Method, device and system for setting up industrial devices

Abstract

The invention relates to a computer-implemented method, a device and a system for training pre-trained AI (10) in order to generate trained AI which is suitable for generating a configuration file (40) for configuring an industrial device (50), wherein the method (10) comprises the use of a processor (60) and a memory (65), comprising: a receiving function for receiving training data (20), wherein the training data (70) comprise an instruction file (81) and a training configuration file (73), wherein the instruction file (81) is related to the training configuration file (73); and a processing function for generating context information (30) on the basis of the training data (70), wherein the context information (90) is designed for use by the trained AI; and an output function for outputting the context information (98).

Description

[0001] Method, device and system for setting up industrial equipment

[0002] TECHNICAL AREA

[0003] This disclosure is aimed at simplifying the parameterization of complex industrial devices that can be integrated into control systems. Industrial devices can, for example, be designed as measuring instruments. They are typically configured with a multitude of parameters, which are selected according to the desired measurement results and their integration into the respective control system.

[0004] BACKGROUND

[0005] Electrical or electronic controls are required for managing system solutions for energy supply, including wind and solar power, as well as for equipment and machine manufacturing. Such controls for plants or processes consist of a variety of industrial devices and other components. They are often housed in control cabinets. An industrial device is an electronic component comprised of hardware and firmware. It can, for example, be a measuring instrument. For use in the control system, the firmware of an industrial device requires configuration data that takes into account both the available input variables and the desired output variables, enabling the industrial device to output appropriate user data.

[0006] The parameterization of industrial equipment when building a new control system or modernizing an existing one with new components is both complex and error-prone due to the large number of parameters to be considered and their interdependencies. Furthermore, the parameter sets for a given or comparable industrial device are similar to one another due to their function. Typically, these parameter sets exhibit numerous mutual dependencies.

[0007] Therefore, there is a need to simplify the parameterization of industrial equipment when setting up a new control system or when modernizing an existing one.

[0008] SUMMARY

[0009] The need is met by the features of the independent claims. Advantageous embodiments and further developments of the invention are specified in the dependent claims.

[0010] Exemplary embodiments of the invention, which can be optionally combined with one another, are disclosed below with partial reference to the figures.

[0011] A first aspect of the invention relates to a computer-implemented method for training a pre-trained AI (artificial intelligence) to generate a trained AI. The trained AI is capable of generating a configuration file for configuring an industrial device. The method comprises the use of a processor and memory. Furthermore, the method includes a receive function for receiving training data. This training data comprises an instruction file and a training configuration file, the instruction file being referenced to the training configuration file. The method further includes a processing function for generating context information based on the training data. This context information is designed for use by the trained AI. Finally, the method includes an output function that outputs the context information.

[0012] In general, an AI model comprises mathematical structures and is characterized by an architecture that defines the structured interaction between model elements in such a way that the input data provided to the AI ​​model is transformed into the desired output generated by the model. AI models are sometimes also referred to as "algorithms." A pre-trained AI is a model that has already been trained on a large dataset and is factory pre-configured, so that its fine-tuning is limited to specific tasks. This fine-tuning can include taking into account contextual information that familiarizes the pre-trained AI with an industrial device to be parameterized. Thus, the pre-trained AI becomes the trained AI, which is trained to determine parameter sets for configuring the industrial device.

[0013] A configuration file is used to configure an industrial device for a specific application. It is transmitted to the device via appropriate means and results in the corresponding parameter settings within the device. The configuration file has a predefined structure, in which different parameter types are defined in various sections.

[0014] Configuration files are transmitted to the industrial device before operation and are generally not changed during operation.

[0015] A training configuration file refers to the industrial device to be parameterized.

[0016] Accordingly, the training configuration file has a structure and necessary parameters appropriate for the industrial device, arranged according to that structure. The training configuration file differs from the standard configuration file in that it is designed for a different application of the industrial device or a different type of industrial device. An industrial device controls a system or process in conjunction with other industrial devices and components. It can be a multifunctional measuring device with numerous adjustable parameters. Other industrial devices can also serve for surge protection, grid connection, or control.

[0017] Processor and memory are electronic components and are generally known.

[0018] A receiving function is designed to receive signals. Reception can be wired or wireless. The receiving function relies on appropriate hardware, which may also include wire fastening elements such as those found in control cabinets. The receiving function detects and makes available the received signals, often in a modified physical form.

[0019] The processing function includes providing contextual information. By using this contextual information, the AI ​​becomes a trained AI. The contextual information is based on the training data.

[0020] The output function provides the context information for further processing, for example in the form of a file.

[0021] Advantageously, this allows the means (the context information) to be provided to generate a trained AI from a pre-trained AI, which is suitable for configuring the industrial device. In exemplary implementations, an instruction file can contain elements for defining a technical field, a task for the AI, and instructions for the AI ​​to create the configuration file.

[0022] Defining a technical field can encompass the technical area and a general task, as well as a human-readable text. This could be, for example, "Technical support for electronic components for measuring, switching, and monitoring." The technical field can also be referred to as a role.

[0023] The task of the class describes its function, for example in text form. This can include a format specification for the configuration file, which is, for example, designed as JSON (JavaScript Object Notation).

[0024] Instructions can specify minimum requirements. For example, it can be specified that a structure always includes a predefined field, such as the 'Info' field. The structure can also include direct instructions to the class, which can be in text format. For example, the instruction might be: 'Analyze the specified sources to understand the structure of a configuration text in JSON format'. Furthermore, the instruction can specify the number of groups in the configuration file to be created. For example, the instruction might be: 'At least one group must be present'. Additionally, the instruction can specify the format of the configuration file to be created, such as JSON (JavaScript Object Notation). Finally, the instruction can include specifications for a register in the configuration file to be created.

[0025] This allows the desired configuration to be specified in the configuration file.

[0026] In other embodiments, the instructions may include specifications for a structure. This structure defines the layout of the configuration file to be generated. Alternatively or additionally, the structure includes the format of the configuration file. Optionally, the configuration file format can be specified as JSON, as already mentioned. Furthermore, the instructions may include specifications for groups and, alternatively or additionally, specifications for registers within the configuration file.

[0027] The "Structure" element describes the structure of the configuration file to be created. Furthermore, the format of the configuration file can be specified, for example, the aforementioned JSON format.

[0028] The `Groups` element specifies the number of groups in the configuration file. This can be set to exact values ​​(`XGroups`), a minimum number (`>= XGroups`), or a maximum number (`< XGroups`). The format can be specified as JSON, as described above. This specification can be in text form, for example, "The configuration file must be created in JSON format."

[0029] The Register element defines the rules for the register in the configuration file. These can include the following instructions, which in turn can be specified in text form.

[0030] Register instructions govern the inclusion of new registers, the use of unused registers, the handling of groups of registers, the handling of key values, and additionally or alternatively, the handling of content values ​​of registers.

[0031] Examples of texts that may be included include: 'No new registers may be added' and additionally or alternatively 'Unused registers may be removed' and additionally or alternatively 'If all registers in a group are unused, the entire group may be removed' and additionally or alternatively 'The "key" values ​​of the registers may not be changed' and additionally or alternatively 'The "content" values ​​of the registers may be adjusted according to the change requests'.

[0032] Advantageously, a detailed configuration file can be specified to be created, which also enables the detection of existing training configuration files.

[0033] In other implementation examples, the training data can include another training configuration file.

[0034] The training configuration file and the secondary training configuration file can describe different configurations of the industrial device. The structure and parameters of the secondary training configuration file can correspond to those of the training configuration file. Optionally, second, third, and fourth secondary training configuration files can also be included in the training data.

[0035] This allows for the provision of a larger number of training data sets, enabling a more detailed description of the industrial device.

[0036] In some implementation examples, the training configuration file can be based on a first XML file. This first XML file can document the configuration of another industrial device and, alternatively or additionally, the industrial device in a different application. The XML file can contain a multitude of parameters in a predefined structure. Optionally, the second training configuration file can be based on a further XML file that documents the configuration of the other industrial device and / or the industrial device in a different application.

[0037] Other industrial equipment may include industrial equipment with comparable functionality, for example, from other manufacturers. It may also include older designs of the industrial equipment that are now to be replaced by newer designs.

[0038] Other applications may include the use of the industrial device in different circuits and, additionally or alternatively, under different conditions. These may include temperature, accuracy, resolution of measuring ranges, and other parameters of the circuit, environment, or parameterization of the industrial device.

[0039] This allows for the advantageous recording of a larger number of parameters to be specified for the configuration file to be created.

[0040] A second aspect of the invention relates to a computer-implemented method for executing a trained AI. This AI is suitable for generating a configuration file for configuring an industrial device. The method comprises the use of a processor and memory. It includes a receiver function for receiving user input, wherein the user input comprises at least two inputs. Optionally, the two inputs can comprise at least two input types from the input types text, speech, graphics, and additionally or alternatively, image. Furthermore, the method comprises a processing function for generating the configuration file based on the user input and context information generated by a pre-trained AI according to the first aspect and additionally or alternatively its associated embodiments. The method further comprises an output function that prints the configuration file.

[0041] A trained AI corresponds to the pre-trained AI according to the first aspect of the invention, supplemented at least by specific contextual information of the industrial device to be configured. This specific contextual information, also called relevant contextual information, encompasses the contextual information according to the first aspect of the invention. Furthermore, with regard to the industrial device to be configured, technical documentation, datasheets, principle diagrams, circuit diagrams, illustrations, photographs, and the like may be included in the contextual information.

[0042] The configuration file to be generated serves to configure a specific industrial device for a specific use case, as already described in the first aspect (pre-trained AI).

[0043] The receiving function is also based on the receiving function described in the first aspect, with the user input additionally including a combination of at least two input types: text, speech, graphics, and optionally or alternatively, images. Text, speech, graphics, and optionally or alternatively, images are used in accordance with common usage.

[0044] User input is enabled by a device that allows a user to provide input for the process. This often also includes user output to inform the user. Together, user input and user output are also referred to in technical terms as MMI (Man-Machine Interface). This is generally known.

[0045] The processing function generates the configuration file for the industrial device to be configured. To do this, it relies at least on user input and contextual information.

[0046] The output function makes the configuration file available for further processing, for example in the form of a file.

[0047] This can advantageously provide the basis for the automatic configuration of the industrial device.

[0048] In some implementation examples, the user input can include a second XML file. This second XML file can document the configuration of the industrial device being configured.

[0049] Additionally or alternatively, the processing function can include a verification function that ensures the configuration file conforms to the industrial device's properties. The second XML file describes a configuration of the industrial device. This can describe the desired target configuration of the industrial device. In particular, the second XML file can contain specific parameters to be considered in the target configuration, for example, in the form of a specific string. Such a parameter can also be included in the configuration file. For example, the parameter could be the web address of a server. This server could, for instance, set or regularly synchronize the industrial device's internal clock. The second XML file can also include configurations for other industrial devices.

[0050] A validation function can check the configuration file to be generated. This can include checks for completeness of the necessary parameters, dependencies between parameters, structural correctness, adherence to predefined value ranges, correct parameter names, and so on. The validation function can be called automatically when the configuration file is generated. It can also be configurable, allowing the scope and accuracy of the check to be set.

[0051] For example, if a web address is specified, a corresponding request can be made to the addressed server, which, as in the example above, queries and evaluates the time.

[0052] Optionally, discrepancies identified during the check can be displayed on a user interface. Based on the detected discrepancy, corresponding corrections can be suggested, which may include at least one of the following pieces of information: designation, limit values, characteristic curves, and / or recommendations for correcting the specified parameters of the industrial device.

[0053] This allows the parameters of the configuration file to be completed and verified. In further embodiments, the configuration file can include parameters that are marked as true or false. Furthermore, the configuration file can contain group information that includes the parameter definitions and, alternatively or additionally, register information that identifies a server. Alternatively or additionally, the configuration file can include information about the current measurement inputs of the industrial device.

[0054] A 'true' or 'false' parameter marking can be done verbally. The marking can be worded 'true' or 'false'. It can also be digitally encoded, for example, with '0' or '1'. The 'true' or 'false' parameter marking can be located in an information field of the configuration file. The parameter definitions can be introduced separately, for example, in a group field of the configuration file that can name the parameters. A registry entry can include, in addition to the server address, other server parameters, such as the server name, the character set used, the server's group membership, and so on.

[0055] Information on current measurement inputs can include value ranges, resolutions of the measurement ranges, alarm thresholds, and the like.

[0056] This allows the configuration file to take on a form that is readable and editable by both humans and machines.

[0057] In other embodiments, the image or graphic can be a circuit diagram of an electrical circuit using the industrial device. Alternatively or additionally, the text, graphic, and image can be presented as manuals, user instructions, measurement results, drawings, topologies, screenshots, and / or architectural representations related to the industrial device. The user instructions can also be in the form of an email, optionally as an image of an email or similar. The image can be a photograph. This advantageously allows for the use of a variety of input formats and media for the process.

[0058] In some embodiments, the industrial device can be configured as a measuring instrument. Optionally, the measuring instrument can be configured as a multifunctional energy meter.

[0059] Multifunctional energy meters can, for example, perform measurements of electrical parameters in low-voltage systems up to 690 V. This can include true RMS measurements, for example, for alternating current in sinusoidal form (50 Hz / 60 Hz). Such a device can also include an integrated Modbus (communication protocol of a client / server architecture) / TCP (Transmission Control Protocol) interface for measuring electrical parameters in low-voltage systems up to 690 V.

[0060] This can make it easier to use existing industrial products, for example for modernization, expansion or as a replacement for outdated control systems.

[0061] In further embodiments, the industrial device can be parameterized based on the output configuration file using a configuration tool. This tool can be a physical device. The parameterization then involves sending commands to the industrial device. Alternatively, the configuration tool can be a server on the internet, with a user interface for configuring the industrial device. Another alternative is that the configuration tool can be a mobile communication device or application software, also with a user interface for configuring the industrial device. Optionally, the application software can be a Windows program, a mobile application, an internet-based application, or a progressive internet-based application.

[0062] The configuration tool feeds the data from the configuration file into the industrial device. It can use a protocol supported by the industrial device, such as Modbus (a communication protocol in a client / server architecture) or TCP (Transmission Control Protocol). The configuration tool can take various forms and can run as a standalone device or as software on a server or mobile device. Optionally, the configuration tool can be designed to configure multiple industrial devices simultaneously based on the configuration specifications.

[0063] The user interface can include input and output for the user. It can point out inconsistencies and suggest corresponding solutions.

[0064] This allows the industrial device to be parameterized in a variety of advantageous ways.

[0065] In other implementations, the user interface of the server, mobile communication device, or application software may include the option to change the configuration. This may include, in particular, the ability to change individual configuration values. For example, a web address may be changed, such as the web address used to provide the time.

[0066] Individual values ​​can encompass all parameters of the configuration file. Following changes, the modified configuration file can be validated. For implausible values, parameters, and the like, suggested changes can be displayed by the settings tool.

[0067] This allows for advantageous post-processing of the configuration file in the event that, in exceptional cases, changes to the configuration file are required for the intended configuration of the industrial device.

[0068] In exemplary embodiments, the method for executing a trained AI can include a monitoring function of the industrial device. This monitoring function can indicate when parameters exceed predefined value ranges. Optionally, an alarm can be triggered when a parameter value range is exceeded. Alternatively or additionally, a recommendation for parameter correction can be provided.

[0069] The monitoring function can evaluate the configuration file, particularly on the web server, and provide feedback and instructions to the operator. For this purpose, the monitoring function can query and consider the identification data of the industrial device. The monitoring function can be parameterized with regard to the scope and level of detail of the monitoring. The monitoring function can make correction suggestions, optionally based on historical data from previous configuration files.

[0070] Inputs from the monitoring function can be received using the receive function to receive user input. Outputs from the monitoring function can be displayed using user output, which can, for example, be used in conjunction with the output function to display the configuration file. User input and user output can also be implemented as the user interface of the server, mobile communication device, or application software. This can be advantageous for identifying errors in the configuration file and supporting its improvement.

[0071] A third aspect of the invention relates to a device for generating a configuration file suitable for configuring an industrial device. The device comprises a processor and a memory. The processor and the memory are configured to implement a receive function, a processing function, and an output function according to the second aspect of the invention or one of its embodiments. It may also be suitable for implementing a method according to the first aspect of the invention or one of its embodiments. The device may include AI-compatible hardware.

[0072] Advantageously, such a device can be used both for training a pre-trained AI according to a method of the first aspect and for executing a trained AI according to a method of the second aspect.

[0073] A fourth aspect of the invention relates to a system for implementing a configuration file in an industrial device. The system comprises a device according to the third aspect of the invention and an adjustment means according to the second aspect of the invention, wherein the industrial device is configured for wireless communication. The adjustment means can be coupled to the industrial device. The coupling between the adjustment means and the industrial device is wired and, additionally or alternatively, wireless. Modbus (a communication protocol of a client / server architecture), optionally in conjunction with TCP (Transmission Control Protocol), can be used as the protocol. The adjustment means can also be coupled to multiple industrial devices. The coupling can be direct or via a network.

[0074] This allows the generation of the configuration file and the configuration of the industrial device to take place seamlessly in one system.

[0075] DETAILED DESCRIPTION

[0076] The invention is explained in more detail below with reference to the accompanying drawings and preferred embodiments, which can optionally be combined with one another. The drawings show...

[0077] Fig. 1 shows a schematic representation of a computer-implemented method for training a pre-trained AI according to the first aspect.

[0078] Fig. 2 shows a schematic representation of an instruction file according to the first aspect,

[0079] Fig. 3 shows a schematic representation of a first and second XML file for a different industrial device or for the industrial device in a different configuration according to the first aspect.

[0080] Fig. 4 shows a schematic representation of a computer-implemented method for executing a trained AI according to the second aspect.

[0081] Fig. 5a a first schematic representation of a configuration file according to the second aspect, Fig. 5b a second schematic representation of a configuration file according to the second aspect, Fig. 6 a schematic representation of the function of an industrial device, Fig. 7a a schematic representation of an industrial device set by a setting device according to the second aspect,

[0082] Fig. 7b shows a schematic representation of user input according to the second aspect,

[0083] Fig. 8 shows a schematic representation of a device for generating a configuration file according to the third aspect,

[0084] Fig. 9 shows a schematic representation of a system for implementing a configuration file according to the fourth aspect.

[0085] Fig. 10 shows a schematic representation of an alternative process of a trained AI according to the second aspect,

[0086] Fig. 11 shows a schematic representation of a first allocation of functional blocks to a hardware system according to the second aspect,

[0087] Fig. 12 shows a schematic representation of a second to fourth assignment of functional blocks to different hardware systems according to the second aspect.

[0088] Fig. 13 shows a schematic representation of a fifth assignment of functional blocks to different hardware systems,

[0089] Fig. 14 shows the components of the industrial device and subsequent components, and

[0090] Fig. 15 shows the sequence of process steps for generating output data of the setting device.

[0091] Fig. 1 shows a schematic representation of a computer-implemented method 10 for training a pre-trained AI according to the first aspect. The method 10 serves to generate a trained AI, which in turn is suitable for generating a configuration file 40 (not shown) for configuring an industrial device 50. The method 10 includes the use of a processor 60 and a memory 65 (not shown). The method 10 further includes a receive function for receiving training data 20, wherein the training data 70 comprise an instruction file 81 and a training configuration file 73. The instruction file 81 is referenced to the training configuration file 73. The method 10 further includes a processing function for generating context information 30 based on the training data 70 received via the receive function 20, wherein the context information 90 is configured for use by the trained AI.Procedure 10 further includes an output function for outputting the context information 98, based on the output of the processing function 30.

[0092] The training configuration file 73 can be structured as follows:

[0093] {

[0094] "info": {

[0095] "MIN_VERSION":"0.8.5",

[0096] "IS_CT":true,

[0097] "IS_RCP": false,

[0098] "IS_RS485": false,

[0099] "IS_RAIL":true,

[0100] "IS_PANEL": false,

[0101] "IS_DISP":false,

[0102] "IS_RT_ETH": false,

[0103] "IS_MID": false,

[0104] "FW_REV_STR": "1.5.3 ","IS_PC":false,

[0105] "hash": "bf24fe9a3b8cad609a3b31451079a09c"

[0106] },

[0107] "groups": {

[0108] "1": [

[0109] "IS_CT",

[0110] "IS_RCP",

[0111] "IS_RS485",

[0112] "IS_RAIL",

[0113] "IS-PANEL",

[0114] "IS_DISP",

[0115] "IS_RT_ETH",

[0116] "IS_MID"

[0117] ]

[0118] },

[0119] "registers": [

[0120] {

[0121] "key": "SNTP SERVER NAME",

[0122] "content": "europe.pool.ntp.org",

[0123] "t": "ASCH",

[0124] "c": 16,

[0125] "group": 1

[0126] }

[0127] ]

[0128] }

[0129] It is an example of the shortest possible structure with only one register. It has a structure with information (info), a version indicator, true or false values ​​for parameters of the industrial device 50, and a hash function for checking data integrity. Furthermore, the training configuration file 73 has group elements (groups) that define the parameters. Finally, the training configuration file 73 has register elements (registers) that contain server information and, for example, the server address as a text string "europe.pool.ntp.org". Fig. 2 shows a schematic representation of an instruction file 81 according to the first aspect. The instruction file 81 includes elements or fields for recording information to determine a technical field 76, a task of the computer 77, and instructions 78 to the computer for creating the configuration file 40.

[0130] Instructions 78 include specifications for a structure (not shown), where the structure defines the layout of the configuration file 40 to be generated. Alternatively or additionally, the structure includes the format of the configuration file 40, with instructions 78 providing further specifications for groups and, additionally or alternatively, specifications for registers of the configuration file 40. Optionally, the format of the configuration file 40 can be specified as JSON format.

[0131] Instruction file 81 can be structured as follows:

[0132] Technical field: Technical supporter for electronic components (switching, measuring and monitoring)

[0133] Task: Creating and customizing configuration files in JSON format. Instructions: 1. Understanding the structure: Analyze the given pages to understand the structure of a configuration file in JSON format. a. Required fields: The configuration file must always contain the "info" field. 2. Groups: At least one group ("groups") must be present.

[0134] 3. Format: The configuration file must be created in JSON format.

[0135] 4. Register:

[0136] a. No new registers may be added. b. Unused registers may be removed.

[0137] c. If all registers in a group are not in use, the entire group may be removed.

[0138] d. The "key" values ​​of the registers must not be changed. e. The "content" values ​​of the registers may be adjusted according to the desired changes.

[0139] Fig. 3 shows a schematic representation of a first XML file 79 for a different industrial device or for the industrial device 50 in a different configuration according to the first aspect. The first XML file 79 documents a configuration of a different industrial device and, alternatively or additionally, of the industrial device 50 in another application. The XML file 79 comprises an XML declaration 82, a title 84 of the XML file, and elements 86 that describe settings (parameters) of at least one industrial device. Optionally, a further training configuration file 75 can be based on a second XML file 80 with the same structure, which also documents a configuration of the other industrial device and / or the industrial device 50 in another application.

[0140] Fig. 4 shows a schematic representation of a computer-implemented method 100 for executing a trained AI according to the second aspect. The computer-implemented method for executing a trained AI 100 is suitable for generating a configuration file 40 for configuring an industrial device 50. The method 100 includes the use of a processor 60 and a memory 65 (not shown). The method 100 further includes a receive function for receiving user input 122, wherein the user input 122 comprises at least two inputs. Optionally, the two inputs comprise two input types from the input types text, speech, graphics, and image. The method 100 further includes a processing function for generating the configuration file 130 based on the user input 122 and on context information 90 generated by a pre-trained AI according to the first aspect or its embodiments.Procedure 100 further includes an output function for outputting the configuration file 140.

[0141] The image or graphic can be designed as a circuit diagram of an electrical circuit using the industrial device 50. Alternatively or additionally, the text, graphic, and image can be designed as manuals, user instructions, measurement results, drawings, topologies, and, additionally or alternatively, architectural representations related to the industrial device 50.

[0142] User input 122 can include an XML file for the trained KL 124. The XML file for the trained KL 124 documents a configuration of the industrial device 50. Additionally or alternatively, the processing function 130 includes a verification function 132 that ensures the conformity of the configuration file 40 with the properties of the industrial device 50 (not shown). The XML file for the trained KL 124 can be structured as follows:

[0143] 1<?xml version = "I.O" encoding="UTF-8" ?>

[0144] 2 <aml>

[0145] 3 clnstance Hierarchy Name="AutomationMLExample">

[0146] 4 clnternal Element Name="EEM-MB370-24DC" I D = " 1 127061 ">

[0147] 5 Attribute Name="Typ" Value=''EEM-MB370-24DC'7>

[0148] 6 <attribute Name="Fabrikat" Value="Phoenix Contact" / >

[0149] 7 Attribute Name = ' , Art-Nr" Value = "1 127061 " / >

[0150] 8 <attribute Name="Bauform" Value="Tragschienengerät ohne Display" / >

[0151] 9 <attribute name=" KommunikationsP rotoko 11 1 ” Value = " modbus tcp '7>

[0152] 10 <attribute name="Kommunikationsprotokoll2" value="REST'7>

[0153] 11 <attribute Name="Eingangsspannungsbereich" Value="35 V AC - 690 V AC (Phase / Phase) " / >

[0154] 12 < Att ri bute Name = "Input signal current l" Value = "l A (secondary) " / >

[0155] 13 < Att ri bute Name = "Input signal Current2" Value = "5 A (secondary) " / >

[0156] 14 < Att ri bute Name = "Input signal Current3' T Value = "1 A - 20000 A (primary)" / > 15 < A tt ri bute Name="Current measurement" Value = "via external converters" / >

[0157] 16 < Att ri bute Name=”Voltage Measurement" Value = "direct / via external converters" / > 17 < Att ri bute Name = "Supply Voltage Range" Value = "18 V DC - 30 V DC" / > 18 < Att ri bute Name = "Accuracy Class" Value = "Class 0? 5 S" / >

[0158] 19 < Att ri bute Name = " connection type" Value = " screw connection 7>

[0159] 20 < Att ri bute Name = "Ambient temperature (operation)" Value = "-25 °C - 70 °C" / > 21< / l nternalElement>

[0160] ​"" ""> < "" " "> < "" " "> < " " " "" " > "" " > < "" " > < " " " "> " " " > < " " " " > < " " " "> < " " " ">

[0161]

[0162] < " " ""> 35 Attribute Name="DHCP Server IP" Value = " 192. 168 . 214. I " / >

[0163] 36 < / l nternalElement>

[0164] 37 < / l nstanceHierarchy>

[0165] 38< / attribute> < / attribute> < / aml>

[0166] The XML file for the trained KL 124 includes, as an example, the documentation for two industrial devices. One is the EEM-MB370-24DC with ID 1127061, a measuring device from Phoenix Contact®. The other is the Beryl with ID GL-MT1300, a router from Gl.iNet®. The latter has a server IP address "192.168.214.1".

[0167] on, which is copied as an example into the configuration file 40 for the assignment of an external time transmitter for the industrial device 50.

[0168] Fig. 5a shows a first schematic representation of a configuration file 40 according to the second aspect.

[0169] The configuration file includes meta information 42 and content 44.

[0170] For example, meta-information can include 42 such as the article type, the minimum required firmware version of the energy meter, and a hash value for checking data integrity.

[0171] These are framework parameters that provide additional information for handling configuration file 40, such as during fault diagnosis or other types of evaluation. Subsequently, values ​​are included that are represented as pairs of keywords and

[0172] Value data can be implemented. Fig. 5b shows a second schematic representation of a configuration file according to the second aspect. The configuration file 40 can be structured according to the training configuration file 73. Accordingly, the configuration file 40 can also include parameters 45 that are marked as true or false. Additionally or alternatively, the configuration file 40 can include group information 46 that contains the parameter definitions. Furthermore, additionally or alternatively, the configuration file 40 can include register information 47 that identifies a server that serves, for example, as an external clock. Additionally or alternatively, the configuration file 40 can also include information on current measurement inputs of the industrial device 50 (not shown).

[0173] Configuration file 40 can be structured as follows:

[0174] {

[0175] "info": {

[0176] "MIN_VERSION":"0.8.5",

[0177] "IS_CT":true,

[0178] "IS_RCP": false,

[0179] "IS_RS485": false,

[0180] "IS_RAIL":true,

[0181] "IS_PANEL": false,

[0182] "IS_DISP":false,

[0183] "IS_RT_ETH": false,

[0184] "IS_MID": false,

[0185] "FW_REV_STR": "1.5.3 ",

[0186] "IS_PC":false,

[0187] "hash": "bf24fe9a3b8cad609a3b31451079a09c"

[0188] },

[0189] "groups": {

[0190] "1": [

[0191] "IS_CT",

[0192] "IS.RCP",

[0193] "IS_RS485",

[0194] "IS_RAIL",

[0195] "IS_PANEL",

[0196] "IS.DISP",

[0197] "IS_RT_ETH",

[0198] "IS_MID"

[0199] ]

[0200] },

[0201] "registers": [

[0202] {

[0203] "key": "SNTP SERVER NAME",

[0204] "content": "192.168.214.1",

[0205] "t": "ASCH",

[0206] "c": 16,

[0207] "group": 1

[0208] }

[0209] For example, the training configuration file 73 differs from the configuration file 40 by replacing the entry "europe.pool.ntp.org" with the entry "192.168.214.1", the actual server address in the Content parameter of the register.

[0210] Fig. 6 shows a schematic diagram of the function of an industrial device 50. The industrial device 50 is designed as a measuring instrument. Optionally, the measuring instrument is designed as a multifunctional energy meter. The energy meter comprises, by way of example, five input terminals 52, which are connected to five output terminals 53. The connections are in turn connected to voltage terminals 54 and, except for the ground wire, to current measuring devices 55.

[0211] Accordingly, various energy measurements can be carried out with the energy measuring device.

[0212] Fig. 7a shows a schematic diagram of an industrial device 50 configured by a setting device 150 according to the second aspect. Fig. 7a shows the parameterization of the industrial device 50 based on the output configuration file 40 with a setting device 150, wherein the setting device 150 is either designed as a device in which the parameterization transmits commands to the industrial device 50. Alternatively, the setting device 150 can be designed as a server on the internet, wherein the server has a user interface 152 for configuring the industrial device 50. Alternatively, the setting device 150 can be designed as a mobile communication device or as application software for the mobile communication device. The mobile communication device or the application software can have a user interface 152 for configuring the industrial device 50.Optionally, the application software can be designed as a Windows program, a mobile application, an internet-based application, or a progressive internet-based application.

[0213] Fig. 7b shows a schematic representation of user input according to the second aspect. A user interface 152 of the server or mobile communication device, or the application software, includes the ability to modify the parameters of the configuration file 40 (not shown), in particular the ability to modify individual values ​​of parameters in the configuration file 40. A mode 153 can be set in a field, for example, the SNTP server mode. The SNTP server field 154 can contain a text entry of the server address. The Check Connection field 154a can perform a check of the text entry. The Query Cycle field 155 can specify a query interval in minutes. The Time Zone field 156 is used to set the time zone. The Daylight Saving Time rule field 157 is used to set the daylight saving time rule.

[0214] The user interface 152 of the server, mobile communication device, or application software can include a monitoring function for the industrial device 50 (not shown). Alternatively or additionally, the inputs of the monitoring function can also be received using the receive function 120 to receive user input 122. Outputs of the monitoring function can be displayed using a user output 122, which, for example, can be used in conjunction with the output function 140 to output the configuration file (not shown). The monitoring function can indicate when parameters exceed predefined value ranges. Optionally, an alarm can be triggered when the parameter value ranges are exceeded. Alternatively or additionally, a recommendation for correcting the parameters can be provided.

[0215] Fig. 8 shows a schematic diagram of a device 200 for generating a configuration file according to the third aspect. The device for generating a configuration file 200 is suitable for configuring an industrial device 50 (not shown). It comprises a processor 60 and a memory 65. The processor 60 and the memory 65 are configured to perform a receive function 120, a processing function 130, and an output function 140 according to the second aspect of the invention or one of its embodiments (not shown). Fig. 9 shows a schematic diagram of a system for implementing a configuration file 300 according to the fourth aspect. The system for implementing a configuration file 300 in an industrial device 50 comprises a device 200 according to the third aspect and an adjustment means 150 according to the second aspect.The industrial device 50 can be configured for wireless communication and, alternatively or additionally, for wired communication. The configuration device 150 can be coupled to the industrial device 50. The configuration device 150 can also be coupled to multiple industrial devices 50 to implement the respective configuration file 300 (not shown). The coupling can be direct or via a network (not shown). Fig. 10 shows a schematic representation of an alternative process of a trained AI 105 according to the second aspect. A receive function of the trained AI 120 serves as input, which can also include multimedia prompts (inputs, documents, etc.). A selection device 134 manages the pool containing the context information 90. A processing function of the trained AI 130 evaluates the context information 90 and processes the multimedia prompts 120.Output 140 displays the result, usually in the form of a file, for example as a configuration file 40. User input 122 includes the evaluation and optionally the processing of the output. Furthermore, it can ask follow-up questions or further inquiries to the trained AI 105.

[0216] Fig. 11 shows a schematic representation of a first arrangement 410 of functional blocks for a first hardware system according to the second aspect. The functional blocks receive function 20 / 120, processing function 30 / 130 and output function 98 / 140 are jointly assigned to one system (dashed lines).

[0217] Fig. 12 shows a schematic representation of a second to fourth arrangement 420 / 430 / 440 of functional blocks for different hardware systems according to the second aspect. The functional blocks receive function 20 / 120, process function 30 / 130, and output function 98 / 140 are arranged in different, mutually communicating systems (dashed lines). In arrangement 420, the functional blocks are each arranged in separate systems (dashed lines). In arrangement 430, the process function 30 / 130 and the output function 98 / 140 are arranged together in one system, and the receive function 20 / 120 is in a separate system. In arrangement 440, the receive function 20 / 120 and the process function 30 / 130 are arranged together in one system, and the output function 98 / 140 is in a separate system.

[0218] Fig. 13 shows a schematic representation of a fifth arrangement 450 of functional blocks for different hardware systems. In the arrangement 450, the receive function 20 / 120 and the output function 98 / 150 are arranged together in one system, and the processing function 30 / 130 is in a separate system.

[0219] Figure 14 shows, in scenario 500, components of the industrial device 50 and subsequent components. The industrial device 50 comprises hardware 510 and firmware 520, which includes the configuration data and the user data. The firmware 520 includes configuration data 530, which makes the generated user data 540 usable for the consumer 560. The industrial device 50 is connected to the consumer 560 of the user data 540 via communication channels 550.

[0220] Fig. 15 shows the sequence of process steps for generating output data of the setting device 150 (not shown). This includes the sequence of receive function 120, processing function 130, output function 140, and setting function 151.

[0221] In other words, the invention can also be described as follows:

[0222] The invention consists of the specific integration of a large language model (ChatGPT, Copilot) into methods and devices for configuring an industrial device 50. This integration makes it possible to easily change the settings of the industrial device 50 and to adapt the industrial device 50 with minimal effort and without transmission errors. The industrial device 50 consists of electronics comprising hardware 510 and firmware components 520. In the firmware 520, setting data must be changed so that the electronics can output suitable user data. The user data consumer 560 can access the user data via communication channels 550, see also Fig. 14.

[0223] The procedure comprises the following steps, which are also outlined in Fig. 15:

[0224] The receive function 120 accepts input in the form of a prompt, which receives artifacts in the form of text, speech, or other formats. These artifacts contain information about the desired settings for the current measurement inputs of the industrial device 50. The processing function 130 identifies the settings. The large language model (LLM) – for example, ChatGPT or Copilot – analyzes the artifacts and identifies the specific settings for the current measurement inputs of the industrial device 50.

[0225] Output function 140 generates middleware-compatible information, the configuration file 40. When the LLM detects a configuration record for the current inputs, it automatically generates a configuration file 40 suitable for further processing by middleware. This configuration file 40 contains the relevant settings.

[0226] The setting device function 151 is implemented, for example, by an automated system. The generated configuration file 40 can be received by an automated system, which sends the commands contained therein to the industrial device 50. This enables the efficient configuration of the electronics.

[0227] Alternatively, the setting function 151 can be performed by a web server of the industrial device 50. The web server of the industrial device 50 can process the configuration file 40. This web server enables the configuration and monitoring of the industrial device 50 via a user interface 152 in the browser.

[0228] Alternatively, the setting tool function 151 can be performed using special setting software. This can be implemented as a Windows program, mobile app, web app, or progressive web app.

[0229] In general, configuring industrial equipment is a frequent process in daily customer interactions. Customers often have XML files from other programs containing the necessary settings information. This data is comparable to a database. However, customers cannot easily and efficiently transfer the configuration to the industrial equipment. Manually transferring the settings data from the XML files to the industrial equipment is time-consuming and prone to errors. Customers need an efficient solution to overcome this hurdle.

[0230] The presented method 10 / 100, which uses large language models such as ChatGPT and Copilot, offers a solution. By automatically identifying the settings for the current measurement inputs of the industrial device 50, customers can conveniently convert the relevant data into a configuration file 40 suitable for middleware.

[0231] When training the system for AI-supported configuration of an industrial device 50, in particular a measuring device, it must be taken into account that the system can be trained using various data materials, including manuals, user instructions, measurement results, circuit diagrams, drawings, topologies and, additionally or alternatively, architectural representations.

[0232] The training process 10 of the KL system, implemented as a language model, comprises several steps: 1. Data preparation: Collection and processing of the relevant data materials. This includes digitizing manuals, user instructions, and, additionally or alternatively, circuit diagrams, as well as formatting the data into a format understandable to the model.

[0233] 2. Data annotation: Tagging and categorizing the data to ensure that the model can correctly interpret the different types of information. This can include annotating text, drawings, and, additionally or alternatively, diagrams.

[0234] 3. Model training: Using machine learning algorithms to train the language model with the prepared and annotated data. Techniques such as supervised and unsupervised learning methods are used to optimize the model.

[0235] 4. Validation and Testing: Verification of model performance using validation and test datasets to ensure the model operates accurately and reliably. Adjustments and fine-tuning are made based on the test results.

[0236] 5. Deployment and Monitoring: Implementation of the trained model into the target system and continuous monitoring of its performance in practical use. Feedback loops and regular updates ensure that the model remains up-to-date and functions optimally.

[0237] Furthermore, in the training process 10, 'learning from a single instance' (technically one-shot learning) and 'learning with a small amount of training data' (technically few-shot learning) can be used in the field of machine learning, so that training can take place with very few examples.

[0238] The invention comprises training the language model and configuring the industrial device 50, for example, the measuring device of the EMpro product family from Phoenix Contact®, using the prepared language model in its trained state. For illustrative purposes, a parameter for the internal device clock, "SNTP server," is chosen as an example, which, at the customer's request, is intended to assume the value of the corresponding IP address. This is explained in detail above in connection with Figures 4 and 5.

[0239] The instructions generated during the training of the language model can be compiled into an instruction file and can form the basic framework for the model to recognize context. A small number of instructions is particularly useful when it is difficult or expensive to collect large amounts of training data. One could also use few-shot learning to train the generative AI with only a few examples of configuration components, enabling it to create new configurations based on customer requirements. An XML file can also be used as training data.

[0240] Based on the configuration file 40, communication with the industrial device 50 is possible during the processing process. For example, in a web-based management system (web server), the configuration file 40 created by the generative AI can be used to modify the configuration of the industrial device 50.

[0241] Another input source could be a screenshot of a customer email, which is unstructured and informally written. Nevertheless, attributes for a device configuration must be extracted. The prompt "Create a configuration file based on the screenshot" leads to a corresponding evaluation of the configuration file, indicating that the customer requests the SNTP server to be set to the address 192.168.214.230. Configuration file 40 is then modified by inserting the value 192.168.214.230, and the completed configuration file 40 can subsequently be downloaded. Importing it into the industrial device 50 produces the desired result. In summary, the presented method, device, and system enable seamless integration of the settings into the industrial device 50. Customers save time and minimize errors by creating and processing configuration data 40 more efficiently.REFERENCE SYMBOL LIST TERMS 90 Context information 10 Procedure pre-trained class 98 Output function training.

[0242] 20 Reception function Training class Contextual text

[0243] 30 Processing function training with Kl 100 procedures with trained Kl

[0244] 40 Configuration file 105 Alternative process of the KL

[0245] 42 Meta-information 120 Reception function trained computer 44 Content 122 User input / User output 45 Parameters 124 XML file for the trained computer 46 Group information 130 Processing function with trained computer

[0246] 47 Register information

[0247] 132 Verification function

[0248] 50 industrial equipment

[0249] 134 selection tools

[0250] 52 input ports

[0251] 140 Output function with trained Kl 53 output connections

[0252] 150 adjustment tools

[0253] 54 voltage connections

[0254] 151 Adjustment device function

[0255] 55 Current measuring instruments

[0256] 152 User interface settings 60 Processor

[0257] 153 Mode

[0258] 65 memory

[0259] 154 SNTP Server

[0260] 70 training data

[0261] 154a Check connection

[0262] 73 Training configuration file

[0263] 155 query cycle

[0264] 75 more training configuration files

[0265] 156 Time zone

[0266] 76 Technical Field

[0267] 157 Daylight Saving Time Rule

[0268] 77 Task of the class

[0269] 200 Device

[0270] 78 instructions

[0271] 300 system consisting of device and

[0272] 79 First XML file for another

[0273] Adjustment device

[0274] Industrial device / other configuration

[0275] 410 first arrangement

[0276] 80 Additional XML file for another

[0277] Industrial device / other configuration 420 second arrangement

[0278] 81 Instruction file 430 third arrangement

[0279] 82 XML declaration 440 fourth arrangement

[0280] 84 titles XML file 450 fifth arrangement

[0281] 86 elements XML file (measuring device, 470 system part 2)

[0282] Router) Scenario Industrial device 550 Communication path Hardware 560 Consumer / User Firmware 600 Data acquisition Industrial device setting data

[0283] Usage data

Claims

REQUIREMENTS 1. Computer-implemented method for training a pre-trained AI (10) to produce a trained AI suitable for generating a configuration file (40) for configuring an industrial device (50), wherein the method (10) comprises the use of a processor (60) and a memory (65), comprising: a receive function for receiving training data (20), wherein the training data (70) comprise an instruction file (81) and a training configuration file (73), wherein the instruction file (81) is related to the training configuration file (73), and a processing function for generating context information (30) based on the training data (70), wherein the context information (90) is trained for use by the trained AI, and an output function for outputting the context information (98).

2. Method (10) according to claim 1, wherein the instruction file (81) contains elements for determining a technical field (76), a task of the Cl (77) and instructions (78) to the Cl for creating the configuration file (40).

3. Method (10) according to claim 1 or 2, wherein the instructions (78) include specifications for a structure, wherein the structure specifies the configuration of the configuration file (40) to be generated and / or wherein the structure includes the format of the configuration file (40), wherein the instructions (78) further include specifications for groups and / or for registers for the configuration file (40), optionally wherein the format of the configuration file (40) is specified as JSON format.

4. Method (10) according to one of claims 1 or 3, wherein the training data (70) comprises a further training configuration file (80).

5. Method (10) according to any one of claims 1 to 4, wherein the training configuration file (73) is based on a first XML file (79) that documents a configuration of another industrial device and / or the industrial device (50) in another application, optionally wherein the further training configuration file (75) is based on a further XML file (80) which documents a configuration of the other industrial device and / or the industrial device (50) in another application.

6. Computer-implemented method for executing a trained AI (100) suitable for generating a configuration file (40) for configuring an industrial device (50), wherein the method (100) comprises the use of a processor (60) and a memory (65), comprising: a receive function for receiving a user input (120), wherein a user input (122) comprises at least two inputs, optionally wherein the two inputs comprise two input types from the input types text, speech, graphics and image, a processing function for generating the configuration file (130) based on the user input (122) and context information (90) generated by a pre-trained AI according to one of claims 1-5, and an output function for outputting the configuration file (140).

7. Method (100) according to claim 6, wherein the user input (122) comprises an XML file for the trained AI (124), wherein the XML file for the trained AI (124) documents a configuration of the industrial device (50) and / or wherein the processing function (130) includes a verification function (132) which ensures the conformity of the configuration file (40) with the properties of the industrial device (50).

8. Method (100) according to one of claims 6 or 7, wherein the configuration file (40) includes parameters (45) that are marked as true or false and / or group information (46) that includes the parameter definitions and / or register information (47) that identifies a server and / or wherein the configuration file (40) includes information on current measurement inputs of the industrial device (50).

9. Method (100) according to any one of claims 6 to 8, wherein the image or graphic is designed as a circuit diagram of an electrical circuit using the industrial device (50) and / or wherein the text, graphic and image are designed as manuals, user instructions, measurement results, drawings, topologies and / or architectural representations relating to the industrial device (50).

10. Method (100) according to any one of claims 6 to 9, wherein the industrial device (50) is designed as a measuring device, optionally the measuring device is designed as a multifunctional energy measuring device.

11. Method (100) according to any one of claims 6 to 10, wherein The industrial device (50) is parameterized based on the output configuration file (40) using a setting device (150), wherein the setting device (150) is designed as an apparatus, wherein the parameterization includes commands to the industrial device (50) or is configured as a server on the Internet, wherein the server has a user interface (152) for configuring the industrial device (50) or is designed as a mobile communication device or as application software and has a user interface (152) for configuring the industrial device (50), optionally wherein the application software is designed as a Windows program, as a mobile application, as an Internet-based application or as a progressive Internet-based application.

12. Method (100) according to claim 11, wherein the user interface (152) of the server or mobile communication device or application software includes a possibility to change the configuration, in particular a possibility to change individual values ​​of the configuration.

13. Method (100) according to one of claims 6 - 12, further comprehensive monitoring function of the industrial device (50), where the monitoring function indicates when parameters exceed predefined value ranges, optionally where an alarm is triggered when the parameter's value range is exceeded and / or a recommendation for correcting the parameters is given.

14. Device for generating a configuration file (200) suitable for configuring an industrial device (50), comprising a processor (60) and a memory (65), the processor (60) and the memory (65) are set up to form a receiving function (120), a processing function (130) and an output function (140) according to one of claims 6 to 10.

15. System for implementing a configuration file (300) in an industrial device (50), comprising: a device (200) according to claim 14 and an adjustment means (150) according to one of claims 11 or 12, wherein the industrial device (50) is designed for wireless or wired communication, wherein the adjusting device (150) can be coupled to the industrial device (50) or a plurality of industrial devices (50), optionally the coupling takes place via a network.

Images