Storage medium, learning device and data collection system

Abstract

The engineering design tool (10) comprises: a device model editing unit (11) which edits first correspondence information based on an instruction from a user, the first correspondence information indicating a correspondence between a first device data type of first collected data collected from a first device and a first reference data type of first reference data interpretable by a first application; and a transformation candidate providing unit (12) which learns a transformation rule for transforming from the first reference data type to the first device data type based on the result of editing the first correspondence information, i.e., a transformation rule, and uses the transformation rule to infer a transformation candidate from a second reference data type of second reference data interpretable by a second application to a second device data type of second collected data collected from a second device.

Description

Technical Field

[0001] The present invention relates to an engineering design tool, a learning device and a data collection system for data collection. Background Art

[0002] In recent years, IoT (Internet of Things) technology has been used to collect data from industrial equipment installed at production sites and to feed back analysis results of the collected data to the production sites, thereby improving productivity at the production sites.

[0003] Production sites often operate in a multi-vendor environment, combining industrial equipment and other devices from various suppliers. Furthermore, the communication protocols used by each device often differ from one supplier to another. Therefore, in order to centrally utilize data collected by an IoT platform independently of the device or communication protocol, even if the definition of data that can be output externally varies by device, data collection must be performed using a unique data definition for applications such as those that analyze the collected data.

[0004] Therefore, it is necessary to collect the collected data from the device after associating the data definition of the reference data that can be interpreted by the application with the data definition of the collected data that can be interpreted by the industrial instrument.

[0005] The computer processing device described in Patent Document 1 uses mapping rules that represent the correspondence between input data and electronic data concepts to select an electronic data concept corresponding to the input data, and uses the selected concept to capture the structure of the input data. In Patent Document 1, input data corresponds to reference data that can be interpreted by an application, while electronic data concepts correspond to collected data.

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 2006-178982 Summary of the Invention

[0007] However, the technology of the above-mentioned patent document 1 has the following problem: for input data in which the concepts of input data and electronic data are pre-associated, the concept of electronic data corresponding to the input data can be provided, but for input data that is not associated with the concept of electronic data, the concept of electronic data cannot be provided. As described above, in the case where the technology of the above-mentioned patent document 1 is applied to a data collection system that collects collected data from a device and provides it to an application, the input data is reference data that can be interpreted by the application, and the concept of electronic data is collected data. Therefore, there is the following problem: in the case where the technology of the above-mentioned patent document 1 is applied to a data collection system, if there is no pre-defined conversion rule between the data type of the reference data that can be interpreted by the application and the data type of the collected data collected from the device, it is impossible to provide a data type of the collected data corresponding to the data type of the reference data.

[0008] The present invention is proposed in view of the above-mentioned problems, and its purpose is to obtain an engineering design tool that can provide data type candidates of collected data corresponding to the data type of benchmark data even if the data type of benchmark data that can be interpreted by the application is not associated with the data type of collected data.

[0009] To solve the above-mentioned problems and achieve the purpose, the engineering design tool of the present invention includes: an editing unit that, based on an instruction from a first user, edits first correspondence information indicating a correspondence between a data type of first collected data collected from a first device, namely, a first device data type, and a data type of first reference data interpretable by a first application, namely, a first reference data type. Furthermore, the engineering design tool of the present invention includes: a conversion candidate providing unit that, based on the result of editing the first correspondence information, learns a conversion rule for converting from the first reference data type to the first device data type, namely, a conversion rule, and uses the conversion rule to estimate conversion candidates for converting from a data type of second reference data, namely, a second reference data type, interpretable by a second application, to a data type of second collected data collected from a second device, namely, a second device data type.

[0010] Effects of the Invention

[0011] The engineering design tool according to the present invention can provide data type candidates of collected data corresponding to the data type of reference data even when the data type of reference data interpretable by an application is not associated with the data type of collected data. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a diagram showing the configuration of a data collection system according to an embodiment.

[0013] Figure 2 This is a diagram showing a configuration of a conversion candidate providing unit included in the engineering tool according to the embodiment.

[0014] Figure 3 This is a diagram showing a configuration of a conversion rule learning unit included in the engineering design tool according to the embodiment.

[0015] Figure 4 This is a diagram showing the structure of a neural network used in the engineering design tool according to the embodiment.

[0016] Figure 5 This is a flowchart showing the flow of operations when machine learning is performed by the engineering design tool according to the embodiment.

[0017] Figure 6 This is a flowchart showing the flow of operations when the engineering design tool according to the embodiment performs data estimation.

[0018] Figure 7 This is a diagram showing a first example of a hardware configuration for realizing a computer that runs the engineering design tool according to the embodiment.

[0019] Figure 8 This is a diagram showing a second example of a hardware configuration for realizing a computer that runs the engineering design tool according to the embodiment. DETAILED DESCRIPTION

[0020] Hereinafter, an engineering design tool, a learning device, and a data collection system according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

[0021] Implementation Method

[0022] Figure 1 1 is a diagram showing the configuration of a data collection system according to an embodiment of the present invention. A data collection system 1 includes an engineering tool 10 , an application 20 , a platform 30 , a communication server 40 , a device 50 , and a network line 60 .

[0023] The data collection system 1 collects device data from various instruments and provides collected data generated from the device data to applications 20. Examples of instruments include working machines and peripheral devices installed at a production site. In this embodiment, the device 50 is used as the instrument for collecting device data. Examples of the device data and collected data include operational data indicating the operating status of the device 50.

[0024] The engineering design tool 10, application 20, platform 30, and communication server 40 are each implemented using a computer such as a PC (Personal Computer). Alternatively, the application 20 and platform 30 may be implemented on the same computer. Alternatively, the platform 30 and communication server 40 may be implemented on the same computer.

[0025] In data collection system 1, platform 30, an IoT platform, obtains and stores collected data from devices 50 via communication server 40. Platform 30 obtains collected data from each device 50 according to a communication protocol. If an application 20 requests data collection, platform 30 provides the collected data to the application 20.

[0026] The engineering tool 10 is a software tool having a function of assisting in data collection settings in the platform 30. The engineering tool 10 transmits collection setting information, which is setting information for collecting data, to the platform 30.

[0027] The collection setting information includes information about the collected data stored by the platform 30. The collection setting information includes the data item of the data requested from the application 20 and information for identifying the collected data corresponding to the data item within the communication server 40. Examples of information for identifying the collected data within the communication server 40 include the identifier of the data item, the data tag name of the data item, the address of the location where the data item is stored, the folder path, the URL (Uniform Resource Locator), etc. The collection setting information also includes information (correspondence information described later) that associates the data type of the collected data stored by the platform 30 with the data type of the data processed by the application 20.

[0028] The engineering tool 10 can be used at a location remote from the production site where the device 50 is installed, and is connected to the platform 30 and the communication server 40 via a network line 60. Examples of the network line 60 are the Internet and a LAN (Local Area Network).

[0029] The engineering design tool 10 obtains a schema definition that defines the schema of the collected data from the communication server 40. This schema definition defines the schema of the collected data processed by the communication server 40, namely, the device schema (data model structure). In the following description, the schema definition of the collected data processed by the communication server 40 is referred to as the device schema definition. The device schema definition includes identifiers such as data tag names. The collected data processed by the communication server 40 is data that the communication server 40 can interpret.

[0030] The device architecture includes a data model of the collected data processed by the communication server 40. Therefore, the device architecture definition includes information defining the data model of the collected data. The data model of the collected data is obtained by modeling the template constituting the device architecture.

[0031] An example of application 20 is an application for monitoring equipment operation introduced for the purpose of improving productivity in the production site. Application 20 performs visualization of production operation status, etc. Application 20 analyzes the collected data collected from device 50, and diagnoses the operation status of the production site, etc. Application 20 processes data according to the architecture definition of the data processed by application 20, that is, the benchmark data. The benchmark data processed by application 20 is data that can be interpreted by application 20. The architecture definition of benchmark data is information that defines the architecture of the benchmark data processed by application 20, that is, the benchmark architecture. In the following description, the architecture definition of the benchmark data processed by application 20 is referred to as the benchmark architecture definition. The benchmark architecture definition can include identifiers such as data tag names, and can also include the data content of the benchmark data.

[0032] The baseline schema definition includes the data model of the baseline data processed by the application 20. Therefore, the baseline schema definition includes information defining the data model of the baseline data. The data model in the application 20 is obtained by modeling the template that constitutes the baseline schema.

[0033] Communication server 40 obtains device data from device 50 and stores it as collected data. If platform 30 requests collected data, communication server 40 transmits the collected data to platform 30. Examples of communication server 40 include MT Connect and OPC UA (Object Linking and Embedding for Process Control Unified Architecture) servers. When edge cross-connect is applied to data collection system 1, communication server 40 is accessed from data collectors supporting various communication protocols.

[0034] Device 50 deployed at a production site includes a device data output unit 51 for outputting device data, such as operating data, to an external device. Operating data is status monitoring data that enables application 20 to determine the operating status of device 50. Examples of operating data include data indicating the operating status of device 50, the operating mode of device 50, the processing status of the workpiece, and the presence or absence of an alarm.

[0035] The communication server 40 includes a device model management unit 41 and a collected data generation unit 42. The device model management unit 41 manages device architecture definitions. To manage these definitions, the device model management unit 41 manages, for example, XML (Extensible Markup Language) documents. In an XML document, each line contains a data item. These data items are assigned a data type representing the device architecture.

[0036] The data type is information that indicates the content of the collected data. Specifically, the data type defines the category, classification, or content of the collected data. In other words, the data type is the definition of the collected data. Examples of data types include coordinates, workpiece counts, and program names.

[0037] The device model management unit 41 stores a device architecture definition described in an XML document, and provides the device architecture definition to the engineering tool 10 in response to a request from the engineering tool 10 .

[0038] The collected data generation unit 42 collects device data from the device 50 based on the device architecture definition stored by the device model management unit 41. The collected data generation unit 42 generates collected data based on the device data based on the device architecture definition. Specifically, if the collected data generation unit 42 receives device data from the device data output unit 51, it shapes the device data into collected data in an output format corresponding to the communication protocol based on the device architecture definition of the device 50. The communication protocol here is a communication protocol used between the platform 30 and the communication server 40. The collected data generation unit 42 outputs the collected data generated by shaping to the platform 30 in response to a request from the platform 30. In addition, the collected data generation unit 42 can also output the generated collected data to the application 20 in response to a request from the application 20.

[0039] The data model of data generally used in communication devices for industrial purposes is roughly determined by the device category of the device 50, the supplier of the device 50, i.e., the device supplier, and the communication protocol. These data models are specified by the device architecture definition possessed by the communication server 40. In the device architecture definition, the data model adapted to each device 50 is constructed and defined using XML documents, etc., in accordance with the meta-structure of the data model determined by the communication protocol. Specifically, in the device architecture definition, for each data item to be collected, the data model is constructed and defined using the tag name or data identifier (ID (Identification) information), data type, subtype, data category, unit, etc. as basic attribute information. Subtypes are used to further classify data types. In the case where the data type is coordinates, examples of subtypes are workpiece coordinates and machine coordinates. Data categories are categories of the programming language that collects data, and examples of data categories are strings, integers, and dates.

[0040] Furthermore, multiple data models (device models) may be defined within a single device architecture. However, in principle, the data identifiers for each data item within each data model must not be duplicated. While basic attribute information such as data identifiers is semantically defined by the communication protocol, the interpretation of the data model and application-level connectivity within the actual product often depend on vendor implementation. Therefore, there is generally no strict answer regarding which data type or subtype is associated with which device data. Application-level connectivity indicates whether a connection is possible for data communication that maintains the data content.

[0041] An example of information that varies between vendors in the device architecture definition is the execution line information of the automatic operation program in the machine tool. The automatic operation program for the machine tool is identified by information such as the program name, sequence number, and module number, independent of the vendor of the numerical control (NC) device. This information is used by the NC device as information for the start point of automatic operation and for searching edit lines. Both the sequence number and module number are information that can be used to identify program lines.

[0042] Sometimes, in a certain communication protocol, the data type representing the row number only has the program name or the program line. In this case, in the baseline architecture, even if the program line is defined as a module number, in the case of a certain supplier (supplier A), it is possible that the data type of the program line is defined as a serial number. In addition, in the case of a certain supplier (supplier B), it is possible that the data type of the program line is defined as an extended data type. Therefore, even if the same data type is defined in the device architecture definition processed by the communication server 40, sometimes different data is collected for each device 50. That is, the content of the data defined by the application 20 in the baseline architecture definition is sometimes different from the content of the data defined by each supplier in the device architecture definition.

[0043] Furthermore, sometimes, extended definitions, which can be called device vendor-specific specifications, are used instead of the standard data types defined by the communication protocol. In these cases, even if the data types are intended to represent the same program line, one (Vendor A's device) may have a data item with a data type representing a module number, while the other (Vendor B's device) may have a data item representing a serial number. In other words, the correspondence between data types and data content (the meaning of the data) is established in various ways. Therefore, application 20 cannot use data types as a clue to treat data from Vendor A's device and data from Vendor B's device as identical data.

[0044] The platform 30 has a collection data setting unit 31 and a collection data storage unit 32. The collection data setting unit 31 receives the collection setting information of the collection data to be stored from the engineering design tool 10, and manages the received collection setting information. In the collection setting information, the data type of the device architecture (hereinafter referred to as the device data type) corresponding to the data type of the benchmark architecture (hereinafter referred to as the benchmark data type) is set. That is, in the collection setting information, the benchmark data type assigned to the data item of the benchmark data is associated with the device data type assigned to the data item of the device data. Specifically, in the collection setting information, the identifier of the data item corresponding to the benchmark data type is associated with the identifier of the data item corresponding to the device data type. In addition, in the collection setting information, the data item of the benchmark data type may be associated with the identifier of the data item corresponding to the device data type.

[0045] When a specific reference data type is specified as a collection target by the application 20, the collected data storage unit 32 extracts the device data type corresponding to the specified reference data type from the collection settings information. The collected data storage unit 32 requests the collected data generation unit 42 of the communication server 40 for the collected data corresponding to the extracted device data type. For example, the collected data storage unit 32 sends the identifier of the data item of the device data type to the collected data generation unit 42, thereby requesting the collected data generation unit 42 for the collected data corresponding to the identifier. In this manner, the collected data storage unit 32 requests the collected data generation unit 42 for the collected data according to the collection settings information.

[0046] The collected data storage unit 32 receives and stores the collected data sent from the collected data generation unit 42. In response to a request from the application 20, the collected data storage unit 32 sends the stored collected data to the application 20. The data type of the collected data sent by the collected data storage unit 32 generally conforms to the definition of data types or subtypes that can be processed by the application 20. The collected data storage unit 32 sends the collected data to the application 20 using a common communication protocol.

[0047] The engineering design tool 10 includes a device model editing unit 11, a conversion candidate providing unit 12, and a device profile output unit 13. The device model editing unit 11 obtains a device architecture definition including a device data type from the device model management unit 41 of the communication server 40. The device data type is used when editing correspondence information indicating the correspondence between the device data type and the reference data type.

[0048] The device model editor 11 edits the correspondence information by editing the device data type within the device architecture definition. The device model editor 11 edits the correspondence information when a user inputs an edit instruction into the device model editor 11. The user edits the correspondence information while referencing the system information (device 50 information, application type, and communication protocol type) described later.

[0049] The correspondence information indicates which data item identifiers within the device architecture should be collected from communication server 40, relative to the identifiers of the data items that need to be converted in the base architecture. In other words, the correspondence information indicates the correspondence between the base architecture definition and the device architecture definition, that is, the correspondence between the architecture definitions.

[0050] The device model editor 11 edits the device models and other items included in the device architecture definition based on user operations, thereby editing the correspondence information. Here, the platform 30 needs to collect data from the device 50 that matches the data type of each data item defined in the base architecture definition.

[0051] However, as mentioned above, the baseline data required by an application is often data with similar or identical meanings between the baseline architecture and the device architecture, but the data type definitions may differ. In such cases, without modifying the application, a system integrator who fully understands the specifications of both the baseline architecture definition and the device architecture definition must modify the device architecture definition used in the communication server and the collection settings information used in the platform. Specifically, the system integrator must modify the device architecture definition for the communication server and the collection settings information for the platform to ensure that the device collects data that matches the data type required by the application.

[0052] In this embodiment, the user edits the correspondence information using the device model editing unit 11, referring to the data type or subtype of the data items that need to be converted between the baseline architecture and the device architecture. In this case, the user edits the device architecture definition (device model, etc.) corresponding to the baseline architecture definition while referencing the information learned from the results of editing the correspondence information, thereby editing the correspondence information.

[0053] The base architecture definition can be input by the user into the device model editor 11, or it can be obtained by the device model editor 11 from an external device such as the application 20. As described above, the correspondence information indicates the correspondence between data types. Therefore, in the example of an automatic operation program for a working machine, if it is necessary to collect device data from supplier A, the device model editor 11 must define a mapping definition for each program line in the base architecture, not for the data item with the serial number, but for the data item with the data identifier representing the module number.

[0054] The device model editor 11 sends the edited result including the edited content of the correspondence information to the conversion candidate provider 12. The conversion candidate provider 12 sends the correspondence information, which maps the device data type to the reference data type for each data item of the collected data, to the device profile output unit 13.

[0055] Furthermore, the conversion candidate providing unit 12 uses the information used to edit the correspondence information and the edited results of the correspondence information to learn conversion rules. In other words, the conversion candidate providing unit 12 learns conversion rules based on the user's edit history of the correspondence information. Conversion rules are rules for converting from a reference data type to a device data type. In other words, conversion rules are rules for associating the reference data type with the device data type. Therefore, learning conversion rules corresponds to learning candidates for the device data type corresponding to the reference data type (described later as conversion candidates).

[0056] In the following description, the information used to edit the correspondence information is referred to as system information. System information includes at least one of the following: "device information" (information about the device 50), "application type" (category of the application 20), and "communication protocol type" (category of the communication protocol between the communication server 40 and the platform 30). "Device information" includes at least one of the following: "device manufacturer type" (category of the manufacturer that manufactured the device 50), "device type" (category of the device 50), and "device structure" (configuration of the device 50). The result of editing the correspondence information is the association of the reference data type with the device data type.

[0057] The "device information," "communication protocol type," and "application type" are input to the conversion candidate providing unit 12 by, for example, a user. Alternatively, the conversion candidate providing unit 12 may extract at least one of the "device information" and "communication protocol type" from the device architecture definition. Furthermore, the conversion candidate providing unit 12 may obtain the "application type" from the application 20.

[0058] The conversion candidate providing unit 12 observes the system information and the reference data type as state variables. In addition, the application 20 obtains the teacher data. Moreover, the conversion candidate providing unit 12 learns the conversion rules according to the data group created based on the combination of the state variables and the teacher data. The teacher data is a device data type associated with the reference data type by the user. In the following description, the device data type associated with the reference data type by the user is referred to as the "converted data type". The "converted data type" as the teacher data is the device data type (data type conversion result) actually set by the user in a manner corresponding to the reference data type.

[0059] Since the device model editing unit 11 edits the device model, the correspondence information includes the edited device model (edited model). The conversion candidate providing unit 12 learns conversion rules, i.e., conversion candidates, that can output device data types whose contents are consistent or similar between the reference data type and the device data type.

[0060] The conversion candidate providing unit 12 observes state variables for each device 50, each communication protocol between the communication server 40 and the platform 30, or each application 20. The conversion candidate providing unit 12 observes state variables based on the device architecture definition obtained from the device model management unit 41 and the edited content in the device model editing unit 11. Specifically, the conversion candidate providing unit 12 observes, for example, "device information," "application type," and "communication protocol type," which are system information, as state variables.

[0061] The conversion candidate providing unit 12 also observes the reference data type set in the correspondence information as a state variable. That is, the conversion candidate providing unit 12 observes the data type conversion result (mapping result) in the device model editing unit 11, that is, the reference data type in the correspondence information, as a state variable.

[0062] The transformation candidate providing unit 12 uses the transformation rules obtained through learning to calculate the candidates of the device data type corresponding to the reference data type (hereinafter referred to as transformation candidates). In other words, the transformation candidate providing unit 12 of the present embodiment calculates the candidates (transformation candidates) of the device data type set in the correspondence information based on the history of the editing results of the correspondence information. The correspondence information containing the device data type mapped to the reference data type includes identifiers such as the data tag name that the application 20 can use to identify each collected data for each data item. In the correspondence information, the identifiers of the data items of the collected data processed by the communication server 40 are associated with the identifiers of the data items of the data processed by the application 20. Among the identifiers included in the correspondence information, the identifiers of the data items processed by the communication server 40 are identifiers included in the device architecture, and the identifiers of the data items processed by the application 20 are identifiers included in the reference architecture.

[0063] When the user specifies a base data type to be mapped, the candidate transformation provider 12 estimates transformation candidates corresponding to the specified base data type. Based on the learned transformation rules, the candidate transformation provider 12 estimates the transformation candidates. The candidate transformations are device data types that are candidates for transformation relative to the base data type. In other words, the candidate transformations are candidates for transformations of the data type of the device architecture associated with the base architecture. The candidate transformation provider 12 sends the candidate transformations to the device model editor 11.

[0064] Furthermore, when the correspondence information for output is sent from the device model editing unit 11 , the conversion candidate providing unit 12 outputs the correspondence information to the device profile output unit 13 .

[0065] The device configuration file output unit 13 generates collection setting information using the correspondence information, converts the collection setting information into a protocol as needed, and sends the information to the collection data setting unit 31 of the platform 30 .

[0066] The engineering design tool 10 displays the device architecture definition, device data type, system information, reference architecture definition, reference data type, conversion rules, edited results of correspondence information, conversion candidates, etc. on a display device such as a liquid crystal monitor (not shown).

[0067] The user edits the correspondence information while referring to the conversion candidates displayed on the display device. In the data collection system 1, the user edits the correspondence information and the engineering design tool 10 learns the conversion rules repeatedly.

[0068] With such a configuration, the engineering tool 10 can provide conversion candidates corresponding to the reference data type even to a user who has insufficient knowledge of both the reference data type and the device data type.

[0069] Next, the detailed structure of the conversion candidate providing unit 12 will be described. Figure 2 The diagram shows the configuration of a conversion candidate providing unit included in the engineering design tool according to the embodiment. The conversion candidate providing unit 12 includes a data selecting unit 121 , a conversion rule learning unit 122 , a conversion candidate estimating unit 123 , and a device model correcting unit 124 .

[0070] The data selection unit 121, the conversion rule learning unit 122, the conversion candidate estimation unit 123, and the device model correction unit 124 are connected to the device model editing unit 11. Furthermore, the conversion candidate estimation unit 123 is connected to the data selection unit 121, the conversion rule learning unit 122, and the device model correction unit 124. Furthermore, the device model correction unit 124 is connected to the device configuration file output unit 13.

[0071] When the user edits the correspondence information, the device model editing unit 11 sends the correspondence information to the device model correction unit 124. In addition, when the conversion candidate providing unit 12 learns the conversion rule (conversion candidate), the device model editing unit 11 sends the correspondence information indicating the edited result to the conversion rule learning unit 122. In addition, when the conversion candidate providing unit 12 estimates the conversion candidate, the device model editing unit 11 sends the edited correspondence information to the data selection unit 121. In addition, the device model editing unit 11 obtains the conversion candidate (in the conversion candidate providing unit 12) from the conversion candidate providing unit 12. Figure 2 , the diagram shows "transformation candidates").

[0072] The transformation rule learning unit 122 is a machine learning device that observes system information and reference data types as state variables and learns transformation rules, which are learning models, based on the state variables and the transformed data types. The transformation rule learning unit 122 uses the results of transforming the data types of the device architecture relative to the reference architecture to learn transformation rules for the data types corresponding to the device 50. The transformation rule learning unit 122 outputs the learned transformation rules to the candidate transformation estimation unit 123.

[0073] When the conversion candidate estimation unit 123 estimates conversion candidates corresponding to the reference data type, the data selection unit 121 obtains editing information indicating the editing status in the device model editing unit 11 from the device model editing unit 11 .

[0074] The data selection unit 121 obtains editing information from the device model editing unit 11, including the reference data type and system information for the reference architecture being edited. From the editing information, the data selection unit 121 selects and extracts a reference data type that maps to the device data type. The reference data type that maps to the device data type is a reference data type whose data type content or data tag name differs between the device data type and the reference data type.

[0075] The reference data type and system information obtained by the data selection unit 121 from the device model editing unit 11 are the same information as the state variables observed by the state observation unit, described later. Hereinafter, the reference data type and system information extracted by the data selection unit 121 are referred to as estimation data. The data selection unit 121 outputs the estimation data to the conversion candidate estimation unit 123.

[0076] The transformation candidate estimation unit 123 estimates transformation candidates to be collected from the device 50 based on the transformation rule, which is the learning model output from the transformation rule learning unit 122, and the estimation data output from the data selection unit 121. The transformation candidate estimation unit 123 may also estimate transformation candidates for a device different from the device 50 used to learn the transformation rule. Furthermore, the transformation candidate estimation unit 123 may also estimate transformation candidates for an application different from the application 20 used to learn the transformation rule. The transformation candidate estimation unit 123 outputs the estimated transformation candidates to the device model editing unit 11 and the device model correction unit 124.

[0077] The device model correction unit 124 determines whether there are any editing defects, such as omissions, in the device model editing unit 11 based on the transformation candidates sent from the transformation candidate estimation unit 123 and the correspondence information sent from the device model editing unit 11. If an editing defect exists in the device model editing unit 11, the device model correction unit 124 automatically corrects the correspondence information and outputs the automatically corrected correspondence information to the device configuration file output unit 13. An example of an editing defect is the omission of a device data type in the correspondence information.

[0078] The content of the device architecture definition in this embodiment is roughly determined by the combination of the type of device 50 for determining the collection data that can be collected, the type of application 20 for determining the collection data used, the supplier of the device 50, the supplier of the application 20, and the type of communication protocol.

[0079] Originally, mapping was required between the data items in the device architecture and the data items in the reference architecture. In this embodiment, the engineering design tool 10 observes "device information" such as "device manufacturer type," "device type," and "device structure" used to characterize the device architecture, the "application type" used to characterize the reference architecture, and the "communication protocol type" used to identify the device architecture as state variables. This improves the learning accuracy of the transformation rules used in the estimation of transformation candidates, allowing the engineering design tool 10 to teach users appropriate transformation candidates for device data types.

[0080] The first user among the users edits the correspondence information before learning the conversion rule, and the second user edits the correspondence information based on the estimated conversion candidates.

[0081] The device 50 targeted for transformation rule learning is the first device, and the device 50 targeted for transformation candidate estimation is the second device. The data collected from the first device is the first collected data, and the data collected from the second device is the second collected data.

[0082] The application 20 to be learned as the transformation rule is the first application among the applications, and the application 20 to be estimated as a transformation candidate is the second application among the applications. Furthermore, the reference data that can be interpreted by the first application is the first reference data, and the reference data that can be interpreted by the second application is the second reference data.

[0083] Furthermore, the correspondence information edited by the first user is the first correspondence information, and the correspondence information edited by the second user is the second correspondence information. In the first correspondence information, the first device data type is associated with the first reference data type, and in the second correspondence information, the second device data type is associated with the second reference data type.

[0084] The system information used to edit the first correspondence information is the first system information, and the system information used to edit the second correspondence information is the second system information. The first system information includes the first device information, the first application type, and the first communication protocol type. The second system information includes the second device information, the second application type, and the second communication protocol type.

[0085] Furthermore, the first user and the second user may be different users or the same user. Furthermore, the first device and the second device may be different devices or the same device. Furthermore, the first application and the second application may be different applications or the same application.

[0086] Figure 3 This is a diagram showing a configuration of a conversion rule learning unit included in the engineering design tool according to the embodiment. Figure 4 This is a diagram showing the structure of a neural network used in the engineering design tool according to the embodiment.

[0087] The conversion rule learning unit 122 includes a data acquisition unit 71, a state observation unit 72, and a learning unit 73. The data acquisition unit 71 acquires the training data from the device model editing unit 11. The training data is the device data type included in the edited correspondence information (the result of the edited correspondence information), that is, the converted data type. The data acquisition unit 71 sends the training data to the learning unit 73.

[0088] The state observation unit 72 obtains the system information from the device model editing unit 11 and extracts the reference data type from the edited correspondence information. The state observation unit 72 observes the system information and the reference data type as state variables and sends the system information and the reference data type to the learning unit 73.

[0089] Learning unit 73 learns transformation rules for deriving transformation candidates (learning content) based on a data set created based on a combination of system information and reference data types output from state observation unit 72, and teacher data, i.e., the transformed data types. Here, a data set is data that associates state variables with teacher data.

[0090] Furthermore, the transformation rule learning unit 122 is not limited to being located within the engineering design tool 10. The transformation rule learning unit 122 may also be located in a device external to the engineering design tool 10. The transformation rule learning unit 122 may also be located in a device that is connectable to the engineering design tool 10 via the network line 60. In other words, the transformation rule learning unit 122 may be a separate component connected to the engineering design tool 10 via the network line 60. Furthermore, the transformation rule learning unit 122 may also reside on a cloud server.

[0091] The transformation rule learning unit 122 learns transformation candidates based on the data types (device data types) of the device models included in the device architecture definitions collected from the communication server 40, using so-called supervised learning, for example, according to a neural network model. Here, supervised learning refers to a model in which a large number of sets of input and result (label) data are fed to a machine learning device to learn features derived from these data sets and infer the results based on the inputs.

[0092] The neural network is composed of the following layers, namely, the input layer X1~Xp (p is a natural number) composed of multiple neurons, the intermediate layer (hidden layer) Y1~Yq (q is a natural number) composed of multiple neurons, and the output layer Z1~Zr (r is a natural number) composed of multiple neurons. The intermediate layer Y1~Yq can also be 1 layer or more than or equal to 2 layers. The input layer X1~Xp is connected to the intermediate layer Y1~Yq, and the intermediate layer Y1~Yq is connected to the output layer Z1~Zr. In addition, Figure 4 The connection between the input layers X1 to Xp and the intermediate layers Y1 to Yq shown is an example, and each input layer X1 to Xp can be connected to any intermediate layer Y1 to Yq. Figure 4 The connection between the intermediate layers Y1 to Yq and the output layers Z1 to Zr shown is an example, and each of the intermediate layers Y1 to Yq can be connected to any output layer Z1 to Zr.

[0093] For example, in Figure 4 In the case of a three-layer neural network as shown, when multiple inputs are input to input layers X1 to Xp, these values are multiplied by weights A1 to Aa (a is a natural number) and then input to intermediate layers Y1 to Yq. The values input to intermediate layers Y1 to Yq are further multiplied by weights B1 to Bb (b is a natural number) and then input to output layers Z1 to Zr, from which they are output. The output results are shown here as transformation candidates T1 to T3. These output results vary depending on the values of weights A1 to Aa and B1 to Bb.

[0094] The neural network of this embodiment learns the conversion rules through so-called taught learning according to the data group created based on the combination of system information and reference data type observed by the state observation unit 72 and the converted data type acquired by the data acquisition unit 71.

[0095] That is, the neural network learns by adjusting weights A1 to Aa and B1 to Bb so that system information and reference data types are input to input layers X1 to Xp and the results output from output layers Z1 to Zr are close to the transformed data types.

[0096] Information input to the input layers X1 to Xp includes, for example, "communication protocol type," "application type," "reference data type n" (n is a natural number), "device manufacturer type," "device type," and "device configuration."

[0097] Examples of "application types" include operations monitoring applications, process management applications, quality management applications, and maintenance applications. Examples of "device types" include machining centers, multi-processing machines, laser processing machines, and electrical discharge machines. Examples of "device configuration" include the number of systems, axis information, and peripheral equipment. "Conversion candidates" are "device data types" that are highly likely to be associated with the "base data type."

[0098] When receiving new system information and a new reference data type, the conversion candidate providing unit 12 uses the learned conversion rules ( Figure 4 The transformation candidates are calculated using the neural network shown in FIG.

[0099] Figure 5 This is a flowchart illustrating the operational flow of machine learning performed by an engineering design tool according to an embodiment. The transformation rule learning unit 122 obtains learning data. Specifically, the transformation rule learning unit 122 obtains the baseline architecture definition, the device architecture definition, and the user's edited correspondence information from the device model editing unit 11 as learning data (step S101).

[0100] The transformation rule learning unit 122 learns the relationship between the data types before and after the transformation from the learning data, generating a transformation rule, or learning model (step S102). The relationship between the data types before and after the transformation is correspondence information indicating the correspondence between the reference data type and the device data type. The transformation rule learned by the transformation rule learning unit 122 is a learning model that can estimate transformation candidates that can be collected from the device 50, based on the reference data type that can be interpreted by the application 20. The transformation rule learning unit 122 learns the transformation rule based on the learning data, for example, through supervised learning.

[0101] In addition, neural networks can also learn transformation candidates through so-called unsupervised learning. Unsupervised learning is a method of learning how the input data is distributed by providing only a large amount of input data to a machine learning device. Even if the corresponding teacher data (output data) is not provided, learning is performed by compressing, classifying, and shaping the input data. In unsupervised learning, data groups with similar features can be clustered, etc. In unsupervised learning, a certain benchmark is set by using the results of the clustering, and the output distribution that optimizes the benchmark is performed, so that the output can be predicted. In addition, as a problem setting between unsupervised learning and supervised learning, there is also learning called semi-supervised learning. Semi-supervised learning is a learning method in which there is only a group of input and output data, and the rest is only input data.

[0102] Furthermore, the learning unit 73 can also use deep learning (Deep Learning) as a learning algorithm to learn the extracted feature quantity itself. In addition, the learning unit 73 can also perform machine learning according to other well-known methods, such as genetic programming, functional logic programming, support vector machines, etc.

[0103] Next, a description will be given of a process in which the engineering tool 10 calculates conversion candidates using the conversion rule. Figure 6 This is a flowchart showing the flow of operations when the engineering design tool according to the embodiment performs data estimation.

[0104] The data selection unit 121 obtains the reference data type and system information currently being edited by the device model editing unit 11 as estimation data (step S201). The reference data type being edited is the reference data type before being associated with the device data type (before conversion). The data selection unit 121 outputs the estimation data to the conversion candidate estimation unit 123.

[0105] The conversion candidate estimation unit 123 receives the estimation data output from the data selection unit 121. The conversion candidate estimation unit 123 also receives the conversion rule, which is a learning model output from the conversion rule learning unit 122.

[0106] The conversion candidate estimation unit 123 estimates the conversion candidates of the device data type using the estimation data and the learning model (step S202). An example of the learning model is Figure 4 In the neural network shown, estimation data is input to the neural network's input layers X1 through Xp. Specifically, system information such as the communication protocol type and application type is input to the neural network's input layers X1 through Xp. Data output from the neural network's output layers Z1 through Zr are transformation candidates.

[0107] The transformation candidate estimation unit 123 teaches the device model editing unit 11 that is editing the reference data type about the transformation candidates (step S203). That is, the transformation candidate estimation unit 123 teaches the device model editing unit 11 that is editing the device data type (data model of the device 50) that should be compatible with the reference data type, about the data that can be collected from the device 50, i.e., the transformation candidates. Specifically, the transformation candidate estimation unit 123 sends the estimated transformation candidates to the device model editing unit 11. The user inputs a selection instruction to select the desired device data type from the transformation candidates into the device model editing unit 11. The device model editing unit 11 associates the device data type with the reference data type being edited according to the selection instruction. Thus, the device model editing unit 11 edits the correspondence information.

[0108] In this way, when multiple conversion candidates are taught, the user's selection operation is reflected in the device model editing unit 11. The conversion rule learning unit 122 performs so-called reinforcement learning, that is, assigning positive evaluations to conversion candidates selected by the user and negative evaluations to unselected candidates. In other words, the conversion rule learning unit 122 relearns the conversion rules using the device data type selected by the user. This allows the conversion rule learning unit 122 to provide conversion rules that match the actual usage frequency of the device data type.

[0109] The device model editing unit 11 sends the correspondence relationship information edited by the user to the device model correction unit 124. In addition, the conversion candidate estimation unit 123 sends the conversion candidates to the device model correction unit 124.

[0110] The device model correction unit 124 uses the conversion candidates, which are the output results of the device model editing unit 11, to modify the device architecture definition (step S204). For example, if the device architecture definition contains defects such as omissions due to editing operations performed by the device model editing unit 11, the device model correction unit 124 automatically corrects the defective device architecture definition to a device architecture definition that complies with appropriate conversion rules. The device model correction unit 124 outputs the correspondence information, including the necessary modifications to the device architecture definition, to the device configuration file output unit 13.

[0111] The device configuration file output unit 13 generates collection setting information including correspondence information. The device configuration file output unit 13 performs protocol conversion on the collection setting information as needed and sends it to the collection data setting unit 31 of the platform 30. Thus, the collection data setting unit 31 sets the collection setting information. Moreover, when there is a data request from the application 20, the collection data storage unit 32 of the platform 30 makes a data request to the communication server 40 in accordance with the collection setting information. Specifically, the collection data storage unit 32 sets the data with the request from the application 20 as data of the reference data type, and requests the communication server 40 for data of the device data type corresponding to the reference data type. The collection data storage unit 32 obtains data of the device data type from the communication server 40 by sending the identifier of the device data type to the communication server 40. The collection data storage unit 32 sends the obtained data of the device data type to the application 20.

[0112] Through these mechanisms, the user of the engineering design tool 10 of this embodiment can perform data collection settings on the platform 30 that are compatible with the data model of the application 20 without understanding the specifications of the baseline architecture definition in the application 20 (the data model in the application 20) and the device architecture definition in the device 50 (the data model in the device 50).

[0113] Even if the device architecture definition (the definition of data that can be output externally) differs depending on "device information," "communication protocol type," or "application type," the platform 30 can collect the collected data in a manner that provides a unique data definition for the application 20. This allows the application 20 to centrally utilize the collected data collected by the platform 30 without relying on "device information," "communication protocol type," or "application type."

[0114] However, typically, when achieving data matching between applications and devices through an IoT platform or communication server, each device is configured at the production site, taking into account the data specifications of both the device and the application. This operation requires a significant amount of man-hours commensurate with the scale of the system, and is particularly problematic when handling communication protocols for which configuration tools are not widely available. Furthermore, when achieving data matching between applications and devices through an IoT platform or communication server, system integrators with a thorough understanding of the data specifications of both devices and applications are involved, centrally performing configuration as part of system construction. This, in turn, increases the cost and lead time associated with system construction.

[0115] On the other hand, in this embodiment, since the engineering design tool 10 estimates the conversion candidates, the user can easily edit the correspondence information in a short time. Therefore, the data collection system 1 can be constructed at low cost and in a short time.

[0116] Furthermore, the data collection system 1 can also be applied to data utilization in IT systems higher than application 20, such as manufacturing execution systems (MES) and enterprise resource planning (ERP). Furthermore, the data collection system 1 can be used for data analysis using edge computing near the production site, and diagnostic results obtained from edge computing can be fed back to the device 50 in real time. Thus, the data collection system 1 can achieve high-efficiency operation of production equipment.

[0117] As described above, according to the embodiment, the engineering tool 10 learns the conversion rules based on the results of editing the correspondence information, and uses the conversion rules to infer conversion candidates from the reference data type that the application 20 can interpret to the device data type. This allows the engineering tool 10 to provide conversion candidates for the device data type that corresponds to the reference data type that the application 20 can interpret. Therefore, even when the reference data type that the application 20 can interpret is not associated with the device data type, conversion candidates corresponding to the reference data type can still be provided.

[0118] Data collection system 1 can automatically configure data collection settings on platform 30 based on learned conversion rules, thereby reducing the effort required for data conversion on platform 30. Even when new devices are connected or new communication protocols are required to be compatible, data collection system 1 does not require modifications in application 20, enabling rapid connection configuration and system construction.

[0119] Furthermore, because the engineering design tool 10 is separate from the platform 30, it can edit and output the conversion rules for the collected data even at a location remote from the device 50. Consequently, the data collection system 1 can flexibly share the roles of suppliers in the setup process, thereby reducing system construction costs and shortening system setup time.

[0120] Furthermore, the device architecture definition corresponding to the conversion rule can be output as a device configuration file obtained using a standard modeling description language, and thus can be applied to various industrial platforms.

[0121] Here, the hardware configuration of a computer that runs the engineering tool 10 will be described. Figure 7This is a diagram showing a first example of a hardware configuration for realizing a computer that runs the engineering design tool according to the embodiment. Figure 8 This is a diagram showing a second example of a hardware configuration for realizing a computer that runs the engineering design tool according to the embodiment.

[0122] The computer running the engineering design tool 10 can be Figure 7 The processor 501, memory 502, and interface 504 shown are implemented. Processor 501 is a CPU (Central Processing Unit, also known as FPGA (Field-Programmable Gate Array), central processing unit, processing unit, computing unit, microprocessor, microcomputer, processor, DSP (Digital Signal Processor)), system LSI (Large Scale Integration), etc. Memory 502 is RAM (Random Access Memory), ROM (Read Only Memory), etc.

[0123] Memory 502 stores programs that execute the functions of engineering tool 10. Processor 501 reads and executes the programs stored in memory 502, thereby executing the processing performed by engineering tool 10. It can be said that the programs stored in memory 502 cause the computer to execute multiple commands corresponding to the flow or method of engineering tool 10. Memory 502 also serves as temporary storage when processor 501 executes various processes.

[0124] The program executed by the processor 501 may be a computer program product having a computer-readable and non-transitory recording medium containing a plurality of instructions for performing data processing. In other words, the engineering design tool 10 may be implemented by a computer-readable medium having the program recorded thereon.

[0125] also, Figure 7 The processor 501 and memory 502 shown can also be replaced by Figure 8 Processing circuit 503 is shown. Processing circuit 503 may be, for example, a single circuit, a complex circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof. Furthermore, the functions of engineering design tool 10 may be partially implemented by dedicated hardware and partially implemented by software or firmware.

[0126] In addition, at least one of the application 20 , the platform 30 , and the communication server 40 may be implemented by the same hardware structure as that of the computer that runs the engineering design tool 10 .

[0127] The configuration shown in the above embodiment is merely an example of the content of the present invention, and may be combined with other known technologies. Part of the configuration may be omitted or modified without departing from the scope of the present invention.

[0128] Description of the label

[0129] 1 Data collection system, 10 Engineering design tool, 11 Device model editing unit, 12 Transformation candidate providing unit, 13 Device configuration file output unit, 20 Application, 30 Platform, 31 Collected data setting unit, 32 Collected data storage unit, 40 Communication server, 41 Device model management unit, 42 Collected data generation unit, 50 Device, 51 Device data output unit, 60 Network line, 71 Data acquisition unit, 72 State observation unit, 73 Learning unit, 121 Data selection unit, 122 Transformation rule learning unit, 123 Transformation candidate estimation unit, 124 Device model correction unit, 501 Processor, 502 Memory, 503 Processing circuit, 504 Interface, T1 to T3 Transformation candidates, X1 to Xp input layer, Y1 to Yq intermediate layer, Z1 to Zr output layer.

Claims

1. A storage medium, characterized in that: A program is provided to cause the processor to execute the following processing: an editing process of editing, based on an instruction from the first user, first correspondence information indicating a correspondence between a data type of first collected data collected from the first device, namely, a first device data type, and a data type of first reference data interpretable by the first application, namely, a first reference data type; as well as A transformation candidate providing process is performed, which learns the rule for transforming the first reference data type to the first device data type, i.e., the transformation rule, based on the editing result of the first correspondence information, and uses the transformation rule to infer a transformation candidate from the data type of the second reference data that can be interpreted by the second application, i.e., the second reference data type, to the data type of the second collected data collected from the second device, i.e., the second device data type.

2. The storage medium according to claim 1, wherein The editing process edits the first correspondence information based on first system information, the first system information including first device information, at least one of a type of a first communication protocol corresponding to the first device, and a type of the first application. The conversion candidate providing process estimates the conversion candidate based on second system information including second device information, a type of a second communication protocol corresponding to the second device, and at least one of a type of the second application.

3. The storage medium according to claim 2, wherein: The first device information includes at least one of a type of a device manufacturer that manufactured the first device, a type of the first device, and a structure of the first device. The second device information includes at least one of a type of a device manufacturer that manufactured the second device, a type of the second device, and a configuration of the second device.

4. The storage medium according to claim 2, wherein: The conversion candidate providing process executes a conversion rule learning process for learning the conversion rule. The transformation rule learning process performs: State observation processing, observing state variables including the first system information and the first reference data type; Data acquisition processing, acquiring the first device data type; and The learning process learns the conversion rule according to a data group created based on a combination of the state variable and the first device data type.

5. The storage medium according to any one of claims 1 to 4, characterized in that If the second user selects the second device data type corresponding to the second reference data type from the conversion candidates, In the editing process, the second correspondence information indicating the correspondence between the selected second device data type and the second reference data type is edited. In the conversion candidate providing process, the conversion rule is relearned based on the editing result of the second correspondence information.

6. The storage medium according to any one of claims 1 to 4, characterized in that The first correspondence information is information corresponding to a device architecture definition representing an architecture definition of the first collected data and a reference architecture definition representing an architecture definition of the first reference data.

7. The storage medium according to claim 5, wherein: The first correspondence information is information corresponding to a device architecture definition representing an architecture definition of the first collected data and a reference architecture definition representing an architecture definition of the first reference data.

8. A learning device, characterized in that: have: a state observation unit that, when correspondence information is edited based on an instruction from a user, observes a state variable, the correspondence information indicating a correspondence between a data type of collected data collected from a device, i.e., a device data type, and a data type of reference data interpretable by an application, i.e., a reference data type, the state variable including the reference data type included in the correspondence information and system information referenced when editing the correspondence information; a data acquisition unit configured to acquire the device data type included in the correspondence information; and The learning unit learns a conversion rule, which is a rule for converting from the reference data type to the device data type, based on a data group created based on a combination of the state variable and the device data type.

9. A data collection system, characterized in that: have: a communication server that collects data from the device; an application for calculating status information of a device equipped with the apparatus based on the collected data; a platform that obtains, from the communication server, collected data corresponding to the data requested by the application, and transmits the data to the application based on correspondence information indicating a correspondence between a data type of the collected data, i.e., a device data type, and a data type of reference data interpretable by the application, i.e., a reference data type; as well as An engineering design tool that edits the correspondence information based on an instruction from a user, The engineering design tool has: an editing unit that edits, based on an instruction from a first user among the users, correspondence information indicating a correspondence between a data type of first collected data collected from a first device among the devices, namely, a first device data type, and a data type of first reference data interpretable by a first application among the applications, namely, a first reference data type; as well as A conversion candidate providing unit learns a rule for converting the first reference data type to the first device data type, i.e., a conversion rule, based on the editing result of the correspondence information, and uses the conversion rule to infer conversion candidates from the data type of the second reference data that can be interpreted by the second application in the application, i.e., the second reference data type, to the data type of the second collected data collected from the second device in the device, i.e., the second device data type.

10. The data collection system according to claim 9, wherein: The conversion candidate providing unit sends the conversion candidates to the editing unit, If a second user among the users selects the second device data type corresponding to the second reference data type from the conversion candidates, The editing unit edits the second correspondence information indicating the correspondence between the selected second device data type and the second reference data type. The platform acquires, from the communication server, the collected data corresponding to the data requested by the second application based on the correspondence information transmitted from the editing unit, and transmits the acquired data to the second application.

Images