, in each of the component mounting lines L1 to L3, correction or adjustment of the component data included in the production data is performed in order to reduce the frequency of defective products. The actual performance production data containing the actual performance component data is generated through such correction or adjustment. The component mounting apparatus M4 or M5 of each of the component mounting lines L1 to L3 uses the actual performance production data to perform mounting of a component P to a substrate B. The data acquisition section 108 acquires the actual performance production data thus generated from each of the component mounting lines L1 to L3.

[0091] The learning unit 104 generates or updates a learning model through machine learning. Hereinafter, the machine learning will be simply referred to as learning. For example, the learning unit 104 generates a learning model through learning, and stores the generated learning model in the learning model holding unit DB2. Further, the learning unit 104 selects one learning model from among the plurality of learning models held in the learning model holding unit DB2, and updates the selected learning model through learning. The actual performance production data used by each of the component mounting lines L1 to L3 and acquired by the data acquisition unit 108 is used in the learning of the learning unit 104.

[0092] That is, the learning unit 104 performs updating of the relationship represented by the learning model corresponding to the actual performance production data acquired by the data acquisition unit 108, among the plurality of learning models held in the learning model holding unit DB2. At this time, the learning unit 104 performs the updating through learning using the actual performance component data included in the actual performance production data as teaching data. Further, the learning model can be, for example, a neural network, a decision tree, or another model.

[0093] The display unit 106 displays the production data held in the production data holding unit DB1, the component library held in the component library holding unit DB3, and the like. Specific examples of the display unit 106 are a liquid crystal display, a plasma display, an organic EL (Electro-Luminescence) display, or the like, but are not limited to these.

[0094] The input / output unit 107, for example, accepts input data based on an operator's operation of the production system 1, and outputs the input data to the control unit 101. Such an input / output unit 107 has, for example, a keyboard, a touch sensor, a touch pad, a mouse, or the like. Further, the input / output unit 107 performs output of data to the component mounting lines L1 to L3 and input of data from the component mounting lines L1 to L3. The production data generated by the data generation unit 102 can be output to each of the component mounting lines L1 to L3 via the input / output unit 107. Further, the input / output unit 107 can acquire the above-described component information based on an operator's operation, and output it to the parameter estimation unit 105.

[0095] The control unit 101 controls each of the constituent elements included in the production data generation apparatus 100, except for the control unit 101. For example, the control unit 101 controls each of the constituent elements based on the input data of the operator accepted by the input / output unit 107, or the like.

[0096] The production data holding section DB1, the learning model holding section DB2, and the component library holding section DB3 are recording media for holding production data, a learning model, and a component library. Such a recording medium is, for example, a hard disk, a ROM (Read Only Memory), a RAM (Random Access Memory), or a semiconductor memory. Such a recording medium can be volatile or non-volatile.

[0097] The component mounting line L1 includes an operation control section 211, an input / output section 212, a display section 213, an operation mechanism 214, and a production data holding section DB4. Each of the components included in the component mounting line L1 except for the operation mechanism 214 can be included in the line management device 200 or in any operation device different from the line management device 200.

[0098] The input / output section 212, like the input / output section 107 of the production data generation device 100, for example, accepts input data based on an operation of an operator of the production system 1 and outputs the input data to the operation control section 211. Such an input / output section 212 can have, for example, a keyboard, a touch sensor, a touch pad, a mouse, or the like. Further, the input / output section 212 performs output of data to the production data generation device 100 and input of data from the production data generation device 100. For example, the input / output section 212 acquires production data from the production data generation device 100 and stores the production data in the production data holding section DB4.

[0099] The display section 213 displays production data and the like held in the production data holding section DB4. A specific example of the display section 213 is a liquid crystal display, a plasma display, an organic EL display, or the like, but is not limited to these.

[0100] The operation mechanism 214 is configured of a mounting head 10 and a feeder 7 and the like mechanism for producing a mounting substrate.

[0101] The job control section 211 controls each of the components included in the component mounting line Ll other than the job control section 211. For example, the job control section 211 controls each of the components based on the input data of the operator accepted by the input / output section 212 or the like. For example, the job control section 211 causes the job mechanism 214 to execute at least one of the solder printing job, the component mounting job, and the reflow soldering job based on the production data held in the production data holding section DB4. Further, the job control section 211 performs correction or adjustment of the production data held in the production data holding section DB4, or the like, in accordance with the input data of the operator accepted by the input / output section 212. Thus, the actual performance production data is generated. The job control section 211 outputs the actual performance production data from the input / output section 212 to the production data generating apparatus 100 by controlling the input / output section 212.

[0102] The production data holding section DB4 is a recording medium for holding production data. For example, such a recording medium is a hard disk, a RAM, a ROM, a semiconductor memory, or the like. In addition, such a recording medium can be volatile or non-volatile.

[0103] [Component Library]

[0104] Figure 5 is a diagram showing an example of a component library.

[0105] The component library is constituted by a plurality of component data Dc. Each of the plurality of component data Dc is data of one kind of component P, and is associated with a component code for identifying the kind of the component P. Such component data Dc has component information d related to the component P and an action condition, i.e., an action parameter m, of the component mounting apparatus M4 or M5 for mounting the component P to the board B. In addition, in the component data Dc shown in Figure 5 The blank portion of each item or the like in the component data Dc shown in

[0106] The component information d includes, for example, a shape diagram dl of the component P, dimension data d2, and component parameters d3.

[0107] The shape diagram dl illustrates the outline of the component P corresponding to the component data Dc. The dimension data d2 represents, in numerical values, information related to the dimensions of the component P, i.e., the outline dimensions, the number of leads, the lead pitch, the lead length, the lead width, the component height, and the like.

[0108] The component parameter d3 is attribute information for the component P. Such a component parameter d3 includes information relating to the component P itself, i.e., component attribute d31, and information relating to the component carrier tape 14 for feeding the component P by the feeder 7, i.e., carrier tape information d32. The component attribute d31 indicates, for example, polarity of the component P, polarity flag, flag position, component category, and shape category. The carrier tape information d32 includes, for example, carrier tape material of the component carrier tape 14, carrier tape width indicating width dimension of the component carrier tape 14, transport interval indicating carrier tape transport pitch of the component carrier tape 14 of the feeder 7, and information relating to color and material of the component carrier tape 14.

[0109] Thus, the component information d in the present embodiment indicates at least one of size, shape, appearance, category, and feeding method for feeding the component P corresponding to the component information d. In addition, the feeding method corresponds to the carrier tape information d32 or the like, for example.

[0110] The action parameter m is a machine parameter that specifies a method of action when the component P is mounted to the board B by the component mounting device M4 or M5. In the example shown here, the action parameter m includes model information ml indicating a kind of the component mounting device M4 or M5 and nozzle setting information m2 indicating a kind of the suction nozzle 10b used. Further, the action parameter m includes a speed parameter m3, recognition information m4, gap information m5, suction information m6, and mounting information m7, etc.

[0111] The speed parameter m3 includes a lift speed when the component P is suctioned by the suction nozzle 10b, a mounting speed when the component P is transported by the mounting head 10, and a carrier tape transport speed when the component carrier tape 14 is transported by the feeder 7. The recognition information m4 is a parameter that specifies a method of component recognition. Specifically, the recognition information m4 includes camera category indicating a kind of the component recognition camera 11 used, illumination mode indicating a method of illumination when the component recognition camera 11 takes an image, and recognition speed indicating a moving speed of the mounting head 10 when taking an image. The gap information m5 includes a suction gap when the component P is suctioned by the suction nozzle 10b and a mounting gap when the suctioned component P is mounted to the board B.

[0112] The suction information m6 includes a suction position offset indicating an offset amount when the component P is suctioned by the suction nozzle 10b and a suction angle. The mounting information m7 indicates a press load as a mounting load when the component P suctioned by the suction nozzle 10b is mounted to the board B.

[0113] Thus, the action parameter m in the present embodiment is a parameter relating to at least one of component P transport, recognition, suction, and mounting by the component mounting device M4 or M5.

[0114] In addition,Figure 5 The component information d and the action parameter m included in the component data Dc are each an example, and can represent Figure 5 information other than the information shown in the above equation (1), and can also be represented Figure 5 the information shown in the above equation (1) and other information, and can also represent Figure 5 part of the information shown in the above equation (1). Furthermore, the number of pieces of information included in each of the component information d and the action parameter m can be one or a plurality of pieces.

[0115] In the present embodiment, the data generation section 102 selects the component data Dc corresponding to the component P mounted on the board B from the component library held in the component library holding section DB3 when generating the production data, and generates the production data including the component data Dc.

[0116] Furthermore, if the component data Dc corresponding to the component P mounted on the board B does not exist in the component library, the data generation section 102 generates the production data using the component data Dc not included in the component library. For example, the data generation section 102 generates the production data using the component data Dc having the component information d acquired by the input / output section 107 and the action parameter m estimated by the parameter estimation section 105 from the component information d. The component information d acquired by the input / output section 107 can be, for example, information acquired from CAD (Computer Aided Design) information related to the component P mounted on the board B, or information input by an operator's operation.

[0117] Furthermore, a default action parameter m can be set to the component data Dc of the component library. If a default action parameter m is set to the component data Dc selected from the component library, the data generation section 102 can cause the parameter estimation section 105 to estimate the action parameter m corresponding to the component information d included in the component data Dc. If the action parameter m is estimated by the parameter estimation section 105, the data generation section 102 replaces the default action parameter m included in the component data Dc with the action parameter m estimated by the parameter estimation section 105. Then, the data generation section 102 generates the production data using the component data Dc in which the replacement of the action parameter m is performed.

[0118] In addition, all of the information (i.e., parameters) included in the action parameter m of the component data Dc can be default, or only a part of the parameters can be default. In the case where only a part of the parameters is default, the parameter estimation section 105 can estimate parameters that are substitutes for the part of the parameters that are default. The data generation section 102 replaces the part of the parameters among the action parameter m included in the component data Dc with the parameters estimated by the parameter estimation section 105. Then, the data generation section 102 generates the production data using the component data Dc in which the replacement of the part of the parameters is performed.

[0119] [Production data]

[0120] Figure 6 is a diagram that shows an example of production data Dp.

[0121] In the production data Dp, for example, the component name and the component code of each of the plurality of components P mounted on the substrate B are arranged in the mounting order of the plurality of components P. In addition, the component code of the component P is a code for specifying the component data Dc of the component P from a component library. Further, the production data Dp indicates, for each of the plurality of components P, the assembly coordinates of the component P, the assembly angle of the component P, the identification information of the feeder 7, and the identification information of the mounting head 10 or the suction nozzle 10b. The assembly coordinates of the component P are the position in the substrate B at which the component P is assembled or mounted, and are also referred to as a mounting point, an assembly position, or a mounting position. The assembly angle of the component P is the angle at which the suction nozzle 10b that sucks the component P rotates with the center axis of the suction nozzle 10b as the rotation axis in order to mount the component P to the substrate B. The identification information of the feeder 7 corresponding to the component P is information for identifying the feeder 7 that supplies the component P. The identification information of the mounting head 10 corresponding to the component P is information for identifying the mounting head 10 that is used in order to mount the component P to the substrate B.

[0122] For example, the production data Dp indicates, for the component P that is initially mounted on the substrate B, the component name "A component", the component code "C001", the assembly coordinates "x1, y1", the assembly angle "θ1", the identification information "F2" of the feeder 7, and the identification information "H3" of the mounting head 10.

[0123] Further, the production data Dp in the present embodiment includes the component data Dc of each of the plurality of components P mounted on the substrate B. For example, the production data Dp includes the component data Dc that corresponds to the component code "C001" possessed by the component P of the component name "A component".

[0124] [Learning model]

[0125] Figure 7A and Figure 7B is a diagram that shows an example of a plurality of learning models managed in units of time, which are held by the learning model holding unit DB2. In addition, the learning model in the present embodiment will also be referred to as an action parameter model hereinafter.

[0126] For example, the plurality of learning models, that is, the plurality of action parameter models Pm11 to Pm14, held in the learning model holding unit DB2 are respectively as follows Figure 7Aare managed in units of time as shown. Specifically, the action parameter model Pm11 is generated by learning using actual performance production data used between 1990 and 1999. Likewise, the action parameter model Pm12 is generated by learning using actual performance production data used between 2000 and 2009, and the action parameter model Pm13 is generated by learning using actual performance production data used between 2010 and now. Further, the action parameter model Pm14 is generated by learning using actual performance production data used between 1990 and now.

[0127] Further, the plurality of learning models, i.e., the plurality of action parameter models Pm21 to Pm25, held in the learning model holding unit DB2 are each managed in units of time as shown, and further, can be managed in chronological order. Figure 7B

[0128] For example, the action parameter model Pm21 is generated by learning using actual performance production data used on July 1, 2019. The action parameter model Pm22 is generated by learning using the action parameter model Pm21 and actual performance production data used on July 2, 2019. That is, the action parameter model Pm22 is generated by learning using actual performance production data used between July 1 and 2, 2019.

[0129] Likewise, the action parameter model Pm23 is generated by learning using the action parameter model Pm22 and actual performance production data used on July 3, 2019. That is, the action parameter model Pm23 is generated by learning using actual performance production data used between July 1 and 3, 2019. The action parameter model Pm24 is generated by learning using the action parameter model Pm22 and actual performance production data used on July 4, 2019. That is, the action parameter model Pm24 is generated by learning using actual performance production data used between July 1 and 2, 2019 and on July 4. Thus, actual performance production data used on July 3, 2019 is not reflected in the action parameter model Pm24.

[0130] The action parameter model Pm25 is generated by learning using the action parameter model Pm24 and actual performance production data used on July 5, 2019. That is, the action parameter model Pm25 is generated by learning using actual performance production data used between July 1 and 2, 2019 and between July 4 and 5. Thus, as with the action parameter model Pm24, actual performance production data used on July 3, 2019 is not reflected in the action parameter model Pm25.​

[0131] Thus, in the present embodiment, the plurality of learning models (i.e., the action parameter models) respectively correspond to mutually different periods. In other words, the plurality of action parameter models are managed in units of time. In this case, the learning unit 104, when performing learning using the actual performance component data contained in the actual performance production data as the teaching data, performs learning on the learning model corresponding to the period in which the actual performance production data was acquired.

[0132] For example, in the example shown in FIG. 10, if the data acquisition unit 108 of the production data generation apparatus 100 acquires actual performance production data in 2011, the learning unit 104 performs learning on the action parameter models Pm13 and Pm14 corresponding to 2011. Further, in the example shown in FIG. 11, if the data acquisition unit 108 acquires actual performance production data on July 3, 2019, the learning unit 104 performs learning on the action parameter model Pm22 or Pm23 corresponding to the period in which the actual performance production data was acquired. Figure 7A Figure 7B

[0133] Figure 8 is a diagram showing an example of the plurality of learning models managed in units of production equipment that are held by the learning model holding unit DB2.

[0134] For example, the plurality of learning models, i.e., the plurality of action parameter models Pm31 to Pm34, held in the learning model holding unit DB2 can each be managed for each production equipment as shown in Figure 8 FIG. 12. Specifically, the action parameter model Pm31 is generated by learning using actual performance production data used in the component mounting line L1. Likewise, the action parameter model Pm32 is generated by learning using actual performance production data used in the component mounting line L2, and the action parameter model Pm33 is generated by learning using actual performance production data used in the component mounting line L3. Further, the action parameter model Pm34 is generated by learning using actual performance production data used in each of all of the component mounting lines L1 to L3.

[0135] ​​Thus, in the present embodiment, the plurality of learning models (i.e., the action parameter models) respectively correspond to mutually different production equipment. In other words, the plurality of action parameter models are managed in units of production equipment. The production equipment can be the component mounting line as described above, or can be a collection of a plurality of component mounting lines. Further, the production equipment can be one or a plurality of component mounting devices, or can be a floor provided with a plurality of component mounting devices or component mounting lines, or can be a factory. In this case, when the learning unit 104 performs learning using actual performance component data contained in actual performance production data as teaching data, the learning unit 104 performs learning on a learning model corresponding to production equipment including a component mounting device M4 or M5 that used the actual performance production data.

[0136] For example, in the example shown in FIG. 6, if the data acquisition unit 108 of the production data generation device 100 acquires actual performance production data from the component mounting line L2, the learning unit 104 performs learning on the action parameter models Pm32 and Pm34 corresponding to the component mounting line L2. Figure 8

[0137] Figure 9A FIG. 7 is a diagram showing an example of a plurality of learning models managed in units of production types, which are held by the learning model holding unit DB2.

[0138] For example, the plurality of learning models, i.e., the plurality of action parameter models Pm41 to Pm44, held in the learning model holding unit DB2 can each be managed for each production type as shown in FIG. 7. The production types are, for example, a trial production type and a mass production type. The trial production type is a type of mounting board produced as a trial product, and the mass production type is a type of mounting board produced as a mass product. In the trial production type, action parameters that place greater emphasis on quality than in the mass production type are set, and in the mass production type, action parameters that place greater emphasis on productivity than in the trial production type are set. Therefore, even if the same mounting board is produced, the action parameters set differ between the trial production type and the mass production type, and thus the learning is performed in a manner of distinguishing by production type, which further improves the estimation accuracy. Specifically, the action parameter model Pm41 is generated by learning using actual performance production data used in production of mounting boards of the trial production type Tl. The action parameter model Pm42 is generated by learning using actual performance production data used in production of mounting boards of a trial production type T2 different from the trial production type Tl. The action parameter model Pm43 is generated by learning using actual performance production data used in production of mounting boards of the mass production type. Further, the action parameter model Pm44 is generated by learning using actual performance production data used in production of mounting boards of all production types. Figure 9A

[0139] Figure 9B ​​is a diagram indicating an example of a plurality of learning models managed by the learning model holding unit DB2 in units of production type and production equipment.

[0140] For example, the plurality of learning models held in the learning model holding unit DB2, i.e., the plurality of action parameter models Pm51 to Pm54, can be managed per combination of each production type and production equipment as shown in Figure 9B In addition, in order to manage per production type, an item that sets the production type can be provided in the component data Dc.

[0141] Specifically, the action parameter model Pm51 is generated by learning using actual performance production data used in production of the trial production type T1's mounting board performed by the component mounting line L1. The action parameter model Pm52 is generated by learning using actual performance production data used in production of the trial production type T2's mounting board performed by the component mounting line L2. The action parameter model Pm53 is generated by learning using actual performance production data used in production of the mass production type's mounting board performed by the component mounting line L3. Further, the action parameter model Pm54 is generated by learning using actual performance production data used in production of all production types' mounting boards performed by all of the component mounting lines L1 to L3.

[0142] As such, in the present embodiment, the plurality of learning models, i.e., the action parameter models, respectively correspond to mutually different production types of mounting boards. In other words, the plurality of learning models are managed in units of production type. In this case, when the learning unit 104 performs learning using actual performance component data contained in actual performance production data as teaching data, the learning unit 104 performs learning on a learning model corresponding to a type of mounting board produced using the actual performance production data.

[0143] For example, in the example shown in Figure 9A If the data acquisition unit 108 of the production data generation apparatus 100 acquires actual performance production data of the trial production type T1, the learning unit 104 performs learning on the action parameter models Pm41 and Pm44 corresponding to the trial production type T1. Further, in the example shown in Figure 9B If the data acquisition unit 108 acquires actual performance production data of the trial production type T2 from the component mounting line L2, the learning unit 104 performs learning on the action parameter models Pm52 and Pm54 corresponding to the trial production type T2 and the component mounting line L2.

[0144] [Outline and flow of processing]

[0145] Figure 10A is a diagram for explaining an outline of the action parameter m's estimation processing in the present embodiment.

[0146] The parameter estimation unit 105 acquires component information d of the component P, for example, from the input / output unit 107. Further, the model selection unit 103 selects one action parameter model Pm from among a plurality of action parameter models Pm held in the learning model holding unit DB2, for example. In addition, the plurality of action parameter models Pm can each be any one of the action parameter models Pm11 to Pm14, Pm21 to Pm25, Pm31 to Pm34, Pm41 to Pm44, and Pm51 to Pm54 illustrated in FIG. 12. The parameter estimation unit 105 estimates an action parameter m of the component mounting device M4 or M5 for mounting the component P indicated by the component information d to the board B, using the acquired component information d and the selected action parameter model Pm. Then, the parameter estimation unit 105 outputs component data Dc including the estimated action parameter m and the component information d. Figure 7A-9B

[0147] For example, in a case where the plurality of action parameter models Pm are managed in units of time as illustrated in FIG. 13, the model selection unit 103 can select one action parameter model Pm corresponding to a manufacturing period of the component P indicated by the component information d. For example, if the manufacturing period is the 1990s, the model selection unit 103 can select the action parameter model Pm11 illustrated in FIG. 14. Further, if the manufacturing period is unknown, the model selection unit 103 can select the action parameter model Pm14 illustrated in FIG. 15. Thereby, a suitable action parameter model Pm for estimating the action parameter m of the component P can be selected. Figure 7A Figure 7A Figure 7A

[0148] Further, in a case where the plurality of action parameter models Pm are managed in chronological order as illustrated in FIG. 16, the model selection unit 103 can select one action parameter model Pm updated on the most recent date. For example, the model selection unit 103 can select the action parameter model Pm25 illustrated in FIG. 17. Further, in a case where a mounting board using a component similar to the component P was mass-produced on July 3, 2019, the model selection unit 103 can select the action parameter model Pm23 illustrated in FIG. 18. Thereby, a suitable action parameter model Pm for estimating the action parameter m of the component P can be selected. Figure 7B Figure 7B Figure 7B

[0149] Further, in a case where the plurality of action parameter models Pm are managed in chronological order as illustrated in FIG. 16, the model selection unit 103 can select one action parameter model Pm updated on the most recent date. For example, the model selection unit 103 can select the action parameter model Pm25 illustrated in FIG. 17. Further, in a case where a mounting board using a component similar to the component P was mass-produced on July 3, 2019, the model selection unit 103 can select the action parameter model Pm23 illustrated in FIG. 18. Thereby, a suitable action parameter model Pm for estimating the action parameter m of the component P can be selected. Figure 8 ​​​​​​​In a case where the production devices are managed as units as illustrated, the model selection unit 103 can select one action parameter model Pm that corresponds to the production device that produces the mounting board using the component P. For example, in a case where the production device that produces the mounting board using the component P is the component mounting line L2, the model selection unit 103 can select the action parameter model Pm32 illustrated. Figure 8 In a case where the production devices are managed as units as illustrated, the model selection unit 103 can select one action parameter model Pm that corresponds to the production device that produces the mounting board using the component P. For example, in a case where the production device that produces the mounting board using the component P is the component mounting line L2, the model selection unit 103 can select the action parameter model Pm32 illustrated. Figure 8 In a case where the production devices are managed as units as illustrated, the model selection unit 103 can select one action parameter model Pm that corresponds to the production device that produces the mounting board using the component P. For example, in a case where the production device that produces the mounting board using the component P is the component mounting line L2, the model selection unit 103 can select the action parameter model Pm32 illustrated.

[0150] In a case where a plurality of action parameter models Pm are managed as units as illustrated, Figure 9A In a case where a plurality of action parameter models Pm are managed as units as illustrated, Figure 9A In a case where a plurality of action parameter models Pm are managed as units as illustrated, Figure 9A In a case where a plurality of action parameter models Pm are managed as units as illustrated,

[0151] In the present embodiment, the model selection unit 103 selects one action parameter model Pm, but is not limited to one, and can select a plurality of action parameter models Pm for estimating mutually different action conditions. For example, the action parameter m includes mutually different parameters such as the speed parameter m3 and the recognition information m4 as illustrated. Figure 5 Therefore, the model selection unit 103 can select, for example, an action parameter model Pm for estimating the speed parameter m3 and an action parameter model Pm for estimating the recognition information m4. In this case, the parameter estimation unit 105 can estimate the speed parameter m3 using the component information d and the action parameter model Pm for the speed parameter m3, and estimate the recognition information m4 using the component information d and the action parameter model Pm for the recognition information m4.

[0152] In addition, the model selection unit 103 can automatically perform the selection of the action parameter model Pm described above, or can perform the selection in correspondence with an operation of the input / output unit 107 by an operator.

[0153] Figure 10B is a diagram for explaining an outline of the learning processing of the action parameter model Pm in the present embodiment.

[0154] The learning unit 104 acquires actual performance production data from any one of the component mounting lines L1 to L3, for example, via the data acquisition unit 108. The actual performance production data includes the actual performance component data Dcu as described above. That is, the learning unit 104 acquires the actual performance component data Dcu. The actual performance component data Dcu is component data Dc used in mounting of the component P to the board B by the component mounting device M4 or M5, and is component data Dc that has been corrected or the like by the use. For example, the actual performance component data Dcu includes the operation parameter mu as the operation parameter m that has been corrected, in which the suction speed as the operation condition is corrected from VI to V2.

[0155] Next, the learning unit 104 selects the operation parameter model Pm corresponding to the actual performance component data Dcu from among the plurality of operation parameter models Pm held in the learning model holding unit DB2. For example, the learning unit 104 selects the operation parameter model Pm corresponding to the period during which the actual performance production data including the actual performance component data Dcu is acquired, as shown in Figure 7A and Figure 7B Alternatively, the learning unit 104 selects the operation parameter model Pm corresponding to the production equipment including the component mounting device M4 or M5 that uses the actual performance production data including the actual performance component data Dcu, as shown in Figure 8 Alternatively, the learning unit 104 selects the operation parameter model Pm corresponding to the production type of the mounted board produced using the actual performance production data including the actual performance component data Dcu, as shown in Figure 9A and Figure 9B

[0156] Then, the learning unit 104 updates the selected operation parameter model Pm by learning using the acquired actual performance component data Dcu as teaching data. That is, the relationship of the component information d and the operation condition represented by the operation parameter model Pm is updated. Thus, the learned operation parameter model Pmu is generated. The learning unit 104 replaces the operation parameter model Pm selected and held in the learning model holding unit DB2 as described above with the learned operation parameter model Pmu. Thus, the learned operation parameter model Pmu is saved in the learning model holding unit DB2 as a new operation parameter model Pm.

[0157] Figure 11 is a diagram showing an example of the processing as a whole in the present embodiment. In the example shown in Figure 11 , the plurality of operation parameter models Pm held in the learning model holding unit DB2 are managed in units of time or in units of production type.

[0158] ​The parameter estimation unit 105 and Figure 10A Similarly to the example shown, the motion parameter model Pm selected by the model selection unit 103 is used to estimate the motion parameter m of the component P, and component data Dc including the component information d of the component P and the motion parameter m is generated. Then, the data generation unit 102 generates production data Dp including the component data Dc, and outputs the production data Dp to, for example, the component mounting line L1 via the input / output unit 107. Figure 11 In the example shown, the production data Dp is output to the component mounting line L1 , but may be output to other component mounting lines L2 or L3 .

[0159] When the component mounting devices M4 and M5 included in the component mounting line L1 receive production data Dp from the input / output unit 107 of the production data generating device 100, they mount at least one component P on a substrate B based on the production data Dp, thereby producing a mounted substrate. At this point, the component mounting line L1 corrects the component data Dc included in the production data Dp, for example, to reduce the defect rate of the mounted substrate. As a specific example, the suction speed V1 included in the operation parameter m of the component data Dc is corrected to V2. As a result, actual performance production data is generated that includes actual performance component data Dcu with the operation parameter mu. The actual performance component data Dcu is stored as new component data Dc in the component library storage unit DB3. Specifically, the data acquisition unit 108 of the production data generating device 100 receives the actual performance production data from the component mounting line L1 and stores the actual performance component data Dcu included in the actual performance production data as new component data Dc in the component library storage unit DB3. Note that the component data Dc in the component mounting line L1 is not necessarily corrected. Without correction, the component mounting line L1 generates actual performance production data including the component data Dc acquired from the production data generating device 100 as actual performance component data Dcu.

[0160] If the data acquisition unit 108 acquires actual production data, the learning unit 104 will Figure 10B Similarly to the example shown, the motion parameter model Pm corresponding to the actual performance component data Dcu included in the actual performance production data is selected from the learning model storage unit DB2. The learning unit 104 then uses the actual performance component data Dcu to learn the selected motion parameter model Pm and stores the learned motion parameter model Pmu in the learning model storage unit DB2.

[0161] Figure 12 FIG is a diagram showing another example of the overall processing in this embodiment. Figure 12In the example shown, the plurality of action parameter models Pm held in the learning model holding unit DB2 are managed in units of production equipment.

[0162] The parameter estimation unit 105 estimates the action parameter m of the component P using the action parameter model Pm selected by the model selection unit 103 for each production equipment. Figure 10A In the example shown, the action parameter m of the component P is estimated using the action parameter model Pm selected by the model selection unit 103 for each production equipment, and component data Dc including the component information d of the component P and the action parameter m is generated.

[0163] For example, the parameter estimation unit 105 estimates the action parameter m and generates component data Dc based on the action parameter model Pm for the component mounting line L1, i.e., the action parameter model Pm31 shown. Figure 8 The data generation unit 102 generates production data Dp including the component data Dc, and outputs the production data Dp to, for example, the component mounting line L1 via the input / output unit 107. Further, the parameter estimation unit 105 estimates the action parameter m and generates component data Dc based on the action parameter model Pm for the component mounting line L2, i.e., the action parameter model Pm32 shown. Figure 8 The data generation unit 102 generates production data Dp including the component data Dc, and outputs the production data Dp to, for example, the component mounting line L2 via the input / output unit 107. Further, the parameter estimation unit 105 estimates the action parameter m and generates component data Dc based on the action parameter model Pm for the component mounting line L3, i.e., the action parameter model Pm33 shown. Figure 8 The data generation unit 102 generates production data Dp including the component data Dc, and outputs the production data Dp to, for example, the component mounting line L3 via the input / output unit 107.

[0164] Alternatively, the parameter estimation unit 105 estimates the action parameter m and generates component data Dc based on the action parameter models Pm for the component mounting lines L1 to L3, i.e., the action parameter model Pm34 shown. Figure 8 The data generation unit 102 generates production data Dp including the component data Dc, and outputs the production data Dp to, for example, the component mounting lines L1 to L3 via the input / output unit 107.

[0165] In each of the component mounting lines L1 to L3, the component mounting devices M4 and M5 install at least one component P to a board B based on the production data Dp if the production data Dp is acquired from the input / output section 107 of the production data generating device 100. Thus, production mounting boards are produced. At this time, in each of the component mounting lines L1 to L3, for example, the component data Dc contained in the production data Dp is corrected so that the failure occurrence rate of the mounting boards is reduced. As a result, actual performance production data containing actual performance component data Dcu is generated. These actual performance component data Dcu are stored as new component data Dc in the component library holding section DB3. Specifically, the data acquisition section 108 of the production data generating device 100 acquires actual performance production data from each of the component mounting lines L1 to L3, and stores the actual performance component data Dcu contained in these actual performance production data as new component data Dc in the component library holding section DB3. In addition, the correction of the component data Dc in each of the component mounting lines L1 to L3 is not necessarily performed. In the case where no correction is performed, each of the component mounting lines L1 to L3 generates actual performance production data containing the component data Dc acquired from the production data generating device 100 as actual performance component data Dcu.

[0166] If the actual performance production data of the component mounting line L1 is acquired by the data acquisition section 108, the learning section 104 selects the action parameter model Pm corresponding to the actual performance component data Dcu contained in this actual performance production data from the learning model holding section DB2, as in the example shown in FIG. 8. Specifically, the learning section 104 selects the action parameter model Pm for the component mounting line L1, i.e., the action parameter model Pm31 shown in FIG. 9. Then, the learning section 104 performs learning with respect to the selected action parameter model Pm using the actual performance component data Dcu of the component mounting line L1. In addition, this actual performance component data Dcu is the component data Dc acquired from the component mounting line L1 by the data acquisition section 108 and stored in the component library holding section DB3 as described above. As a result, the learning section 104 updates the selected action parameter model Pm for the component mounting line L1 stored in the learning model holding section DB2 to a learned action parameter model Pmu for the component mounting line L1. Figure 10B Figure 8

[0167] ​​The learning unit 104 also updates the operation parameter model Pm for each of the component mounting lines L2 and L3, as with the component mounting line LI described above. That is, the learning unit 104 updates the operation parameter model Pm for the component mounting line L2 stored in the learning model holding unit DB2 to the learned operation parameter model Pmu for the component mounting line L2. Further, the learning unit 104 updates the operation parameter model Pm for the component mounting line L3 stored in the learning model holding unit DB2 to the learned operation parameter model Pmu for the component mounting line L3.

[0168] Further, if the actual performance production data for any one of the component mounting lines LI to L3 is acquired by the data acquisition unit 108, the learning unit 104 selects the operation parameter model Pm for all of the component mounting lines LI to L3, that is, the operation parameter model Pm34 shown in FIG. 12. In this case, the learning unit 104 performs learning with respect to the selected operation parameter model Pm using the actual performance component data Dcu for any one of the component mounting lines LI to L3. Note that the actual performance component data Dcu is the component data Dc acquired from any one of the component mounting lines LI to L3 by the data acquisition unit 108 and stored in the component library holding unit DB3 as described above. As a result, the learning unit 104 updates the operation parameter model Pm for the selected component mounting lines LI to L3 stored in the learning model holding unit DB2 to the learned operation parameter model Pmu for the component mounting lines LI to L3. Figure 8

[0169] Further, the data generation unit 102 in the present embodiment can import production data Dp from production equipment of other factories and the like other than the factory having the production system 1, and can export production data Dp to production equipment of other factories and the like.

[0170] Figure 13 is a flowchart showing the processing operation of the production data generation apparatus 100 in the present embodiment.

[0171] The input / output unit 107 of the production data generation apparatus 100 accepts component information d (step Sll). The component information d can be generated and accepted by operation of an operator on the input / output unit 107, or can be accepted by selection from a plurality of component information d. Further, the input / output unit 107 can select component data Dc having a default operation parameter m from a plurality of component data Dc included in the component library, and extract component information d from the component data Dc, thereby accepting the component information d. Note that the component information d includes, for example, size data d2 and component attribute d31 and the like as shown in FIG. 10. Figure 5

[0172] ​​Next, the model selection unit 103 selects at least one action parameter model Pm from among the plurality of action parameter models Pm held in the learning model holding unit DB2 (step S12).

[0173] Next, the parameter estimation unit 105 estimates an action parameter m based on the at least one action parameter model Pm selected in step S12 and the component information d received in step Sll (step S13). The action parameter m is an action condition of the component mounting device M4 or M5 for mounting the component P determined based on the component information d to the board B. Then, the parameter estimation unit 105 generates component data Dc having the component information d and the action parameter m (step S14).

[0174] Next, the data generation unit 102 generates production data Dp including the component data Dc generated in step S14 (step S15). Then, the data generation unit 102 outputs the production data Dp to the component mounting lines Ll to L3, respectively. That is, the component mounting lines Ll to L3 each download the production data Dp from the data generation unit 102, and start production of mounting boards using the production data Dp (step S16).

[0175] Then, the learning unit 104 performs relearning of the action parameter model Pm using the component data Dc (i.e., actual performance component data Dcu) included in the production data Dp used in each of the component mounting lines Ll to L3 as teaching data (step S17). The action parameter model Pm to be the object of the relearning is, for example, an action parameter model Pm corresponding to a period in which the used production data Dp (i.e., actual performance production data) is obtained, and the like.

[0176] As described above, in the production data generation apparatus 100 in the present embodiment, at least one action parameter model Pm is selected from among a plurality of action parameter models Pm. Then, based on the at least one selected action parameter model Pm and component information d of a mounting target component P, an action parameter m for mounting the mounting target component P to a board B is estimated.

[0177] Thus, since at least one action parameter model Pm is selected from among a plurality of action parameter models Pm and used in estimation of an action parameter m, it is possible to improve the likelihood of estimating a suitable action parameter m for a mounting target component P. Therefore, it is possible to set a suitable action parameter m. Further, since component data Dc having such an action parameter m and component information d is included in production data Dp, in a case where the production data Dp is used in mounting of a component P to a board B by a component mounting device M4 or M5, it is possible to produce a mounted board of good quality. That is, it is possible to seek quality improvement of a mounted board.

[0178] Furthermore, the production data generation device 100 of this embodiment acquires actual production data including actual component data Dcu used by the component mounting device M4 or M5. Then, using this actual component data Dcu as teaching data for learning, the motion parameter model Pm corresponding to the acquired actual production data among the plurality of motion parameter models Pm is updated.

[0179] The motion parameters mu in the actual performance component data Dcu included in the actual performance production data are used during the installation of the mounted component P, and are corrected at this time. Specifically, these motion parameters mu are corrected to produce higher-quality mounted substrates. Therefore, by using the actual performance component data Dcu with these motion parameters mu as teaching data for learning the motion parameter model Pm, further optimization of the motion parameter model Pm can be achieved. Consequently, when this motion parameter model Pm is selected by the model selection unit 103, the estimation accuracy of the motion parameters m can be improved.

[0180] Furthermore, in the production data generating device 100 of the present embodiment, a plurality of operation parameter models Pm are associated with different periods, and the operation parameter model Pm corresponding to the period in which the actual performance production data is acquired is learned.

[0181] For example, Figure 7A As shown, one of the motion parameter models Pm11-Pm14 corresponds to the entire period (e.g., the entire period from 1990 to the present). Furthermore, the remaining motion parameter models Pm11-Pm13 correspond to different eras. Examples of different eras include the 1990s, 2000s, and 2010s. Thus, a motion parameter model Pm corresponding to the entire period or a specific era is selected from these motion parameter models Pm11-Pm14 and used to estimate the motion parameter m. This allows for estimation of the appropriate motion parameter m for the target component P, corresponding to the period.

[0182] In addition, in the production data generating device 100 of this embodiment, as Figure 8 As shown, the motion parameter models Pm31 to Pm34 are associated with different production equipment. Then, the motion parameter model Pm corresponding to the production equipment including the component mounting device M4 or M5 using the actual production data is learned.

[0183] Thus, the motion parameter model Pm corresponding to all or any one of the component mounting lines is selected from the motion parameter models Pm31 to Pm34 and used to estimate the motion parameter m. Therefore, the appropriate motion parameter m for the mounting target component P according to the production equipment can be estimated.

[0184] Further, in the production data generation apparatus 100 in the present embodiment, as shown in Figure 9A , the operation parameter models Pm41 to Pm44 are respectively established in correspondence with mutually different types of mounting boards. Then, the operation parameter model Pm corresponding to the type of mounting board produced using actual performance production data is learned.

[0185] Thus, the operation parameter model Pm corresponding to, for example, a mass production type or a trial production type is selected from among the operation parameter models Pm41 to Pm44 and used in the estimation of the operation parameter m. Therefore, the appropriate operation parameter m corresponding to the type of mounting board can be estimated for the mounting target component P.

[0186] As such, in the present embodiment, the operation parameter model Pm specialized for the period, the production equipment, or the type of mounting board, etc. is used, and as a result, the appropriate operation parameter m for that period, production equipment, or type of mounting board can be estimated.

[0187] (Modified Example of Embodiment 1)

[0188] In the above-described embodiments, as shown in Figure 7A-9B , a plurality of operation parameter models Pm are managed in units of time, production equipment, or production type. However, the manner of management is not limited to these, and a plurality of operation parameter models Pm can be managed in other units. Further, in the example shown in Figure 9B , a plurality of operation parameter models Pm are managed in units of a combination of production equipment and production type, but the combination is not limited to this and can be any combination.

[0189] Further, in the example shown in Figure 7B in the above-described embodiments, the model selection section 103 selects one operation parameter model Pm updated on the most recent date. Here, in the case where the failure rate of the mounting board produced using production data Dp including the operation parameter m estimated based on this operation parameter model Pm is high, the model selection section 103 can reselect the operation parameter model Pm previously selected as an operation parameter model Pm updated before the most recent date. Further, in the examples shown in Figure 7A , Figure 8 , Figure 9A , and Figure 9B , the model selection section 103 can also reselect the operation parameter model Pm, for example, in correspondence with the failure rate. The reselection can be performed randomly or in accordance with a predetermined order.

[0190] Further, the input / output section 107 of the production data generation apparatus 100 in the above-described embodiment can export at least one action parameter model Pm held in the learning model holding section DB2 to another facility other than the facility having the production system 1. The facility can be a factory, or a floor. Furthermore, the input / output section 107 can import at least one action parameter model Pm from another facility and store it in the learning model holding section DB2. Thereby, further optimization of the action parameter model Pm can be sought. Further, the input / output section 107 can perform import and export of the production data Dp held in the production data holding section DB1, and can perform import and export of the component data Dc held in the component library holding section DB3.

[0191] Further, the learning section 104 of the production data generation apparatus 100 in the above-described embodiment performs learning of the action parameter model Pm corresponding to the actual performance production data acquired by the data acquisition section 108, among the plurality of action parameter models Pm held in the learning model holding section DB2. However, the learning section 104 can also switch the action parameter model Pm to be set as the learning target by an operation of the operator accepted by the input / output section 107. Thereby, learning of the action parameter model Pm specified by the operator is performed.

[0192] Further, the action parameter m estimated by the parameter estimation section 105 can be associated with the identification information of the action parameter model Pm used in the estimation of the action parameter m and the date and time of the estimation, which are included in the component data Dc of the component library. Thereby, the action parameter m can be appropriately managed.

[0193] Further, the parameter estimation section 105 can not use all the information included in the component information d shown in FIG. 6, but can estimate the action parameter m using only a part of the information. For example, the input / output section 107 can accept a part of the information included in the component information d used in the estimation of the action parameter m, in correspondence with an operation of the operator. In a case where such a part of the information is accepted, the parameter estimation section 105 estimates the action parameter m using only the accepted part of the information. Furthermore, the parameter estimation section 105 can not estimate all the parameters included in the action parameter m shown in FIG. 6, but can estimate only a part of the parameters. For example, the input / output section 107 can accept a designation of a part of the parameters included in the action parameter m to be estimated, in correspondence with an operation of the operator. In a case where such a part of the parameters is designated, the parameter estimation section 105 estimates only the designated part of the parameters included in the action parameter m. Further, the parameter estimation section 105 can perform principal component analysis on all the information included in the component information d, and estimate the action parameter m in correspondence with the analysis result. Figure 5 Figure 5 Figure 5 ​​​

[0194] (Embodiment 2)

[0195] In this embodiment, filtering of the operation parameter mu contained in the actual performance production data output from each of the component mounting lines L1 to L3 is performed.

[0196] [Production system]

[0197] Figure 14 is a diagram showing an example of the structure of the production system in this embodiment.

[0198] The production system 2 in this embodiment is provided with three component mounting lines L1 to L3, a production management device 100a, a data management device 300, and three inspection devices 401 to 403. That is, the production system 2 in this embodiment has: component mounting devices M4 and M5 that produce mounting boards by mounting components P on boards B; and a data management device 300 or a processing device such as the inspection devices 401 to 403 that performs processing related to the production of the mounting boards.

[0199] In addition, regarding the constituent elements among the constituent elements in this embodiment that are the same as in Embodiment 1, the same reference numerals are assigned as in Embodiment 1, and detailed description is omitted.

[0200] The three component mounting lines L1 to L3 are the same as the three component mounting lines L1 to L3 of the production system 1 in Embodiment 1.

[0201] The production management device 100a manages the production of mounting boards in the production system 2. Specifically, the production management device 100a is provided with the same functions as the production data generation device 100 in Embodiment 1, and further has a function of filtering the operation parameter mu contained in the actual performance production data.

[0202] The data management device 300 is connected to the production management device 100a and the component mounting lines L1 to L3, respectively, and manages the production data Dp of each of the component mounting lines L1 to L3. The production data Dp can be actual performance production data used in each of the component mounting lines L1 to L3. Further, the data management device 300 in this embodiment generates filtering information used in the filtering of the production management device 100a, and outputs the filtering information to the production management device 100a.

[0203] The inspection devices 401 to 403 perform inspection of the mounting boards produced by the component mounting lines L1 to 3, respectively. That is, the inspection device 401 inspects the mounting boards of the component mounting line L1, the inspection device 402 inspects the mounting boards of the component mounting line L2, and the inspection device 403 inspects the mounting boards of the component mounting line L3. Further, the inspection devices 401 to 403 in the present embodiment are each connected to the production management device 100a, and generate the above-mentioned screening information based on the inspection results of the mounting boards, and output the screening information to the production management device 100a.

[0204] In addition, the data management device 300 and the inspection devices 401 to 403 in the present embodiment are each a processing device that performs processing related to production of mounting boards.

[0205] [Functional Configuration of Production Management Device, Component Mounting Line, and Processing Device]

[0206] Figure 15 is a block diagram showing the functional configuration of each of the production management device 100a, the component mounting lines L1 to L3, and the processing device. In the present embodiment, as shown in Figure 15 the processing device 500 is constituted by the data management device 300 and the inspection devices 401 to 403.

[0207] The production management device 100a, like the production data generation device 100 of Embodiment 1, is provided with the control section 101, the data generation section 102, the model selection section 103, the learning section 104, the parameter estimation section 105, the display section 106, the input / output section 107, the data acquisition section 108, the production data holding section DB1, the learning model holding section DB2, and the component library holding section DB3. Further, the production management device 100a is provided with a screening section 109 that screens the action parameters mu included in the actual performance production data.

[0208] Specifically, the data acquisition section 108 of the production management device 100a in the present embodiment acquires, from the component mounting lines L1 to L3, production data Dp, i.e., actual performance production data, used in production of mounting boards by the component mounting devices M4 or M5, respectively. The actual performance production data includes, for at least one kind of component P, an action condition, i.e., an action parameter mu, of the component mounting device M4 or M5 for mounting the component P to a board B. Further, the data acquisition section 108 acquires screening information from the processing device 500.

[0209] The screening section 109 selects one or more action parameters mu by performing screening of at least one action parameter mu included in the acquired actual performance production data using the screening information obtained from the processing device 500.

[0210] The learning unit 104 generates or updates the learning model, i.e., the movement parameter model Pm, held in the learning model holding unit DB2, by learning using the selected one or more movement parameters mu as teaching data. The movement parameter model Pm represents the relationship between the movement condition of the component mounting device M4 or M5 for mounting the component P to the board B and the component P. In the above learning, specifically, the actual performance component data Dcu including the selected one or more movement parameters mu is used as teaching data for each of the selected one or more movement parameters mu.

[0211] Further, the parameter estimation unit 105 in the present embodiment, like the embodiment 1, estimates the movement condition, i.e., the movement parameter m, of the component mounting device M4 or M5 for mounting the yet-to-be-mounted mounting target component P to the board. The estimation of the movement parameter m is based on the movement parameter model Pm held in the learning model holding unit DB2 and the component information d related to the mounting target component P mounted to the board B.

[0212] The processing device 500 is provided with the inspection devices 401 to 403 and the data management device 300.

[0213] The inspection device 401 is provided with an inspection control unit 411, an input / output unit 412, a display unit 413, an inspection mechanism 414, and an inspection data holding unit DB5.

[0214] The input / output unit 412, for example, accepts input data based on the operation of the operator of the production system 2 and outputs the input data to the inspection control unit 411. Such an input / output unit 412 can have, for example, a keyboard, a touch sensor, a touch pad, a mouse, or the like. Further, the input / output unit 412 performs output of data to the production management device 100a and input of data from the production management device 100a.

[0215] The inspection mechanism 414 is composed of a mechanism including, for example, a camera or the like for inspecting the mounting board and stores inspection data indicating the inspection result in the inspection data holding unit DB5.

[0216] The inspection data holding unit DB5 is a recording medium for holding inspection data. Such a recording medium is, for example, a hard disk, a RAM, a ROM, a semiconductor memory, or the like. Further, such a recording medium can be volatile or non-volatile.

[0217] The display unit 413 displays the inspection data held in the inspection data holding unit DB5 and the like. Specific examples of the display unit 413 include a liquid crystal display, a plasma display, an organic EL display, or the like, but are not limited to these.

[0218] The inspection control section 411 controls the input / output section 412, the display section 413, the inspection mechanism 414, and the inspection data holding section DB5, respectively. For example, the inspection control section 411 causes the inspection mechanism 414 to start the inspection of the mounting board in response to the operation of the operator accepted by the input / output section 412. Further, the inspection control section 411 in the present embodiment generates the screening information and outputs it to the data acquisition section 108 of the production management apparatus 100a via the input / output section 412.

[0219] The inspection apparatuses 402 and 403 also have the same structure as the above-described inspection apparatus 401.

[0220] The data management apparatus 300 is provided with a data control section 311, an input / output section 312, a display section 313, and a data holding section DB6.

[0221] The input / output section 312, for example, accepts input data based on the operation of the operator of the production system 2 and outputs the input data to the data control section 311. Such an input / output section 312 can have, for example, a keyboard, a touch sensor, a touch pad, a mouse, or the like. Further, the input / output section 312 performs output of data to and input of data from the production management apparatus 100a and the component mounting lines L1 to L3.

[0222] The data holding section DB6 is a recording medium for holding data. The data is, for example, screening information. Further, such a recording medium can be a hard disk, a RAM, a ROM, a semiconductor memory, or the like, and can be volatile or non-volatile.

[0223] The display section 313 displays data and the like held in the data holding section DB6. A specific example of the display section 313 is a liquid crystal display, a plasma display, an organic EL display, or the like, but is not limited to these.

[0224] The data control section 311 controls the input / output section 312, the display section 313, and the data holding section DB6, respectively. Further, the data control section 311 in the present embodiment can generate screening information similarly to the above-described inspection control section 411 and output it to the data acquisition section 108 of the production management apparatus 100a via the input / output section 312.

[0225] For example, the data control section 311 in the present embodiment can generate substrate identification information for identifying the substrate B as the screening information. In this case, the plurality of actual performance production data Dpu corresponding to the plurality of different types of mounting substrates are screened by the screening information. Therefore, the data management apparatus 300 can be said to be an apparatus that manages a plurality of actual performance production data Dpu corresponding to a plurality of different types of mounting substrates. Alternatively, the data control section 311 can generate one or more component identification information for identifying the type of the component P as the screening information.

[0226] [Outline of the processing]

[0227] Figure 16 is a diagram showing an example of the overall processing in the present embodiment.

[0228] In the present embodiment, as in Embodiment 1, at least one production data Dp is generated, and actual performance production data Dpu is output from the component mounting lines L1 to L3 based on the at least one production data Dp, respectively.

[0229] The screening section 109 of the production management apparatus 100a performs screening with respect to at least one action parameter mu included in the actual performance production data Dpu. At this time, the screening section 109 acquires screening information Df from the processing apparatus 500, and performs screening based on the screening information Df, thereby selecting one or more action parameters mu. Then, the screening section 109 stores actual performance component data Dcu including the selected one or more action parameters mu as new component data Dc in the component library holding section DB3, respectively.

[0230] [Screening information]

[0231] Figure 17A is a diagram showing an example of the screening information Df generated based on the inspection result of the mounting substrate.

[0232] For example, the inspection control section 411 of each of the inspection apparatuses 401 to 403 included in the processing apparatus 500 generates the screening information Df shown in Figure 17A and outputs it to the screening section 109.

[0233] That is, the inspection control section 411 generates information indicating a quality index of each of at least one type of component P mounted to the mounting substrate as the screening information Df, by the inspection of the mounting substrate performed by the inspection mechanism 414. The quality index indicated by the screening information Df is also called a mounting quality index, and for example, a larger value is shown as the mounting state of the component P corresponding to the quality index is better.

[0234] More specifically, when the inspection mechanism 414 includes a camera, the inspection control unit 411 calculates an installation quality index representing the positional deviation of the installed component P based on the image of the mounting substrate captured by the camera. The positional deviation of component P is the difference between the mounting position of component P on substrate B, as represented by the image, and the assembly coordinates (or mounting position) of component P, as represented by the production data Dp. For example, the inspection control unit 411 calculates the installation quality index as a value closer to 1 as the smaller the positional deviation of component P is, and conversely, calculates the installation quality index as a value closer to 0 as the larger the positional deviation of component P is. In other words, the installation quality index can be normalized to a value ranging from 0 to 1. Furthermore, the installation quality index can also be referred to as a score or evaluation value.

[0235] By calculating the installation quality index in this way, the inspection control unit 411 Figure 17A As shown, filtering information Df is generated that indicates the mounting quality index for each of the multiple types of parts P. This filtering information Df indicates the mounting quality index for each part P by part name and part code. For example, filtering information Df indicates "0.95" as the mounting quality index for the type of part P identified by part name "Part A" and part code "C001."

[0236] If the screening unit 109 of the production management device 100a obtains Figure 17A The filtering information Df shown in FIG. 1 is used to perform filtering. That is, the filtering unit 109 selects one or more action parameters mu corresponding to the type of component P whose mounting quality index is above the threshold value through this filtering. For example, the filtering unit 109 selects one or more action parameters mu corresponding to the type of component P whose mounting quality index is above the threshold value "0.85". Figure 17A In the example shown, the filtering unit 109 selects the action parameter mu corresponding to the component P with the component name "Part A" and the component code "C001," and the action parameter mu corresponding to the component P with the component name "Part G" and the component code "C034." Specifically, the filtering unit 109 selects the actual performance component data Dcu with the component code "C001" and the actual performance component data Dcu with the component code "C034" from the plurality of actual performance production data Dpu.

[0237] Thus, in this embodiment, the motion parameters mu corresponding to the types of components P with a high mounting quality index are selected through screening. Thus, one or more motion parameters mu corresponding to the types of components P with good mounting conditions are selected through screening and used for learning, while motion parameters mu corresponding to the types of components P with poor mounting conditions are not used in learning. Consequently, a motion parameter model Pm can be generated for estimating appropriate motion parameters m for achieving a good mounting condition.

[0238] Figure 17B This is a diagram showing an example of filtering information Df generated based on the actual mounting performance of the component mounting lines L1 to L3.

[0239] In addition, the data control unit 311 of the data management device 300 included in the processing device 500 can generate Figure 17B The filtering information Df shown is output to the filtering unit 109.

[0240] Specifically, the data control unit 311 obtains information indicating the operating status of the component mounting devices M4 and M5 included in each component mounting line from component mounting lines L1 to L3 via the input / output unit 312. Based on this operating status information, the data control unit 311 generates information indicating an actual mounting performance index for each of at least one component type as filtering information. The actual mounting performance index is an index related to errors incurred in component mounting devices M4 and M5 due to operations based on actual performance production data Dpu of component mounting devices M4 and M5. For example, the fewer errors, the smaller the value. Furthermore, the actual mounting performance index can also be referred to as a score or evaluation value.

[0241] More specifically, the mounting performance index is an index related to errors caused by the component mounting devices M4 and M5, such as missed pickup of components P, dropped components P, or missed supply from the feeder 7 to the mounting head 10. For example, the data control unit 311 expresses this mounting performance index as a percentage. Specifically, the data control unit 311 calculates the mounting performance index as a value closer to 0% as the number of errors decreases, and conversely, as a value closer to 100% as the number of errors increases.

[0242] By calculating the actual performance index in this way, the data control unit 311 Figure 17B As shown, filtering information Df is generated that indicates the actual installation performance index for each of the multiple types of parts P. This filtering information Df indicates the actual installation performance index for each part P by part name and part code. For example, filtering information Df indicates "0.5%" as the installation quality index for a part P of the type specified by part name "Part A" and part code "C001."

[0243] If the screening unit 109 of the production management device 100a obtains Figure 17BThe selection information Df shown is used for the selection. That is, the selection unit 109 selects one or more action parameters mu corresponding to the kind of the component P having the installation actual performance index below the threshold value by the selection. For example, the selection unit 109 selects one or more action parameters mu corresponding to the kind of the component P having the installation actual performance index below the threshold value "1%" by the selection. In Figure 17B In the example shown, the selection unit 109 selects the action parameters mu corresponding to the component P of the component name "A component" and the component code "C001", and the action parameters mu corresponding to the component P of the component name "B component" and the component code "C102". That is, the selection unit 109 selects the actual performance component data Dcu of the component code "C001" and the actual performance component data Dcu of the component code "C102" from the plurality of actual performance production data Dpu.

[0244] As such, in the present embodiment, the action parameters mu corresponding to the kind of the component P having a small installation actual performance index are selected by the selection. Thereby, one or more action parameters mu corresponding to the kind of the component P having a small error in the component mounting device M4 or M5 are selected by the selection and used in the learning, and the action parameters mu corresponding to the kind of the component P having a large error are not used in the learning. Therefore, the action parameter model Pm used to estimate the appropriate action parameter m for reducing the occurrence of the error can be generated.

[0245] Figure 18A is a diagram showing an example of the selection information Df generated by the selection of the component P.

[0246] The data control unit 311 of the data management device 300 included in the processing device 500 can correspond to Figure 18A The selection information Df shown in (b) is generated from the operation result of the operator shown in (a) by the data control unit 311 of the data management device 300 included in the processing device 500, and output to the selection unit 109.

[0247] For example, the data control unit 311 outputs the selection information Df to the selection unit 109. Figure 18AThe component selection screen shown in (a) is displayed on the display unit 313. The component selection screen shows the component name, component code, and at least a part of the component information d (such as the external dimensions and the number of leads) of each component P processed by the production system 2. The operator inputs a learning mark for the desired component P by operating the input-output unit 312 while viewing the component selection screen. For example, in a mounting substrate produced by the installation of component P, when the quality of the installation state of component P is good, the operator inputs a learning mark for component P. Alternatively, when component P used in the past is an exceptional special component, the operator does not input a learning mark for component P. In addition, when the action parameter mu of component P contained in the actual performance production data Dpu is set in other factories, the operator does not input a learning mark for component P. Figure 18A In the example shown in (a) of FIG, the operator inputs a learning mark for each of the parts P named "Part A," "Part B," "Part D," and "Part F." The operator then further operates the input / output unit 312 to select a decision button displayed on the part selection screen. As a result, the data control unit 311 generates a data file corresponding to the learning mark input to the part selection screen. Figure 18A The filtering information Df shown in (b) is shown.

[0248] Filtering information Df indicates, for each learning mark input by the operator, the part name and part code of the part P corresponding to the learning mark, as part identification information for the part P. For example, filtering information Df indicates the part name "Part A" and part code "C001," the part name "Part B" and part code "C002," the part name "Part D" and part code "C003," and the part name "Part F" and part code "C005" as part identification information for each of four parts P. In this manner, the data management device 300 outputs filtering information Df including one or more pieces of part identification information for identifying the type of each part P.

[0249] If the screening unit 109 of the production management device 100a obtains Figure 18A The filtering information Df shown in (b) is used for filtering. That is, the filtering unit 109 selects the action parameter mu corresponding to the type of the component P identified by the component identification information for each of the one or more component identification information indicated by the filtering information Df. Figure 18AIn the example shown in (b), the screening unit 109 selects the actual performance component data Dcu of the component code “C001”, the actual performance component data Dcu of the component code “C002”, the actual performance component data Dcu of the component code “C003”, and the actual performance component data Dcu of the component code “C005” from the plurality of actual performance production data Dpu.

[0250] Thus, in this embodiment, the motion parameters mu corresponding to the type of component P specified by the operator are selected through screening. Thus, one or more motion parameters mu corresponding to the type of component P identified by the component identification information specified by the operator are selected through screening and used for learning, while motion parameters mu corresponding to other types of component P are not used for learning. Therefore, a motion parameter model Pm can be generated for estimating appropriate motion parameters mu for a specific component P.

[0251] Figure 18B FIG. 1 is a diagram showing an example of filtering information Df generated by selection of substrate B. FIG.

[0252] The data control unit 311 of the data management device 300 included in the processing device 500 may also correspond to Figure 18B The filtering information Df shown in (b) is generated based on the operation result of the operator shown in (a) and is output to the filtering unit 109.

[0253] For example, the data control unit 311 Figure 18B The substrate selection screen shown in (a) is displayed on the display unit 313. The substrate name, substrate code, and auxiliary information of each substrate B processed by the production system 2 are displayed on the substrate selection screen. The operator inputs a learning mark for the desired substrate B by operating the input-output unit 312 while viewing the substrate selection screen. For example, the operator inputs a learning mark for each substrate B named "A substrate", "B substrate", "D substrate", and "F substrate". Then, the operator selects the decision button shown in the substrate selection screen by further operating the input-output unit 312. As a result, the data control unit 311 generates a learning mark corresponding to the learning mark input to the substrate selection screen. Figure 18B The filtering information Df shown in (b) is shown.

[0254] The screening information Df shows the substrate name and the substrate code and the like of the substrate B corresponding to each learning mark input by the operator as the substrate identification information of the substrate B. For example, the screening information Df shows the substrate name "A substrate" and the substrate code "B001", the substrate name "B substrate" and the substrate code "B002", the substrate name "D substrate" and the substrate code "B004", and the substrate name "F substrate" and the substrate code "B006" as the substrate identification information of the respective four substrates B. In this way, the data management device 300 outputs the screening information Df including one or more substrate identification information for identifying the kind of the substrate B, respectively.

[0255] The screening section 109 of the production management device 100a performs screening using the screening information Df shown in (b) if the screening information Df is acquired. Figure 18B That is, in a case where the actual performance production data group is constituted by the plurality of actual performance production data Dpu acquired by the data acquisition section 108, the screening section 109 selects one or more action parameters mu from at least one action parameter mu included in the actual performance production data group in the screening. These selected one or more action parameters mu correspond to the kind of the component P mounted on the substrate B of the kind identified from the substrate identification information included in the screening information Df, respectively.

[0256] In the example shown in (b) of the production data Dp and the actual performance production data Dpu, the substrate code of the substrate B used in the production of the mounting substrate can be shown. In this case, the screening section 109 selects the actual performance production data Dpu showing the substrate codes "B001", "B002", "B004", and "B006", respectively, from the actual performance production data group, and extracts the actual performance component data Dcu from these selected actual performance production data Dpu. Figure 18B

[0257] In this way, in the present embodiment, the action parameter mu corresponding to the kind of the component P mounted on the substrate B specified from the operator is selected by the screening. That is, one or more action parameters mu corresponding to the kind of the component P mounted on the substrate B specified from the operator are selected by the screening and used in the learning, and the action parameters mu corresponding to the kind of the component P mounted on other substrates B are not used in the learning. Therefore, the action parameter model Pm for estimating the appropriate action parameter mu for the specific substrate B can be generated.

[0258] [Processing Flow] ​

[0259] Figure 19 is a flowchart showing the processing action of the production management apparatus 100a in the present embodiment.

[0260] The data acquisition section 108 of the production management apparatus 100a acquires actual performance production data Dpu from the component mounting lines L1 to L3 (step S21). Further, the data acquisition section 108 acquires screening information Df from the processing apparatus 500 (step S22).

[0261] Next, the screening section 109 screens each action parameter mu included in the actual performance production data Dpu acquired in step S21 using the screening information Df acquired in step S22 (step S23). Thus, the action parameters mu used in learning are selected from the actual performance production data Dpu.

[0262] Next, the learning section 104 generates or updates the action parameter model Pm by learning using the action parameters mu selected by the screening of step S23 (step S24). In this learning, the selected action parameters mu and the component information d included in the actual performance component data Dcu together with the action parameters mu are used as teaching data. Further, in the update of the action parameter model Pm, the action parameter model Pm corresponding to the actual performance production data Dpu held in the learning model holding section DB2 is updated.

[0263] Then, in a case where the selected component P is one for which the action parameter m is not determined by the input / output section 107, the parameter estimation section 105 estimates the action parameter m of the selected component P using the action parameter model Pm generated or updated in step S24 (step S25).

[0264] As described above, in the production management apparatus 100a in the present embodiment, actual performance production data Dpu is acquired, and screening is performed on at least one action parameter mu included in the acquired actual performance production data Dpu. That is, one or more action parameters mu are selected from the actual performance production data Dpu using the screening information Df obtained from the processing apparatus 500. Then, the generation or update of the action parameter model Pm is performed by learning using the selected one or more action parameters mu as teaching data.

[0265] Thus, since one or more action parameters mu selected by the screening are used in the learning, and the unselected action parameters mu are not used in the learning, it is possible to seek the proper action parameter model Pm. As a result, if this action parameter model Pm is used, it is possible to estimate the appropriate action parameter mu, which is set to the production data Dp used later in the component mounting device M4 or M5. Thus, it is possible to produce a mounted board of good quality. That is, it is possible to seek the quality improvement of the mounted board.

[0266] That is, in the present embodiment, it is possible to control the action parameter mu used as the teaching data in the learning, and to perform the learning expected by the user.

[0267] Further, in the production management device 100a in the present embodiment, the action parameter m for mounting the mounting target component P to the board B is estimated based on the generated or updated action parameter model Pm and the component information d. Thus, it is possible to estimate the appropriate action parameter m for the mounting target component P, and to set it.

[0268] (Embodiment 2)

[0269] In the above-described embodiments, as shown in Figure 18A and Figure 18B , the screening based on the kind of the specified component P or the kind of the board B is performed, but the screening based on the product series of the mounted board can be performed. Further, the screening based on the date at which the actual achievement production data Dpu is acquired can be performed. For example, it is possible to select only the action parameters mu contained in the actual achievement production data Dpu acquired on the latest date by the screening. Further, the screening based on the specified component mounting line can be performed. For example, in the case where the component mounting line L1 is specified, it is possible to select only the action parameters mu contained in the actual achievement production data Dpu acquired from this component mounting line L1 by the screening.

[0270] Further, in the examples shown in Figure 17A and Figure 17B , the screening information Df indicates the mounting quality index or the mounting actual achievement index, but these are examples, and other indexes can be indicated. Further, these indexes can be indicated in PPM (parts per million).

[0271] Further, in the present embodiment, at least one action parameter model Pm can be selected from the plurality of action parameter models Pm as in Embodiment 1. That is, the action parameter model Pm held in the learning model holding section DB2 can be only one.

[0272] (Other Modification)

[0273] The production data generation apparatus and the production management apparatus and the like related to one or more of the modes are described above based on the embodiments and modifications thereof, but the present disclosure is not limited to these embodiments and modifications thereof. As long as the gist of the present disclosure is not deviated from, a mode obtained by applying various modifications thought by those skilled in the art to each of the embodiments or the modifications thereof, a mode constructed by combining the constituent elements in each of the embodiments and the modifications thereof can also be included in the scope of the present disclosure.

[0274] In addition, in each of the above embodiments and the modifications thereof, each of the constituent elements can be constituted by a dedicated hardware or realized by executing a software program suitable for each of the constituent elements. Each of the constituent elements can be realized by a program execution unit such as a CPU (Central Processing Unit) or a processor reading out and executing a software program recorded in a recording medium such as a hard disk or a semiconductor memory. Here, the software of the apparatus and the like realizing each of the above embodiments and the modifications thereof is a program causing a computer to execute each of the steps included in the flowcharts shown in the above embodiments and the modifications thereof. Figure 13 Or Figure 19 the program of each of the steps included in the flowcharts shown in the above embodiments and the modifications thereof.

[0275] In addition, the following cases are also included in the present disclosure.

[0276] (1) Each of the above apparatuses is specifically a computer system constituted by a microprocessor, a ROM, a RAM, a hard disk assembly, a display assembly, a keyboard, a mouse, and the like. A computer program is stored in the RAM or the hard disk assembly. Each of the apparatuses achieves its function by the microprocessor acting in accordance with the computer program. Here, the computer program is constituted by combining a plurality of command codes indicating instructions for a computer in order to achieve a given function.

[0277] (2) A part or all of the constituent elements constituting each of the above apparatuses can be constituted by one system LSI (Large Scale Integration). The system LSI is a super multifunctional LSI in which a plurality of structural parts are integrated on one chip, and is specifically a computer system including a microprocessor, a ROM, a RAM, and the like. A computer program is stored in the RAM. The system LSI achieves its function by the microprocessor acting in accordance with the computer program.

[0278] (3) A part or all of the components that constitute each of the above-described devices can be constituted by an IC card or a module that can be attached to and detached from each of the devices. The IC card or the module is a computer system constituted by a microprocessor, a ROM, a RAM, and the like. The IC card or the module can include the above-described super multifunction LSI. The IC card or the module achieves its function by acting in accordance with a computer program by the microprocessor. The IC card or the module can have tamper resistance.

[0279] (4) The present disclosure can be the above-described methods. In addition, the methods can be provided as a computer program implemented by a computer, and can be provided as a digital signal constituted by the computer program.

[0280] In addition, the present disclosure can be provided as a scheme in which the computer program or the digital signal is recorded in a computer-readable recording medium such as a flexible disc, a hard disc, a CD-ROM, an MO, a DVD, a DVD-ROM, a DVD-RAM, a BD (Blu-ray (registered trademark) Disc), a semiconductor memory, and the like. In addition, the present disclosure can be provided as the digital signal recorded in the recording medium.

[0281] In addition, the present disclosure can be provided as a scheme in which the computer program or the digital signal is transmitted via an electric communication line, a wireless or wired communication line, a network typified by the Internet, a data broadcast, and the like.

[0282] In addition, the present disclosure can be a computer system provided with a microprocessor and a memory that stores the above-described computer program, the microprocessor acting in accordance with the computer program.

[0283] In addition, the program or the digital signal can be implemented by an independent other computer system by being recorded in the recording medium and being transferred, or by being transferred via the network or the like.

[0284] (5) The above-described embodiments and the above-described modified examples can be combined, respectively.

[0285] Industrial applicability

[0286] The present disclosure can be utilized in a system for producing a mounting substrate by mounting components on a substrate, and the like.

[0287] Explanation of reference numerals

[0288] 1, 2 Production system

[0289] 7 Feeder

[0290] 10 Mounting head

[0291] 10a adsorption assembly

[0292] 10b adsorption nozzle

[0293] 11 component recognition camera

[0294] 12 substrate recognition camera

[0295] 14 component carrier tape

[0296] 100 production data generation device

[0297] 100a production management device

[0298] 101 control section

[0299] 102 data generation section

[0300] 103 model selection section

[0301] 104 learning section

[0302] 105 parameter estimation section

[0303] 106, 213, 313, 413 display section

[0304] 107, 212, 312, 412 input / output section

[0305] 108 data acquisition section

[0306] 109 screening section

[0307] 200 wire management device

[0308] 211 work control section

[0309] 214 work mechanism

[0310] 300 data management device

[0311] 311 data control section

[0312] 401 to 403 inspection device

[0313] 411 inspection control section

[0314] 414 inspection mechanism

[0315] d component information

[0316] DB1, DB4 production data holding section

[0317] DB2 learning model holding section

[0318] DB3 component library holding section

[0319] DB5 inspection data holding section

[0320] DB6 data holding section

[0321] Dc component data

[0322] Dcu actual performance component data

[0323] Dp production data

[0324] Dpu actual performance production data

[0325] L1 to L3 component mounting lines

[0326] m, mu action parameters

[0327] M4, M5 component mounting devices

Claims

1. A production data generation device, comprising: a model selection section that selects at least one learning model from among a plurality of learning models that respectively correspond to relationships between action conditions of a component mounting device that respectively mounts components to a board and the components; a parameter estimation section that estimates, based on the selected at least one learning model and component information related to a mounting target component mounted to a board, an action condition, i.e., an action parameter, of the component mounting device for mounting the mounting target component to the board; and a data generation section that generates production data including component data having the component information and the action parameter.

2. The production data generation device according to claim 1, wherein the production data generation device further comprises: a data acquisition section that acquires actual performance production data used by a component mounting device including actual performance component data having component information related to a mounted component and an action parameter used in mounting the mounted component; and a learning section that performs updating of the relationships represented by a learning model corresponding to the acquired actual performance production data among the plurality of learning models through learning using the actual performance component data as teaching data.

3. The production data generation device according to claim 1, wherein the component information represents at least one of a size, a shape, an appearance, a category, and a supply method for supplying the component corresponding to the component information.

4. The production data generation device according to claim 1, wherein the action parameter is a parameter related to at least one of transfer, recognition, suction, and assembly of the component performed by the component mounting device.

5. The production data generation device according to claim 2, wherein the learning section learns the learning model corresponding to a period in which the actual performance production data is acquired.

6. The production data generation device according to claim 2, wherein the plurality of learning models respectively correspond to different production equipment, the learning section learns the learning model corresponding to production equipment including the component mounting device that uses the actual performance production data.

7. The production data generation device according to any one of claims 2, 5, and 6, wherein the plurality of learning models respectively correspond to different types of mounting boards, the learning section learns the learning model corresponding to a type of mounting board produced using the actual performance production data.

8. A production data generation method, comprising: selecting at least one learning model from among a plurality of learning models that respectively correspond to relationships between action conditions of a component mounting device that respectively mounts components to a board and the components, estimating, based on the selected at least one learning model and component information related to a mounting target component mounted to a board, an action condition, i.e., an action parameter, of the component mounting device for mounting the mounting target component to the board, ​ production data including component data having the component information and the action parameters is generated.

9. The production data generation method according to claim 8, wherein In the production data generation method, further, actual performance production data used by the component mounting device, including actual performance component data having component information related to a mounted component and action parameters used in mounting of the mounted component, the relational update indicated by the learning model corresponding to the actual performance production data that is acquired, among the plurality of learning models, is performed by learning using the actual performance component data as teaching data.

10. The production data generation method according to claim 8, wherein The component information indicates at least one of a size, a shape, an appearance, a category, and a supply method for supplying the component, of the component corresponding to the component information.

11. The production data generation method according to claim 8, wherein The action parameters are parameters related to at least one of transfer, recognition, suction, and assembly of the component, by the component mounting device.

12. The production data generation method according to claim 9, wherein In the learning, the learning model corresponding to a period during which the actual performance production data is acquired is learned.

13. The production data generation method according to claim 9, wherein The plurality of learning models respectively correspond to mutually different production equipment, In the learning, the learning model corresponding to production equipment including the component mounting device that uses the actual performance production data is learned.

14. The production data generation method according to claim 9, wherein The plurality of learning models respectively correspond to mutually different types of mounting boards, In the learning, the learning model corresponding to the type of mounting board produced using the actual performance production data is learned.

15. A recording medium storing a program that causes a computer to execute the production data generation method according to any one of claims 8 to 14.

Images