Component data management method, component data management program, component data management device, and component data management system

Abstract

This component data management method for managing component data each obtained by associating, with component information of a component, an operation parameter that is an operation condition for a component mounting device for mounting the component on a substrate includes: storing, in a database, at least the component data used in the component mounting device and a learning model type for the component data in association with each other (ST3, ST4); and producing a trained model by using, among the component data stored in the database, the component data associated with one of a plurality of the learning model types (ST8).

Description

Component Data Management Method, Component Data Management Program, Component Data Management Device, and Component Data Management System

[0001] The present disclosure relates to a component data management method, a component data management program, a component data management device, and a component data management system for managing component data used in a component mounting device that mounts components on a substrate.

[0002] A component mounting device that mounts components on a substrate controls component mounting operations based on a number of operation parameters including operation conditions related to component suction by a nozzle, imaging of components by a camera, and component mounting on the substrate. These operation parameters are set with appropriate values for each component as component data associated with component information including information such as the shape of the component. Patent Document 1 discloses creating performance information that aggregates performance data when mounting using the component data based on the creation information of the component data in order to generate a learned model for creating component data with high productivity and mounting quality by machine learning.

[0003] Japanese Patent No. 7535735

[0004] By the way, even for the same component, the operation parameters prioritized when creating component data differ depending on what product the mounting substrate on which the component is mounted is used for. However, in the prior art including Patent Document 1, it is difficult to appropriately prepare the component data used for machine learning in consideration of the circumstances of each user such as the use of the mounting substrate for the product when generating a learned model for creating component data, and there was room for further improvement.

[0005] Therefore, an object of the present disclosure is to provide a component data management method, a component data management program, a component data management device, and a component data management system that can appropriately create a learned model for creating component data according to the application.

[0006] The component data management method of this disclosure is a component data creation method for managing component data in which operating parameters, which are the operating conditions of a component mounting device for mounting the component onto a circuit board, are linked to component information of the component, and includes a storage step of storing the component data used by the component mounting device and the learning model type of the component data in a database, linking them together at least; and a learning step of creating a trained model using the component data stored in the database that is linked to one of the learning model types among a plurality of learning model types.

[0007] The parts data management program of this disclosure causes a computer to execute the parts data management method described in any one of claims 1 to 8.

[0008] The component data management device of this disclosure is a component data management device that manages component data in which operating parameters, which are the operating conditions of a component mounting device for mounting the component onto a circuit board, are linked to component information of a component, and includes a storage processing unit that stores in a database at least the component data used by the component mounting device and the learning model type of the component data, and a learning unit that creates a trained model using component data from the component data stored in the database that is linked to one of the learning model types among a plurality of learning model types.

[0009] The component data management system of this disclosure is a component data management system that manages component data in which operating parameters, which are the operating conditions of a component mounting device for mounting the component onto a circuit board, are linked to component information of the component, and includes a storage processing unit that stores in a database at least the component data used by the component mounting device and the learning model type of the component data, and a learning unit that creates a trained model using component data from the component data stored in the database that is linked to one of the learning model types among a plurality of learning model types.

[0010] According to this disclosure, it is possible to appropriately create a trained model that generates component data according to the application.

[0011] A diagram illustrating the configuration of a production system according to one embodiment of this disclosure. A block diagram showing the configuration of a production system according to one embodiment of this disclosure. A diagram illustrating the configuration of component data used in a production system according to one embodiment of this disclosure. A diagram illustrating the component data creation process in a production system according to one embodiment of this disclosure. A diagram showing an example of a type selection screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of a model selection (mounting board type) screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of a model selection (mounting request type) screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of a model selection (model creation date) screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of a data selection screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of a parameter selection screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A diagram showing an example of an execution screen displayed on the display unit of a production management device according to one embodiment of this disclosure. A flowchart of the first embodiment of the component data management method according to one embodiment of this disclosure. A flowchart of the second embodiment of the component data management method according to one embodiment of this disclosure. A flowchart of the third embodiment of the component data management method according to one embodiment of this disclosure.

[0012] An embodiment of this disclosure will be described in detail below with reference to the drawings. The configurations, shapes, etc. described below are illustrative examples for illustrative purposes and can be modified as appropriate according to the specifications of the production system (parts data management system), parts mounting line (production line), parts mounting equipment, production management equipment, and parts data management equipment. In the following, all corresponding elements are denoted by the same reference numerals in all drawings, and redundant explanations are omitted.

[0013] First, let's explain the configuration of Production System 1 with reference to Figure 1. Production System 1 consists of customer factories F1 to F3 and a support center S that supports the customer's production activities. Each factory F1 to F3 is equipped with a component mounting line L1 that produces mounted circuit boards as a production line for producing manufactured goods. The support center S may be located away from factories F1 to F3, or it may be located within factories F1 to F3. In addition, the support center S may be set up for each customer, or it may be set up to support multiple customers at once. Furthermore, the functions of the support center S may be configured using cloud computing.

[0014] Hereafter, Factory F1 will be referred to as "Factory 1 F1," Factory F2 as "Factory 2 F2," and Factory F3 as "Factory 3 F3." Figure 1 explains the configuration of Factories F1 to F3 using Factory 1 F1 as an example.

[0015] In Figure 1, the first factory F1 is equipped with one component mounting line L1, which is composed of multiple component mounting devices M1 and M2, an inspection device M3, and other production equipment connected together. The component mounting line L1 has the function of sequentially mounting components onto a solder-printed substrate using component mounting devices M1 and M2, and then inspecting the condition of the components mounted on the substrate with the inspection device M3 to produce mounted substrates. Note that the component mounting line L1 installed in the first factory F1 does not have to be just one; there may be two or more. Also, the component mounting devices M1 and M2 that make up the component mounting line L1 do not have to be two; there may be one or three or more.

[0016] Each production device within Factory F1 is connected to an internal communication network 2, such as a LAN (Local Area Network), and is connected to a production management device 3 via an internal communication unit 4. The production management device 3 has the function of creating the data and parameters necessary for the operation of the production devices on component mounting line L1 and transmitting them to each production device. Specifically, the production management device 3 creates component data, production data, etc., used by component mounting devices M1 and M2 on component mounting line L1. The production management device 3 also has the function of collecting actual data from each production device, including the operating status of each production device, work history, and information on events that occurred at the production device. In addition to the production management device 3, Factory F1 may also be configured to have a line management device for each component mounting line L1 that manages the production of mounted boards.

[0017] In Figure 1, a parts data management device 7 is installed in the support center S. Multiple factories F1 to F3 are equipped with external communication units 5 connected to production management devices 3. In addition, an external communication unit 8 connected to the parts data management device 7 is installed in the support center S. External communication units 5 and 8 are connected to an off-site communication network 6, such as the internet or a mobile communication line. With this configuration, the production management device 3 and the parts data management device 7 can exchange information via the off-site communication network 6.

[0018] The parts data management device 7 has the function of acquiring production equipment performance data and parts data from the production management devices 3 of each factory F1 to F3 and storing them in a database. The parts data management device 7 also has the function of learning from the data stored in the database and creating a trained model. Furthermore, the parts data management device 7 has the function of generating and transmitting (outputting) parts data in response to requests from the production management devices 3.

[0019] Furthermore, the production management device 3 and the parts data management device 7 are not limited to a configuration in which they directly exchange information via the external communication network 6; they may also exchange information via the cloud. That is, information transmitted from each device may be stored in the cloud, and the information may be transmitted from the cloud to each device upon request. Alternatively, information may be sent and received using communication tools such as email or data communication lines.

[0020] Next, with reference to Figure 1, the component mounting line L1 will be described. Component mounting devices M1 and M2 are production devices that have the function of mounting components onto substrates on which solder has been printed using a printing device (not shown).

[0021] The component mounting devices M1 and M2, based on the operation parameters contained in the component data set for each component to be mounted on the substrate, pick up components supplied by the feeder using vacuum suction with a nozzle on the mounting head, capture an image of the state of the component held by the nozzle with a component recognition camera, and mount the component at the mounting position on the substrate at a specified mounting angle. The component mounting devices M1 and M2 are equipped with multiple sensors that monitor the suction operation when the nozzle picks up the component, the component recognition operation when the component recognition camera captures and recognizes the picked-up component, and the mounting operation when the nozzle mounts the component to the substrate, as well as any work errors or operational errors during the component mounting process.

[0022] In Figure 1, the inspection device M3 uses an inspection camera to capture images of the components mounted on the substrate by the component mounting devices M1 and M2, and determines whether the mounting status of the components, such as their mounting position and orientation, is acceptable or unacceptable, as well as whether the mounted substrate is acceptable or unacceptable. Based on the mounting status of the components mounted on the substrate, errors such as mounting mistakes in the component mounting devices M1 and M2 are detected.

[0023] Next, referring to Figure 2, the configuration of the production system 1 (parts data management system) equipped with the parts data management device 7 will be described. Here, among the multiple functions of the production system 1, the configuration of the function that stores the parts data used by the parts mounting devices M1 and M2, the type of learning model for the parts data, and the actual data of the parts mounting devices M1 and M2 in a database, creates a learned model, and creates parts data will be described. Furthermore, the production management devices 3 installed in factories F1 to F3 have a similar configuration, and here, the first factory F1 will be used as an example.

[0024] The production management device 3 installed in the first factory F1 is connected to an internal communication unit 4, an external communication unit 5, an input unit 9, and a display unit 10. The input unit 9 is an input device such as a keyboard, touch panel, or mouse, and is used for inputting operation commands and data. The display unit 10 is a display device such as an LCD panel, which displays various data stored in the production management storage unit 11, as well as various information such as operation screens and input screens for operations performed by the input unit 9.

[0025] In Figure 2, the production management device 3 comprises a production management storage unit 11, a data acquisition unit 15, a display processing unit 16, a selection processing unit 17, and a production control unit (not shown). The processing units, such as the data acquisition unit 15, the display processing unit 16, the selection processing unit 17, and the production control unit, are realized by, for example, a memory that stores the control programs executed by each processing unit, and a processor that executes the control programs. The production control unit is, for example, a CPU (Central Processing Unit) and controls the entire production management device 3.

[0026] The production management memory unit 11 is a storage device that stores various types of information, including a production library 12, a parts library 13, and performance data 14. The production management memory unit 11 is implemented using, for example, flash memory or an HDD (Hard Disk Drive).

[0027] In Figure 2, the production library 12 stores production data 12a used in the production of mounted boards by component mounting devices M1 and M2, for each production model name of the mounted board. The production data 12a includes the component name that identifies the component to be mounted on the board, the component code that associates the component with the component data 13a in the component library 13, the mounting position and mounting angle of the component on the board, the component arrangement indicating the position of the feeder that supplies the component in the component mounting devices M1 and M2, and the nozzle arrangement indicating the position of the nozzle that picks up the component in the mounting head. In this way, the production data 12a is associated with the component data 13a of the components to be mounted on the mounted board. Furthermore, the number of components used in a single board can be calculated from the production data 12a.

[0028] The parts library 13 stores multiple parts data 13a, in which operating parameters are linked to parts information. The parts data 13a are associated with the production data 12a of the production library 12 by their parts codes. The parts data 13a also contains creation information related to the creation of the parts data 13a, including the date and time of creation and information identifying the creator. The parts library 13 also stores model data 13b. Model data 13b is information that associates the parts code of the parts data 13a with the learning model type of the parts data 13a, which will be described later.

[0029] Here, with reference to Figure 3, an example of part data 13a stored in the part library 13 will be described. The part data 13a is associated with the production data 12a of the production library 12 by the "part code" 41 contained in the part data 13a.

[0030] The part data 13a consists of a shape diagram 42, size data 43, part parameters 44, operation parameters 47, etc. Images, numerical values, and terms are entered in the blank spaces of each item. Note that the "numerical values" used here are not limited to numerical data, but also include the selection results of quantitatively and qualitatively expressed choices such as yes / no, inexpensive / expensive, high speed / medium speed / low speed, etc. The shape diagram 42, size data 43, and part parameters 44 are part information that identifies the characteristics of the part. The shape diagram 42 illustrates the external shape of the part in question. The size data 43 shows the size information of the part, i.e., external dimensions, number of leads, lead pitch, lead length, lead width, part height, etc., as numerical data.

[0031] In Figure 3, the part parameter 44 is attribute information about the part, and includes basic information 45, which is information about the part itself, and tape information 46, which is information about the carrier tape for supplying the part by the feeder. The basic information 45 shows the polarity of the part, polarity mark, mark position, part type, shape type, and price information. The tape information 46 includes the tape material of the carrier tape, the tape width indicating the width dimension of the carrier tape, the feed interval indicating the tape feed pitch, and color and material information, which is information related to the characteristics when the carrier tape is the target of image recognition.

[0032] The operation parameters 47 are machine parameters that define the operating mode when the component is subjected to component mounting work by component mounting devices M1 and M2. In the example shown here, the model 47a indicates the type of component mounting devices M1 and M2, and the nozzle setting 47b indicates the type of nozzle used. Furthermore, the operation parameters 47 include speed parameters 47c, recognition 47d, gap 47e, suction 47f, mounting 47g, etc.

[0033] In Figure 3, the speed parameter 47c includes the suction speed when the nozzle picks up a component, the mounting speed when the mounting head transfers a component, and the tape feed speed when the feeder feeds the carrier tape. Recognition 47d is a parameter that defines the mode of component recognition and includes the camera type indicating the type of component recognition camera used, the illumination mode indicating the illumination type during imaging, and the recognition speed, which is the movement speed of the nozzle during imaging. The recognition speed can be set from high speed, medium speed, and low speed. Note that the speed parameters may be numerical values ​​(1 to 100%) or options (high speed, medium speed, low speed, etc.).

[0034] The gap 47e includes the suction gap when the nozzle picks up the component and the mounting gap when the held component is mounted on the substrate. Suction 47f specifies the suction position offset, which indicates the amount of offset when the nozzle picks up the component, and the suction angle. Mounting 47g specifies the pressing load when mounting the component held by the nozzle onto the substrate.

[0035] Thus, the operation parameters 47 include nozzle parameters (nozzle setting 47b) related to the nozzle that picks up the part, suction parameters related to suction when the nozzle picks up the part (suction speed, suction gap, suction 47f), recognition parameters (recognition 47d) for recognizing the shape of the part, and mounting parameters (mounting speed, mounting gap, mounting 47g) for mounting the part. Furthermore, the suction speed and mounting speed are parameters related to the movement speed of the part included in the operation parameters 47. Note that the part parameters 44 and operation parameters 47 shown in the part data 13a of Figure 4 are examples of the relevant items, and various other parameters may be set as needed.

[0036] For example, these include the suction hold time, which is the time the nozzle is in contact with the component when suctioning the component; the mounting hold time, which is the time the component is in contact with the substrate when mounting the component to the substrate; the number of component recognitions, which is the number of times the recognition camera recognizes the component; suction check ON / OFF, which determines whether to check whether the component has been suctioned; thickness variation tolerance, which sets the tolerance value when measuring the thickness of the component; component suction state detection ON / OFF, which determines whether to detect the suction state of the component; simultaneous component suction / mounting ON / OFF, which determines whether to suction or mount the components at the same time; automatic component suction position learning ON / OFF, which determines whether to automatically set the suction position of the component; the number of component suction retries, which is the number of times suction is attempted again if suction fails; and the number of recognition retries, which is the number of times recognition is attempted again if recognition fails.

[0037] The operating parameters 47 may be changed even for components with the same component information, depending on the model of the component mounting equipment M1 and M2 used to mount the components onto the board, the material of the board, the electrodes on the board, etc., or to improve mounting quality or mounting error rate. Furthermore, even for components with the same component information, different sets of operating parameters 47 are created depending on the application of the mounting board (mounting board type) and the requirements for the mounting board, such as whether productivity or quality is prioritized (mounting requirement type).

[0038] When the operating parameter 47 of the component is changed, component data 13a is created (updated) with the changed operating parameter 47 linked to it, without changing the component parameters 44 and other information. At this time, the operating parameter before modification is distinguished from the component code of the component data 13a by assigning a new component code 41 to the component code. In this way, the component data 13a links the component information of the component (shape diagram 42, size data 43, component parameters 44) with the operating parameter 47, which is the operating condition of the component mounting devices M1 and M2 for mounting the component onto the circuit board.

[0039] In Figure 2, the data acquisition unit 15 collects work results, event information, inspection information, etc., from the production equipment (component mounting devices M1, M2, inspection device M3) of the component mounting line L1 installed in the first factory F1. The data acquisition unit 15 associates the collected information with information that identifies the mounted board produced, information that identifies the production equipment that performed the component mounting work, and information that identifies the component data 13a, and stores it in the production management storage unit 11 as performance data 14.

[0040] The performance data 14 includes information such as the production start date and time, production end date and time, number of units produced, information identifying the mounted board and the number of components mounted on the board, working time, information on the component data 13a used, and information on the nozzle and feeder used. The performance data 14 also includes event information such as the number of work errors, error rate (frequency, spoilage rate), number of operation errors, and the time and content of the errors. Furthermore, the performance data 14 includes inspection information such as information identifying the mounted board inspected by the inspection device M3, the pass / fail judgment result, and the pass / fail judgment result for the mounting status of the components mounted on the board. Thus, the performance data 14 includes information on mounting errors that occurred in the component mounting devices M1 and M2.

[0041] In Figure 2, the display processing unit 16 displays a setting screen (not shown) on the display unit 10, which allows the user (administrator) of the first factory F1 to store the component data 13a used in the component mounting devices M1 and M2 into the database of the component data management device 7, to set which learning model types of component data 13a should be stored or not stored. The display processing unit 16 also transmits the information entered from the setting screen to the component data management device 7.

[0042] The selection processing unit 17 displays a selection screen, which will be described later, on the display unit 10. On the selection screen, the learning model type for the component data 13a to be machine-learned by the component data management device 7 is selected. The selection processing unit 17 transmits the information entered from the selection screen to the component data management device 7.

[0043] In FIG. 2, the component data management device 7 installed in the support center S has a function of acquiring production data 12a, component data 13a, model data 13b, performance data 14, etc. from the customer's factories F1 to F3 and storing them in a database. Further, the component data management device 7 has a function of creating a learned model by machine learning based on the information stored in the database and creating the component data 13a used in the component mounting devices M1 and M2 of the factories F1 to F3.

[0044] An external communication unit 8 and a storage device 20 are connected to the component data management device 7. The storage device 20 stores a database 21 for storing information acquired from a plurality of factories F1 to F3, a learned model 22, and the like. The storage device 20 is realized by, for example, a flash memory or an HDD. The database 21 may include the component data 13a used in a plurality of factories F1 to F3 and a database prepared in advance by an EDA (Electronic Design Automation) vendor.

[0045] In FIG. 2, the component data management device 7 includes information processing devices such as an acquisition unit 30, a storage processing unit 31, a determination unit 32, a learning unit 33, a component data creation unit 34, an output unit 35, and a management control unit (not shown). Each information processing device is realized not only as an independent hardware asset but also by a memory that stores a control program executed by each information processing device, a processor that executes the control program, and the like. The management control unit controls the entire component data management device 7. Further, the component data management device 7 does not necessarily have to be composed of one computer and may be composed of a plurality of devices. For example, all or part of the storage device 20 and the information processing device may be provided in the cloud.

[0046] The acquisition unit 30 acquires component data 13a used in the component mounting devices M1 and M2 from a plurality of factories F1 to F3 via the external communication unit 8, and model data 13b including the learning model type of the component data 13a. Further, the acquisition unit 30 acquires performance data 14 including mounting errors obtained when components are mounted by the component mounting devices M1 and M2 using the component data 13a, and stores the performance data 14 in the database 21. The acquisition of data by the acquisition unit 30 is executed, for example, once a day or at a predetermined time such as the switching of a production lot. Further, the acquisition unit 30 acquires data by the operation of the production management device 3 by users from the factories F1 to F3.

[0047] In FIG. 2, the storage processing unit 31 executes a storage process of associating the component data 13a used in the component mounting devices M1 and M2 acquired by the acquisition unit 30 with the learning model type of the component data 13a specified by the model data 13b acquired by the acquisition unit 30, and storing the data in the database 21. The storage processing unit 31 stores all of the component data 13a acquired by the acquisition unit 30 in the database 21, and also stores only the component data 13a associated with the learning model type set by the user using a setting screen or the like in the database 21.

[0048] In FIG. 2, the learning unit 33 uses, as learning data, the component data 13a associated with the selected learning model type among the plurality of component data 13a (component parameters 44, operation parameters 47) stored in the database 21 and the performance data 14, and creates a learned model 22 that estimates the operation parameters 47 that conform to the selected learning model type by a learning algorithm using machine learning or the like. The generated learned model 22 is associated with the learning model type and stored in the storage device 20.

[0049] The learning unit 33 uses the component data 13a and performance data 14 stored in the database 21 to further train the pre-trained model 22 stored in the storage device 20. If a pre-trained model 22 of the selected learning model type is not stored in the storage device 20, the learning unit 33 uses the component data 13a and performance data 14 stored in the database 21 to further train a general-purpose pre-trained model 22 supplied by the manufacturer of the component mounting equipment M1 and M2. If neither a pre-trained model 22 of the selected learning model type nor a general-purpose pre-trained model 22 exists, the learning unit 33 uses the component data 13a and performance data 14 associated with the selected learning model type stored in the database 21 to create a pre-trained model 22 using machine learning.

[0050] The learning algorithms that can be used include neural networks (including deep learning using multi-layer neural networks), genetic programming, decision trees, Bayesian networks, and support vector machines (SVMs). The user can select the learning model type using a selection screen, or the determination unit 32 can automatically select it based on production data 12a and performance data 14. Details on how to select the learning model type will be described later.

[0051] The part data creation unit 34 estimates (calculates) the operation parameters 47 based on the created trained model 22 and the part information (shape diagram 42, size data 43, part parameters 44) of the parts transmitted from the production management devices 3 of factories F1 to F3, and creates part data 13a that matches the trained model type. The output unit 35 transmits the created part data 13a to the production management devices 3 of the requesting factories F1 to F3 via the external communication unit 8.

[0052] Thus, in the parts data management system (production system 1) equipped with the parts data management device 7, parts data 13a that conforms to the learning model type is created using a trained model 22 that conforms to the learning model type learned by machine learning at the support center S based on requests from factories F1 to F3.

[0053] Here, referring to Figure 4, the processing steps for creating part data 13a that matches the learning model type in the part data management system (production system 1) will be described. The acquisition unit 30 of the part data management device 7 acquires actual data 14 from the production management device 3 and stores it in the database 21. The acquisition unit 30 also acquires part data 13a and model data 13b used by the part mounting devices M1 and M2 from the part library 13 of the production management device 3. The storage processing unit 31 links the learning model types contained in the acquired part data 13a and model data 13b and stores them in the database 21.

[0054] The learning unit 33 further trains the trained model 22 of the previously created learning model type selected by the selection processing unit 17 using the component data 13a associated with the selected learning model type from the component data 13a stored in the database 21, and the actual data 14 using the component data 13a, to create a trained model 22. The component data creation unit 34 creates component data 13a that is suitable for the learning model type based on the created trained model 22 and component information. The output unit 35 transmits (outputs) the created component data 13a to the production management device 3. The transmitted component data 13a is stored in the component library 13.

[0055] The learning unit 33 may also create a trained model 22 by performing additional training using a trained model 22 of a selected group of previously created learning model types. In this case, additional training is performed using the component data 13a associated with the selected group of learning model types from the component data 13a stored in the database 21, and the actual data 14 using the component data 13a.

[0056] Furthermore, if multiple learning model types are selected, additional learning may be performed by weighting based on features such as external dimensions, lead count, lead pitch, lead width, and part height, which are size data of the part data 13a. In other words, the closer the operating parameter 47 is to the size data of the part data 13a to be estimated, the greater the weight given to it. Alternatively, the part data 13a used for learning may be statistically aggregated by part size, part type, etc., and part data 13a that fall within a predetermined range of that distribution may be used for additional learning.

[0057] Next, referring to Figures 5 to 11, we will describe examples of various screens for creating component data 13a by selecting a learning model type that is set when the selection processing unit 17 creates the learned model 22 displayed on the display unit 10. The learning model type selected by the user may be one that has been created in advance by the mounting machine manufacturer, or one that has been created or updated by the user.

[0058] First, with reference to Figure 5, an example of the type selection screen 50 displayed on the display unit 10 by the selection processing unit 17 will be described. The type selection screen 50 is provided with a step display area 51, a part name specification area 52, a type specification area 53, and a next button 54.

[0059] In Figure 5, the step display area 51 displays the operation steps for causing the component data management device 7 to create a learned model 22. The step display area 51 has frames that display "Type Selection," "Model Selection," "Data Selection," "Parameter Selection," and "Execution" in the order of the operation steps. The step display area 51 is also displayed on the screens shown in Figures 6 to 11, which will be described later, and the frame of the current operation step is inverted in black and white. On the type selection screen 50, "Type Selection" is displayed inverted to indicate that it is the type selection screen 50.

[0060] In the part name specification area 52, the part name for which the part data 13a will be created is entered using the input unit 9. In this example, "D001" is entered as the part name. In the type specification area 53, "mounting board type," "mounting request type," and "model creation date" are displayed as major categories of learning model types. In the type specification area 53, a major category is selected by selecting the radio button 53a displayed using the input unit 9. In this example, "mounting board type" is selected as the major category. When the next button 54 is operated, the information selected and entered on the type selection screen 50 is confirmed, and the display switches to the model selection screen for the next step.

[0061] Next, with reference to Figure 6, an example of the model selection (mounted board type) screen 55 displayed on the display unit 10 by the selection processing unit 17 will be described. The model selection (mounted board type) screen 55 is displayed when "mounted board type" is selected on the type selection screen 50. The model selection (mounted board type) screen 55 is provided with a step display area 51, a component name specification area 52, a model specification area 56, a previous button 57, and a next button 58.

[0062] In the model selection area 56, the learning model types "Automotive board," "Device board," "Electronic equipment board," and "Mobile phone board" are displayed. In the model selection area 56, the learning model type is selected by selecting the radio button 56a displayed using the input unit 9. In this example, "Automotive board" is selected as the learning model type. When the previous button 57 is operated, the display switches to the previous type selection screen 50. When the next button 58 is operated, the information selected and entered on the model selection (mounting board type) screen 55 is confirmed, and the display switches to the data selection screen for the next step.

[0063] Thus, the learning model types are broadly categorized and correspond to the mounting board types. Each mounting board type includes at least one learning model type from among automotive boards, device boards, electronic equipment boards, and mobile phone boards.

[0064] Next, with reference to Figure 7, an example of the model selection (implementation request type) screen 59 displayed on the display unit 10 by the selection processing unit 17 will be described. The model selection (implementation request type) screen 59 is displayed when "implementation request type" is selected on the type selection screen 50. The model selection (implementation request type) screen 59 is provided with a step display area 51, a part name specification area 52, a model specification area 60, a previous button 61, and a next button 62.

[0065] In the model selection area 60, "Productivity-focused" and "Quality-focused" are displayed as learning model types. In the model selection area 60, the learning model type is selected by selecting the radio button 60a displayed using the input unit 9. In this example, "Productivity-focused" is selected as the learning model type. When the previous button 61 is operated, the display switches to the previous type selection screen 50. When the next button 62 is operated, the information selected and entered on the model selection (implementation requirement type) screen 59 is confirmed, and the display switches to the data selection screen for the next step.

[0066] Thus, learning model types are broadly categorized and correspond to implementation requirement types. Furthermore, implementation requirement types include at least one learning model type that is either productivity-focused or quality-focused.

[0067] Next, with reference to Figure 8, an example of the model selection (model creation date) screen 63 displayed on the display unit 10 by the selection processing unit 17 will be described. The model selection (model creation date) screen 63 is displayed when "model creation date" is selected on the type selection screen 50. The model selection (model creation date) screen 63 is provided with a step display area 51, a part name specification area 52, a model creation date specification area 64, a previous button 65, and a next button 66.

[0068] In the model creation date specification area 64, the model creation date on which the trained model 22 was created is entered using the input unit 9. In this example, "July 1, 2024" is entered as the model creation date. When the previous button 65 is operated, the display switches to the previous type selection screen 50. When the next button 66 is operated, the information selected and entered on the model selection (model creation date) screen 63 is confirmed, and the display switches to the data selection screen for the next step.

[0069] Thus, the learning model type is associated with the creation date and time of the trained model 22 used to create the operating parameters 47.

[0070] Next, with reference to Figure 9, an example of the data selection screen 67 displayed on the display unit 10 by the selection processing unit 17 will be described. The data selection screen 67 is provided with a step display area 51, a part name specification area 52, a data specification area 68, a previous button 69, and a next button 70.

[0071] The data specification area 68 displays "Factory 1 Line 1", "Factory 2 Line 1", "Factory 2 Line 2", and "Factory 3 Line 1" as the ranges of data to be used to create the trained model 22. In the data specification area 68, the range of data to be used is selected by selecting the displayed check button 68a using the input unit 9. In this example, "Factory 1 Line 1" and "Factory 3 Line 1" are specified as the ranges of data to be used.

[0072] When the "Previous" button 69 is pressed, the display switches to one of the previous screens: the Model Selection (Mounting Board Type) screen 55, the Model Selection (Mounting Request Type) screen 59, or the Model Selection (Model Creation Date) screen 63. When the "Next" button 70 is pressed, the information selected and entered on the Data Selection screen 67 is confirmed, and the display switches to the parameter selection screen for the next step.

[0073] Next, with reference to Figure 10, an example of the parameter selection screen 71 displayed on the display unit 10 by the selection processing unit 17 will be described. The parameter selection screen 71 is provided with a step display area 51, a part name specification area 52, a parameter specification area 72, a previous button 73, and a next button 74.

[0074] In the parameter specification area 72, "All," "Suction Speed," "Implementation Speed," and "Recognition Speed" are displayed as the operation parameters 47 to be learned. In the parameter specification area 72, the operation parameters 47 to be learned are selected by selecting the displayed check button 72a using the input unit 9. In this example, "All" is specified as the operation parameter to be learned. When the previous button 73 is operated, the display switches to the previous data selection screen 67. When the next button 74 is operated, the information selected and entered in the parameter selection screen 71 is confirmed, and the display switches to the execution screen for the next step.

[0075] Next, with reference to Figure 11, an example of the execution screen 75 displayed on the display unit 10 by the selection processing unit 17 will be described. The execution screen 75 is provided with a step display area 51, a part name display area 76, a learning model type display area 77, a learning data display area 78, a learning target parameter display area 79, a previous button 80, and an execution button 81.

[0076] The part name display area 76 displays the part name specified in the part name specification area 52. In this example, the part with the part name "D001" is specified. The learning model type display area 77 displays the learning model type specified in the type selection screen 50, model selection (mounting board type) screen 55, model selection (mounting request type) screen 59, and model selection (model creation date) screen 63. In this example, "automotive board" is specified as the learning model type. The learning data display area 78 displays the range of data used to create the learned model 22, as specified in the data selection screen 67. In this example, "Factory 1 Line 1" and "Factory 3 Line 1" are specified as the range of data to be used.

[0077] The learning target parameter display area 79 displays the operating parameters 47 to be learned, as specified by the parameter selection screen 71. In this example, "All" is specified as the operating parameters 47 to be learned. When the previous button 80 is operated, the display switches to the previous parameter selection screen 71. When the execute button 81 is operated, the information displayed on the execution screen 75 is sent to the parts data management device 7, and processing in the parts data management device 7 is executed.

[0078] In this example, the learning unit 33 of the component data management device 7 creates a trained model 22 using machine learning with all the operating parameters 47 of the component data 13a of "Factory 1 Line 1" and "Factory 3 Line 1" associated with the "in-vehicle circuit board". Then, the component data creation unit 34 uses the created trained model 22 and the component information (shape diagram 42, size data 43, component parameters 44) of the component named "D001" to create component data 13a that is suitable for mounting the component named "D001" on the "in-vehicle circuit board".

[0079] Next, following the flow in Figure 12, a first embodiment of a component data management method (data management program) for managing component data 13a, which links component information (shape diagram 42, size data 43, component parameters 44) with operating parameters 47 that are the operating conditions for component mounting devices M1 and M2 for mounting components onto a substrate, will be described. In the first embodiment of the component data management method, the user sets the learning model type using a type selection screen 50 or the like.

[0080] In Figure 12, first, the acquisition unit 30 acquires the component data 13a used by the component mounting devices M1 and M2, the model data 13b which includes information on the learning model type of the component data 13a, and the actual data 14 (ST1: acquisition process). Next, if the user has set the system to store all component data 13a (Yes in ST2), the storage processing unit 31 links all the component data 13a used by the component mounting devices M1 and M2 with the learning model type of the component data 13a and stores them in the database 21 (ST3: total storage process).

[0081] If the user has set the component data 13a to be stored (No in ST2), the component data 13a used by the component mounting devices M1 and M2 are linked to the user-set component data 13a and the learning model type of the component data 13a and stored in the database 21 (ST4: partial storage step). In this way, in the storage steps (ST2 to ST4), the component data 13a linked to the user-set learning model type is stored in the database 21 (ST4).

[0082] In Figure 12, the user then selects one of several trained model types using the Type Selection Screen 50 (Figure 5), Model Selection (Mounting Board Type) Screen 55 (Figure 6), Model Selection (Mounting Request Type) Screen 59 (Figure 7), Model Selection (Model Creation Date) Screen 63 (Figure 8), etc. (ST5: Selection Process). The user also selects the range of data to be used to create the trained model 22 using the Data Selection Screen 67 (Figure 9), etc. (ST6: Data Range Selection Process). By selecting the range of data to be used, a highly accurate trained model 22 can be created using data from the high-performing component mounting line L1. Furthermore, a trained model 22 can be created that generates component data 13a that reflects the customer's unique setting rules.

[0083] Furthermore, the user selects the operating parameters 47 to be learned in the creation of the trained model 22 using the parameter selection screen 71 (Figure 10), etc. (ST7: Operating parameter selection step). The selection processing unit 17 transmits the information selected by the user in the selection step (ST5), data range selection step (ST6), and operating parameter selection step (ST7) to the component data management device 7. In this way, the selection processing unit 17 selects one of several trained model types.

[0084] In Figure 12, the learning unit 33 then creates a trained model 22 using the part data 13a stored in the database 21 that is associated with the training model type selected in the selection step (ST5) (ST8: training step). Next, the part data creation unit 34 creates part data 13a based on the created trained model 22 and the part information of the part (ST9: part data creation step). Next, the output unit 35 transmits (outputs) the created part data 13a to the requesting production management device 3 (ST10: output step).

[0085] In addition, the output unit 35 may output only the operation parameters 47 created in the part data creation process (ST9) as part data 13a, or it may output only the operation parameters 47 selected by the user in the operation parameter selection process (ST7) from among the created operation parameters 47.

[0086] Next, a second embodiment of the component data management method (data management program) will be described following the flow chart in Figure 13. In the second embodiment, the determination unit 32 of the component data management device 7 determines, based on the production data 12a, which of the learning model types included in the mounting board type the learning model type of the component data 13a used in the production of the mounting board belongs to. This differs from the first embodiment. Hereafter, the same reference numerals are used for the same steps as in the first embodiment, and detailed explanations will be omitted.

[0087] In Figure 13, first, the acquisition unit 30 acquires component data 13a used by component mounting devices M1 and M2, production data 12a of mounted boards produced using component data 13a, and performance data 14 (ST21: first acquisition step). That is, in the first acquisition step (ST21), production data 12a using component data 13a is acquired. Note that in the first acquisition step (ST21), instead of acquiring production data 12a, other information including the size of the board, the number of components mounted on the board, and the size of the components mounted on the board may be acquired.

[0088] Next, the determination unit 32 determines which learning model type the acquired component data 13a belongs to, based on the size of the substrate, the number of components mounted on the substrate, and the size of the components mounted on the substrate, as included in the acquired production data 12a (ST22: determination step).

[0089] For example, the determination unit 32 determines that the learning model type is "automotive-mounted circuit board" if the size of the circuit board is larger than a predetermined size. Also, the determination unit 32 determines that the learning model type is "mobile phone-mounted circuit board" if the size of the circuit board is smaller than a predetermined size. Also, the determination unit 32 determines that the learning model type is "device-mounted circuit board" if the number of components mounted on the circuit board is less than a predetermined number. Also, the determination unit 32 determines that the learning model type is "electronic device-mounted circuit board" if the number of components mounted on the circuit board is less than a predetermined number and components larger than a predetermined size (large components) are mounted.

[0090] In Figure 13, the storage processing unit 31 then links the acquired component data 13a used by the component mounting devices M1 and M2 with the learning model type of the component data 13a determined in the determination step (ST22) and stores them in the database 21 (ST23: storage step). Next, the learning unit 33 creates a trained model 22 using the component data 13a stored in the database 21 that is linked to the learning model type determined in the determination step (ST22) (ST24: learning step). As a result, a trained model 22 that conforms to the learning model type is automatically created without user specification. Next, the component data creation step (ST9) and the output step (ST10) are executed.

[0091] Next, a third embodiment of the component data management method (data management program) will be described following the flow chart in Figure 14. In the third embodiment, the determination unit 32 of the component data management device 7 determines, based on the actual data 14, which of the learning model types included in the mounting request type the learning model type of the component data 13a used in the production of the mounted board is. This is different from the first embodiment. Hereafter, the same reference numerals are used for the same steps as in the first embodiment, and detailed explanations will be omitted.

[0092] In Figure 14, first, the acquisition unit 30 acquires the component data 13a used by the component mounting devices M1 and M2, and the actual data 14 (ST31: second acquisition step). That is, in the second acquisition step (ST31), the actual data 14 obtained when components were mounted by the component mounting devices M1 and M2 using the component data 13a is acquired. Next, the determination unit 32 determines which learning model type the component data 13a belongs to, based on the parameters related to the movement speed of the components (mounting speed, recognition speed, etc.) included in the operation parameters 47 of the acquired component data 13a, and the mounting error information included in the actual data 14 (ST32: determination step).

[0093] For example, the determination unit 32 determines that the learning model type is "productivity-oriented" if the mounting speed when the components are transferred by the mounting head is "high speed" and the number of mounting errors is greater than a predetermined threshold. Also, the determination unit 32 determines that the learning model type is "quality-oriented" if the mounting speed is "low speed" and the number of mounting errors is less than a predetermined threshold.

[0094] In Figure 14, the storage processing unit 31 then links the acquired component data 13a used by the component mounting devices M1 and M2 with the learning model type of the component data 13a determined in the determination step (ST32) and stores them in the database 21 (ST33: storage step). Next, the learning unit 33 creates a trained model 22 using the component data 13a stored in the database 21 that is linked to the learning model type determined in the determination step (ST32) (ST34: learning step). As a result, a trained model 22 that conforms to the learning model type is automatically created without user specification. Next, the component data creation step (ST9) and the output step (ST10) are executed.

[0095] As explained above, this disclosure discloses the following technical concepts.

[0096] (Technology 1) A component data creation method for managing component data 13a, which is obtained by linking component information (shape diagram 42, size data 43, component parameters 44) of a component with operating parameters 47 that are the operating conditions of component mounting devices M1 and M2 for mounting a component onto a substrate, comprising: a storage step (ST3, ST4, ST23, ST33) of storing in a database 21 at least the component data 13a used by component mounting devices M1 and M2 and the learning model type of the component data 13a linked together; and a learning step (ST8, ST24, ST34) of creating a trained model 22 using component data 13a linked to one of a plurality of learning model types among the component data 13a stored in the database 21.

[0097] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the application.

[0098] (Technology 2) The component data management method described in Technology 1, wherein the learning model type is associated with the mounting board type, and the mounting board type includes at least one learning model type from among automotive boards, device boards, electronic equipment boards, and mobile phone boards.

[0099] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the application (type of mounting board) of the mounting board.

[0100] (Technology 3) A component data management method according to Technology 1 or 2, wherein the learning model type is associated with the implementation requirement type, and the implementation requirement type includes at least one learning model type that is productivity-oriented or quality-oriented.

[0101] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the requirements for the mounting board (mounting requirement type).

[0102] (Technology 4) The learning model type is associated with the creation date and time of the trained model 22 used to create the operating parameters 47, as described in any of the component data management methods described in Technology 1 to 3.

[0103] This makes it possible to appropriately create a trained model 22 that generates part data 13a according to the creation date and time of the trained model 22.

[0104] (Technology 5) A component data management method according to any one of Techniques 1 to 4, wherein in the storage process (ST4), component data 13a associated with the learning model type set by the user is stored in a database 21.

[0105] This prevents unnecessary part data 13a from being stored in the database 21, and prevents the creation of a low-accuracy trained model 22 using inappropriate part data 13a.

[0106] (Technology 6) A component data management method according to any one of Technologies 1 to 5, further comprising a selection step (ST5) of selecting one of a plurality of learning model types, and in a learning step (ST8), a trained model 22 is created using component data 13a associated with the learning model type selected in the selection step (ST5).

[0107] This allows for the appropriate creation of a trained model 22 that generates component data 13a according to the learning model type selected by the user.

[0108] (Technology 7) A component data management method according to any one of Techniques 2 to 5, further comprising a first acquisition step (ST21) for acquiring production data 12a using component data 13a, a determination step (ST22) for determining which learning model type the component data 13a belongs to based on at least one of the size of the substrate, the number of components mounted on the substrate, and the size of the components mounted on the substrate included in the production data 12a, and a learning step (ST24) for creating a trained model 22 using the component data 13a associated with the determined learning model type.

[0109] This allows for the automatic determination of the learning model type included in the mounting board type, and the appropriate creation of a trained model 22 for generating component data 13a.

[0110] (Technology 8) A component data management method according to any one of Techniques 3 to 5, further comprising a second acquisition step (ST31) for acquiring actual data 14 obtained when a component is mounted by component mounting devices M1 and M2 using component data 13a, a determination step (ST32) for determining which learning model type the component data 13a is included in the mounting request type based on parameters related to the movement speed of the component included in the operation parameters 47 and information on mounting errors included in the actual data 14, and a learning step (ST34) for creating a trained model 22 using the component data 13a associated with the determined learning model type.

[0111] This allows for the automatic determination of the learning model type included in the implementation requirement type, and the appropriate creation of a trained model 22 for generating the component data 13a.

[0112] (Technical 9) A component data management program for causing a computer to execute the component data management method described in any one of Technical 1 to 8.

[0113] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the application.

[0114] (Technical 10) A component data management device 7 for managing component data 13a which associates the component information of a component with operating parameters 47 which are the operating conditions of component mounting devices M1 and M2 for mounting components onto a circuit board, comprising: a storage processing unit 31 which stores in a database 21 at least the component data 13a used by the component mounting devices M1 and M2 and the learning model type of the component data 13a, and a learning unit 33 which creates a trained model 22 using the component data 13a stored in the database 21 that is associated with one of a plurality of learning model types.

[0115] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the application.

[0116] (Technology 11) A component data management system (production system 1) for managing component data 13a that associates component information with operating parameters 47 which are the operating conditions of component mounting devices M1 and M2 for mounting components onto a circuit board, comprising: a storage processing unit 31 that stores in a database 21 at least the component data 13a used by component mounting devices M1 and M2 and the learning model type of the component data 13a, and a learning unit 33 that creates a trained model 22 using component data 13a that is associated with one of a plurality of learning model types from among the component data 13a stored in the database 21.

[0117] This makes it possible to appropriately create a trained model 22 that generates component data 13a according to the application.

[0118] Although various embodiments have been described above with reference to the drawings, it goes without saying that the present invention is not limited to these examples. It is clear to those skilled in the art that various modifications or alterations can be conceived within the scope of the claims, and these will naturally also fall within the technical scope of the present invention. Furthermore, the components of the above embodiments may be combined in any way without departing from the spirit of the invention.

[0119] This application is based on Japanese Patent Application No. 2024-164030 filed on September 20, 2024, the contents of which are incorporated herein by reference.

[0120] The component data management method, component data management device, component data management program, and component data management system disclosed herein have the effect of being able to appropriately create trained models that generate component data according to the application, and are useful in the field of mounting components on a substrate.

[0121] 1. Production System (Parts Data Management System) 7. Parts Data Management Devices M1, M2 Parts Mounting Devices

Claims

1. A component data creation method for managing component data, which associates component information of a component with operating parameters that are operating conditions for a component mounting device for mounting the component onto a circuit board, comprising: a storage step of storing in a database at least the component data used by the component mounting device and the learning model type of the component data in association with each other; and a learning step of creating a trained model using the component data stored in the database that is associated with one of the learning model types among a plurality of learning model types.

2. The component data management method according to claim 1, wherein the learning model type is associated with a mounting board type, and the mounting board type includes at least one of the learning model types of an automotive board, a device board, an electronic equipment board, or a mobile phone board.

3. The component data management method according to claim 1, wherein the learning model type is associated with an implementation requirement type, and the implementation requirement type includes at least one of the learning model types that is productivity-oriented or quality-oriented.

4. The component data management method according to claim 1, wherein the learning model type is associated with the creation date and time of the trained model used to create the operation parameters.

5. The component data management method according to claim 1, wherein in the storage step, the component data associated with the learning model type set by the user is stored in the database.

6. The component data management method according to claim 1, further comprising a selection step of selecting one of a plurality of learning model types, wherein in the learning step, the trained model is created using the component data associated with the learning model type selected in the selection step.

7. The component data management method according to claim 2, further comprising a first acquisition step of acquiring production data using the component data, further comprising a determination step of determining which of the learning model types included in the mounting board type the component data is based on at least one of the size of the substrate included in the production data, the number of components mounted on the substrate, and the size of the components mounted on the substrate, and in the learning step, creating the trained model using the component data associated with the determined learning model type.

8. The component data management method according to claim 3, further comprising a second acquisition step of acquiring actual data obtained when a component is mounted by a component mounting device using the component data, further comprising a determination step of determining which of the learning model types included in the mounting request type the component data is based on a parameter relating to the movement speed of the component included in the operation parameters and information on mounting errors included in the actual data, and in the learning step, creating the trained model using the component data associated with the determined learning model type.

9. A component data management program for causing a computer to execute the component data management method according to any one of claims 1 to 8.

10. A component data management device for managing component data, which associates component information of a component with operating parameters that are operating conditions for a component mounting device for mounting the component onto a circuit board, comprising: a storage processing unit that stores in a database at least the component data used by the component mounting device and the learning model type of the component data in association with each other; and a learning unit that creates a trained model using component data from the component data stored in the database that is associated with one of the learning model types among a plurality of learning model types.

11. A component data management system for managing component data in which operating parameters, which are the operating conditions of a component mounting device for mounting the component onto a circuit board, are linked to component information of a component, the system comprising: a storage processing unit that stores in a database at least the component data used by the component mounting device and the learning model type of the component data linked to each other; and a learning unit that creates a trained model using component data from the component data stored in the database that is linked to one of the learning model types among a plurality of learning model types.

Images