Device for controlling and / or monitoring technical equipment

Abstract

Device and method for controlling and / or monitoring technical equipment for metal production and / or metal processing, with an auxiliary program, which has at least one interface for connecting with an application program, wherein the auxiliary program obtains a query for information about the equipment from at least one querying application program via the interface, wherein the auxiliary program has access to a data model, wherein the data model gives a hint as to which information at least one other application program can provide, wherein the auxiliary program determines, depending on the query and depending on the data model, which application program can provide the queried information, wherein the auxiliary program transmits the query to at least one determined application program which can provide the queried information, wherein the auxiliary program receives a response from the determined application program, and wherein the auxiliary program outputs the received response to the querying application program.

Description

TECHNICAL FIELD

[0001] The invention relates to a device for controlling and / or monitoring a technical plant for carrying out a metal production and / or a metal processing and to a method for controlling and / or monitoring a technical plant for carrying out a metal production and / or a metal processing. BACKGROUND

[0002] A device and a method for controlling and / or monitoring a technical plant for carrying out a metal production and / or a metal processing are known from EP 3 293 594 A1.

[0003] US 20170346768 describes a conversational interface system that enables remote access to information about manufacturing processes by exchanging messages in simple language. A cloud-based conversational interface service connects with an instant messaging application and receives a plaintext query from the interface of the instant messaging application that requests information about one or more industrial systems. The cloud-based system synchronizes the query with one or more agent devices on site of the conversational interface located in one or more plants. The on-site agent devices translate the query and apply the translated query to local sources of manufacturing operation data. The on-site agent devices then generate a response message. The cloud-based system forwards the response message to the sender of the query through the instant messaging interface.

[0004] US 20180231954 A1 describes a method for receiving a text message from a client device. The text message contains a query for information of an industrial process control and automation system. The method further comprises evaluating the text message in order to identify the requested information. The method furthermore comprises transmitting one or more queries for the requested information and obtaining the requested information. Further, the method comprises generating a natural language response containing the requested information and transmitting the natural language response for delivery to the client device. SUMMARY

[0005] It is an object of the present invention to provide an improved device and an improved method for controlling and / or monitoring a technical plant for carrying out a metal production and / or a metal processing.

[0006] The object of the present invention is solved by the independent patent claim.

[0007] Advantageous embodiments of the device and the method are specified in the dependent claims.

[0008] A device for controlling and / or monitoring a technical installation for producing and / or processing metal is proposed. The device has a computing system with at least one interface for connecting to at least one other computing system of the installation and / or to at least one sensor device of the installation and / or to at least one human-machine interface and / or to at least one application program. The computing system is configured to obtain an inquiry for information about the installation from at least one inquiring other computing system and / or from an inquiring sensor device of the installation and / or from an inquiring human-machine interface and / or from an inquiring application program via the interface. The computing system has access to a data model via the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program, which data model gives a hint as to which information the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program have access to. The computing system is configured to determine the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program that can provide the searched information from the data model.

[0009] Furthermore, the computing system is configured to obtain the inquired information from at least one determined other computing system and / or from at least one determined sensor device and / or from at least one determined human-machine interface via the interface. Furthermore, the computing system is configured to output the obtained information to the inquiring other computing system and / or to the inquiring sensor device of the installation and / or to the inquiring human-machine interface and / or to the inquiring application program via the interface. The obtained information can be used, inter alia, by one of the computing systems in order to intervene in the control of the functions of the installation and to change the control. For example, a computing system that executes the automation of the installation can use the obtained information to change the control of the installation.

[0010] In particular, because it is first determined how the inquiry is optimally executed, i.e. by which source or by which sources the searched information can be provided, in particular with a greater probability, the information searched with the inquiry can be found more quickly, more simply and, in particular, more appropriately by means of the data model used. Only after the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program that can provide the searched information has been determined, the inquiry is executed by the determined other computing system and / or the determined sensor device and / or the determined human-machine interface and / or the determined other application program.

[0011] In this way a modular device is provided which enables information and / or data and / or knowledge of other computing systems of the device and / or sensor devices and / or human-machine interfaces and / or other applications of the device to be determined in a simple manner. For this purpose, it is not necessary for the information and / or knowledge to be centrally gathered in the computing system. By connecting the computing system with other computing systems and / or sensor devices and / or human-machine interfaces of the device, it is possible to acquire information and / or knowledge about the entire device and to pass it on to the human-machine interface and / or other computing systems and / or sensor devices and / or applications and / or for further processing.

[0012] In this way, the device can be flexibly supplemented with other computing systems, sensor devices and / or human-machine interfaces and / or applications, or other computing systems, sensor devices and / or human-machine interfaces and / or applications of the device can be replaced, without the corresponding data or information having to be deleted or changed in the central computing system. Since other computing systems and / or sensor devices and / or human-machine interfaces and / or other applications can be accessed to determine data, data updates in the central computing system can not be necessary. Furthermore, it is possible in this way to release or not release information of other computing systems, sensor devices and / or human-machine interfaces and / or applications for access depending on predefined boundary conditions. Furthermore, it is possible with the proposed modular system to keep the information, data and knowledge of the individual other computing systems, sensor devices and / or human-machine interfaces and / or applications separate and to pass on only specific information, data, control data and / or technical knowledge of the individual other computing systems, sensor devices and / or human-machine interfaces and / or applications to the computing system. Mixing of information, data, control data and / or technical knowledge of other computing systems, sensor devices and / or human-machine interfaces and / or applications is thus avoided. Furthermore, this modular approach enables a better overview and separation of the functions of the other computing systems, sensor devices and / or human-machine interfaces. This is particularly advantageous when searching for malfunctions of the device, since the exchange of data between the other computing systems, sensor devices and / or human-machine interfaces and / or applications makes it easy to find malfunctions of the technical device.

[0013] The information can represent, for example, a status of at least one sub-device of the device or a status of the device. Furthermore, the information can represent a control value of the device, which is used by the computing system or another computing system to manipulate the device. In an embodiment, the information can be based on camera data, which optically acquires a region of the device. Thus, the information can represent at least one image of a portion of the device or a video sequence showing a portion of the device. The information based on the camera data, in particular the image, enables, for example, an optical evaluation of the status of the device, wherein artificial intelligence can be used in evaluating the camera data. Furthermore, the image or video sequence, for example, represents information that is quickly and simply understandable for an operator, which is well suited for evaluating the status of the device or for evaluating the control variable of the device. The result of the analysis can also be used by the computing system or another computing system to change the control data for the device. Thus, the operability and control of the technical device is simplified both for the operator and for the computing system.

[0014] In an embodiment, the computing system is configured to output the acquired information in the form of at least one image or video sequence, in particular to a human-machine interface. The human-machine interface can have, for example, at least one screen or a plurality of screens. Furthermore, depending on a setting by the operator, the desired information can be represented in the form of an image on one screen or on a plurality of screens. The selection of the information output via the human-machine interface can be specified by the operator or automatically by the computing system depending on the operating parameters or status of the device and / or depending on the control data of the device. An optimized selection of the information output or represented via the human-machine interface is thereby achieved. Thus, important information for a respective status or a respective control data can be represented or output with few screens, in particular with only one screen. A screen can thereby be saved and / or an improved representation or output of information can be achieved. Via the human-machine interface, not only images can be represented, but also data, status, control data, etc.

[0015] In one embodiment, the computing system is configured to determine, from the query, information searched with the query. The computing system is furthermore configured to determine which other computing system and / or which sensor device and / or which human-machine interface and / or which other application program can provide the searched information. Here, the computing system can determine, for example, a probability with which the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program can provide the searched information. Generally, the computing system forwards the query to the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program with the highest probability. Thereby, the effort for a successful search is reduced. For example, the computing system can be configured to determine, from a data model, a probability with which the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program can determine the information searched with the query. To this end, the computing system can use, for example, a trained neural network.

[0016] In another embodiment, the computing system is configured to determine, from a query existing in textual form via an NLU procedure, information searched with the query.

[0017] In one embodiment, the data model has prescribed examples, in particular example sentences, for queries for the other computing system and / or the sensor device and / or the human-machine interface and / or the other application program with which information can be successfully searched in the other computing system and / or in the sensor device and / or in the human-machine interface and / or in the other application program. Instead of example sentences, the examples can also only comprise individual words and / or graphical and / or acoustic signals and / or speech signals and / or images. Depending on the selected embodiment, the query can be configured as a word or as text in the form of multiple words, in particular in the form of a sentence, and / or in the form of a graph and / or in the form of an acoustic signal and / or in the form of a speech signal and / or in the form of an image.

[0018] By comparing the prescribed examples, in particular example sentences, with the query, in particular taking into account the information searched with the query, the computing system determines which other computing system and / or which sensor device and / or which human-machine interface and / or which other application program can provide the information searched with the query. Furthermore, the computing system can determine which other computing system and / or which sensor device and / or which human-machine interface can provide the searched information with which probability.

[0019] In one embodiment, the computing system receives further inquiries for further information about the device from the inquired other computing systems and / or sensor devices and / or human-machine interfaces and / or other applications in response to the transmitted inquiry of the information. The computing system determines from the further inquiries which other computing system and / or which sensor device and / or which human-machine interface and / or which other application is capable of responding to the further inquiries. The computing system is configured to retrieve the inquired further information from at least one determined other computing system and / or from at least one determined sensor device and / or from at least one determined human-machine interface and / or at least one other application via the interface. The computing system outputs the retrieved further information to the inquiring other computing system and / or to the inquiring sensor device and / or to the inquiring human-machine interface and / or to the inquiring application via the interface. Thus, the information can be determined and transmitted to the computing system, for example, by a plurality of other computing systems and / or sensor devices and / or human-machine interfaces and / or other applications together.

[0020] In another embodiment, the computing system is configured to retrieve the specified information from at least one other computing system and / or sensor device and / or human-machine interface and / or other application via the interface depending on at least one predefined operating state of the device and / or on a predefined control value for the device. The computing system can output the independently retrieved information to the other computing system and / or to the sensor device and / or to the human-machine interface and / or to the application via the interface.

[0021] The desired information, data and / or images can be determined, for example, automatically depending on the specified condition, such as a predefined disturbance of the device, and output to the other computing system and / or to the sensor device and / or to the human-machine interface and / or to the application. An adaptation of the determined and provided information is thus achieved automatically for the specified condition. By automatically adapting the information, an improved control and / or monitoring of the device is achieved by means of the other computing system and / or by means of the operating personnel via the human-machine interface and / or by means of the control device and / or by means of the application.

[0022] Depending on the selected embodiment, it can be specified, for example, by the operating personnel, in which operating states and / or in the case of which predefined control values which specified information is to be retrieved from at least one other computing system and / or at least one sensor device and / or human-machine interface and / or application via the first interface and transmitted to which other computing system and / or to which sensor device and / or to which human-machine interface and / or to which application. The information retrieval and the information flow can thus be configured, for example, by the operating personnel. To this end, the human-machine interface has corresponding input means. In this way, the monitoring and control of the device and in particular the human-machine interface can be adapted individually.

[0023] In another embodiment, the computing system is configured to be able to convert an inquiry in the form of an acoustic inquiry, in particular a voice inquiry, into text and to further process the text as an inquiry. By converting the acoustic inquiry into text, which can represent program instructions, the information desired in the acoustic inquiry can be searched for and retrieved via an interface. In this way, simplified control and monitoring of the device can be achieved. Furthermore, the computing system is configured to output the information retrieved via the first interface to other computing systems and / or sensor devices and / or application programs via the first interface and / or to a human-machine interface via the second interface. For example, the human-machine interface has a microphone for voice input. Furthermore, the computing system can have a data memory in which stored acoustic voice commands are assigned to prescribed program instructions. In this way, the conversion of acoustic inquiries into program instructions of the computing system is simplified. Here, the acoustic voice signal, i.e. the language, can first be converted into text and then the text can be converted into program instructions for a program, in particular a program of the computing system or of another computing system. The program can be configured, for example, as an application program or as an auxiliary program.

[0024] In another embodiment, the computing system is configured to at least at prescribed points in time, in particular continuously, to acquire and store the operating state of the device and / or the production quality of the device and / or the maintenance state of the device. In another embodiment, the computing system is configured to further process or compare with one another the operating state of the device and / or the production quality of the device and / or the maintenance state of the device acquired at different times. Furthermore, depending on the embodiment chosen, the acquired operating state of the device and / or the acquired production quality of the device and / or the acquired maintenance state of the device can be output to other computing systems and / or sensor devices and / or application programs via the first interface and / or to a human-machine interface via the second interface. As a result, important parameters of the device can be automatically notified to other computing systems and / or sensor devices and / or a human-machine interface and thus to an operator.

[0025] In another embodiment, the computing system is configured to analyze the acquired operating states and / or production quality and / or maintenance states. For example, the operating states of the device acquired at different times can be compared with one another. For example, the product quality of the device acquired at different times can be compared with one another. For example, the maintenance states of the device acquired at different times can be compared with one another. Here, for example, a database, stored evaluation methods, artificial intelligence, etc. can be used. By means of the analysis, additional information about the device and / or action recommendations can be determined and output to other computing systems and / or sensor devices and / or human-machine interfaces and / or applications. Furthermore, the analyzed data can also be stored in the data memory of the computing system. In this way, later analysis of the functioning of the device, in particular error searches, can be improved.

[0026] In another embodiment, other computing systems for sub-devices of the device can be connected to the computing system via an interface, wherein the other computing systems perform a device control and / or a state monitoring and / or a maintenance system for at least one sub-device of the device and / or have a spare parts catalog for at least one sub-device of the device, wherein each sub-device has a data memory with documentation about the sub-device. The computing system is configured to access the documentation of the sub-devices of the device via the interface, search for information in the documentation and output to the other computing systems and / or sensor devices and / or human-machine interfaces and / or applications via the interface. In this way, the computing system can search for information in the documentation of the sub-devices and hand over to the other computing systems and / or sensor devices and / or applications via a first interface and / or to the human-machine interfaces via a second interface. In this way, it is not necessary to collect information about the sub-devices centrally. When required, individual information of the sub-devices can be accessed, in particular the documentation of the sub-devices connected to the device.

[0027] Furthermore, the computing system can be configured to determine at least one information about the device by means of a maintenance / operation program, output instructions via the interface, document the performed steps and / or intervene in the control of the device. Here, the maintenance / operation program can draw on stored knowledge and / or learned knowledge. In particular, the maintenance / operation program can use artificial intelligence and, in particular, take into account knowledge learned by means of artificial intelligence when documenting or controlling the device. For example, the maintenance / operation program can perform evaluations with data analysis methods, such as deep learning, machine learning, support vector machines, etc.

[0028] In another embodiment, the computing system is configured to communicate, that is to say exchange data and information, with at least one other computing system and / or sensor device and / or with a human-machine interface by means of standardized messages and / or standardized communication protocols via the interface.

[0029] In one embodiment, the data model has a first data model set for a plurality of other computing systems and / or sensor devices and / or human-machine interfaces and / or other applications. The computing unit is configured to determine, in a first step, at least one other computing system and / or at least one sensor device and / or at least one human-machine interface and / or at least one application that can provide the information searched with the query, according to the query and by means of the first data model.

[0030] The data model has a second data model for the respective other computing systems and / or for the respective sensor devices and / or for the respective human-machine interfaces and / or for the respective other applications. A second data model can be set for each other computing system and / or for each sensor device and / or for each human-machine interface and / or for each other application.

[0031] The second data model has for the other computing systems or sensor devices or human-machine interfaces or other applications a prescribed example, in particular an example sentence, with which the information can be successfully searched in the other computing systems or in the sensor devices or in the human-machine interfaces by means of the functionality of the application. Instead of an example sentence, the example of the second data model can also only comprise individual words / or graphics and / or sound signals and / or speech signals and / or images.

[0032] The computing unit determines, according to the second data model of the other computing systems and / or sensor devices and / or human-machine interfaces and / or applications determined in the first step, which application of the other computing systems and / or sensor devices and / or human-machine interfaces determined in the first step and / or which functionality of the determined other applications can be used to determine the searched information.

[0033] After the application or the functionality of the application has been determined with which the searched information can be determined, the query is handed over to the determined functionality or to the determined application. Subsequently, the computing system receives the information from the determined functionality or from the determined application.

[0034] In another embodiment, the data model is configured by means of artificial intelligence, in particular as a trained neural network. The artificial intelligence is configured to determine the other computing systems and / or sensor devices and / or human-machine interfaces and / or applications that can provide the information searched with the query. The neural network can have been trained by means of a supervised learning method with pre-given examples in order to determine, according to the query, the other computing systems and / or sensor devices and / or human-machine interfaces and / or applications that can provide the information searched with the query. The examples can be configured similarly or identically to the query.

[0035] In another embodiment, the first data model is configured by means of artificial intelligence, in particular as a trained neural network, wherein the artificial intelligence is configured for determining other computing systems and / or sensor devices and / or human-machine interfaces and / or applications which are able to provide the searched information. The neural network can have been trained by means of a supervised learning method using examples. The examples can be configured analogously or identically to the query.

[0036] In another embodiment, the second data model is configured by means of artificial intelligence, in particular as a trained neural network, wherein the artificial intelligence is configured for determining applications and / or functions of applications of other computing systems and / or sensor devices and / or human-machine interfaces which are able to provide the searched information. The neural network can have been trained by means of a supervised learning method using examples. The examples can be configured analogously or identically to the query.

[0037] In an embodiment, the computing system has access to a data store or has a data store. In the data store, information is stored as a data model about which other computing systems and / or which sensor devices and / or which human-machine interfaces are connected via the interface and which information the other computing systems and / or the sensor devices and / or the human-machine interfaces and / or other applications can have or can access. In this way, in the case of a query, the computing system can search for the information desired in the query at at least one other computing system and / or sensor device and / or human-machine interface and / or application in a targeted manner. The computing system and / or the sensor device and / or the human-machine interface and / or the other application at which the search is carried out can change, in particular during the runtime of the program carrying out the search. In this way, the search for the information desired in the query is simplified and can be carried out more quickly.

[0038] Depending on the selected embodiment, the functional manner of the computing system and / or of the other computing systems can be realized in the form of electronic circuits and / or in the form of computing programs.

[0039] In an embodiment, the computing system has an auxiliary program. Furthermore, the at least one other computing system and / or the sensor device and / or the human-machine interface has at least two applications. The auxiliary program is configured for receiving a query for information via the interface. Furthermore, the auxiliary program is configured for determining, from the information searched by the query, an application which is able to provide the searched information. Furthermore, the auxiliary program is configured for forwarding the query to the application which is able to provide the information. Furthermore, the auxiliary program is configured for receiving the searched information from the application. The auxiliary program can output the information received from the application to the other computing system making the query and / or to the sensor device making the query and / or to the human-machine interface making the query via the interface.

[0040] The application program is configured to search for the desired information according to the transferred query. If the application program finds the desired information, the found information is returned to the assistant program. The assistant program is configured to output the information obtained from the application program via the first and / or second interface. The obtained information is output to the other computing system, the human-machine interface and / or the sensor device, depending on whether the other computing system, the sensor device and / or the human-machine interface has transferred the query to the assistant program. The assistant program can be configured to implement free-field search, voice input and / or machine-to-machine interface. Furthermore, the assistant program can be programmed based on a network.

[0041] In one embodiment, two application programs are configured to search for different information and / or information with different representations and / or information of different sub-devices of a device and / or information of different databases of a device. For example, by means of the application programs, specific operating data, specific control data, specific quality data, specific error reports, data of different device parts of a device and data of different databases of a device can be searched. Furthermore, the information can differ in the type of data, such as data values, images, analyzed data, learned data, aggregated data, etc. Furthermore, the information can be present in the form of documents, drawings, images, abbreviations, etc.

[0042] In another embodiment, the computing system, in particular by means of the assistant program, is configured to output images, documents, information and / or web addresses of data via the first and / or second interface. Thus, for example, information in the form of images, in the form of documents or in the form of web addresses of information can be output via the first and / or second interface.

[0043] A computer-implemented method for controlling and / or monitoring a technical device for producing and / or processing metal is provided, which uses an assistant program having at least one interface for connecting with application programs. The assistant program obtains a query for information about the device from at least one querying application program via the interface. The assistant program has access to a data model, wherein the data model gives a hint as to which information at least one other application program can provide. The assistant program determines, according to the query and according to the data model, which application program can provide the queried information. The assistant program then transfers the query to at least one determined application program that can provide the information. The assistant program receives a response from the determined application program and outputs the received response to the querying application program. The acquired information can be used, in particular by the assistant program and / or one of the application programs, in order to intervene in the control of the functioning of the device and to change the control. For example, the acquired information can be used by a device automation to change the control of the device.

[0044] In an embodiment, the data model has a first data model set for a plurality of other applications. Furthermore, the data model has a second data model for at least one other application, in particular for each of the other applications. The assistant first determines one or more other applications that can provide the desired information from the query and by means of the first data model. Subsequently, the assistant determines the functions of the determined other applications that can provide the searched information from at least one second data model of the at least one other application determined in the first step. Subsequently, the assistant hands over the query to the determined functions of the determined other applications. Then, the assistant receives the information searched with the query from the determined functions as a response.

[0045] In an embodiment, the data model is constructed by means of artificial intelligence. The artificial intelligence can be constructed as a trained neural network. The artificial intelligence is constructed for determining one or more other applications that can provide the searched information and / or functions of the other applications. The neural network can have been trained by means of a supervised learning method with examples. The examples can be constructed similarly or identically to the query.

[0046] In an embodiment, the first and / or second data model is constructed by means of artificial intelligence. The artificial intelligence of the first and / or second data model can be constructed as a trained neural network. The artificial intelligence is constructed for determining one or more other applications that can provide the searched information and / or functions of the other applications. The neural network can have been trained by means of a supervised learning method with examples. The examples can be constructed similarly or identically to the query.

[0047] In an embodiment, the assistant determines the information searched with the query from the query. Then, the assistant determines by means of the searched information and the data model which application can provide the searched information. Here, the assistant can in particular check which application can provide the searched information with which probability. Usually, the assistant hands over the query to the application with the highest probability. Thereby, the effort for a successful search is reduced.

[0048] In an embodiment, the assistant program determines the information searched by the query from the query existing in text by means of a recognition program, in particular an NLU program. The recognition program can be constructed by means of artificial intelligence, in particular as a trained neural network. The assistant program can access a data model for the application programs. The data model contains a hint which information the application programs can access with which the assistant program can establish a communication with the application programs. The assistant program determines in particular by comparing the query, in particular the searched information, with examples of the data model which of the application programs can provide the searched information. For this purpose, a trained neural network can also be used as a data model which determines from the query which application program is best suited to execute the query. Here, the assistant program can determine for example a probability with which the application program can execute the query and can determine for example the expected information.

[0049] In an embodiment, the data model has for at least a portion of the application programs respectively a prescribed example for searching information with the application program. The prescribed example can be constructed for example in the form of example words or at least one example sentence, in particular in the form of a plurality of example sentences. The prescribed example represents a search query with which information can be searched by means of the application program. For example, the NLU program can check by comparing the prescribed example sentence with the searched information which application program is able to provide the searched information. Here, the NLU program can also determine for each application program a probability with which the application program can execute the query and can provide for example the searched information.

[0050] In an embodiment, the assistant program receives from the queried application program in response to the query for information a further query for further information about the device. The assistant program determines from the further query which application program is able to respond to the further query. Subsequently, the assistant program acquires the further information from at least one determined application program, wherein the assistant program outputs the acquired further information to the queried application program. Thus, there is the possibility that the application program queried for information can itself query for further information. Thus, the information can be determined for example together by a plurality of application programs and transmitted to the assistant program. Thus, the query can be responded to by the application program even if the application program is only able to respond to a portion of the query and acquires further information from a further responding program for responding to the query.

[0051] In an embodiment, the assistant program acquires the prescribed information from at least one application program depending on at least one predefined operating state of the device and / or depending on at least one predefined control value for the device. The assistant program subsequently outputs the independently acquired information to a further application program or stores the independently acquired information in a data store.

[0052] In one embodiment, the information is based on the operating state of the device or on a control variable of the device. For example, the searched information can be based on camera data or video data and can in particular represent an image or a video sequence.

[0053] In one embodiment, the assistant program forwards the acquired information to the inquiring application program. Furthermore, instead of the information itself, the assistant program can forward a web address or a memory address to the inquiring application program. In this case, the application program can acquire, provide, output and / or store the searched information itself by accessing the web address or the memory address.

[0054] In one embodiment, the assistant program converts an inquiry in the form of an acoustic inquiry, in particular as a voice inquiry, into a program inquiry and searches for the information expected in the acoustic inquiry by means of the application program, wherein the assistant program outputs the found information to the application program.

[0055] In one embodiment, the assistant program acquires the operating state of the device and / or the production quality of the device and / or the maintenance state of the device at least at specified points in time, in particular continuously, and stores these in a data memory and / or forwards these to the application program.

[0056] In one embodiment, the assistant program analyzes the acquired operating state and / or production quality and / or maintenance state by means of technical knowledge and determines additional information and / or action recommendations and / or control values for the device and stores these and / or outputs these to the application program, in particular for monitoring and / or controlling the device.

[0057] In one embodiment, a plurality of application programs of sub-devices of the device are connected to the assistant program. The application programs of the sub-devices can for example have device control means and / or device state monitoring means and / or device maintenance systems and / or a spare parts list of at least one sub-device of the device. Furthermore, the application programs of the sub-devices can have or have access to documentation about the sub-devices. For example, the assistant program can access the documentation of the sub-devices, search for information in the documentation and forward the information of the documentation to the application program.

[0058] In one implementation, the auxiliary program uses a maintenance / operation program to determine information about the device and transfers the information, or instructions and / or control values ​​determined based on the information, to the application program. Furthermore, the auxiliary program may record the steps performed. Additionally, the auxiliary program can intervene in the control of the device's functions based on the information and / or instructions and / or control values, and change the control of the device's functions. Furthermore, the application program can determine the control values ​​for the device based on the information and / or instructions themselves. Furthermore, the application program can use the control values ​​to intervene in the control of the device's functions and change the control of the device's functions.

[0059] Here, the auxiliary program and / or application may take into account stored knowledge and / or learned stored knowledge, especially stored knowledge learned using artificial intelligence, and particularly perform evaluations using data analysis methods. For example, the application may be a maintenance / running program.

[0060] In one implementation, the auxiliary program communicates with the application using standardized messages and / or standardized communication protocols.

[0061] In one implementation, the auxiliary program accesses a data storage device, wherein information about which applications are connected via a first interface is stored in the data storage device, and particularly wherein information about what data and / or information the applications possess is stored in the data storage device.

[0062] In one implementation, the auxiliary program communicates with at least two applications, wherein the auxiliary program receives queries for information from the application that made the query. Based on the information searched, the auxiliary program forwards the query to one of the two applications. The application whose query has been forwarded searches for the desired information based on the forwarded query and returns the found information as a response to the auxiliary program. The auxiliary program outputs the information obtained from the application to the application that made the query.

[0063] In one implementation, two applications are configured to search for different information and / or information with different representations and / or information from different sub-devices of the device and / or information from different databases of the device.

[0064] In one implementation, the assistant program forwards the found information to the application making the query; the found information may be, for example, an image, a document, or data. Alternatively, instead of the information itself, the assistant program may forward the URL or storage location where the information can be found to the application making the query.

[0065] Inquiries may include performing pre-defined controls on the equipment, changing the controls of the equipment, and / or providing information, for example, to operators or to applications that perform monitoring and / or control of the equipment.

[0066] The application can be executed by one or more of other computing systems and / or sensor devices and / or human-machine interfaces. BRIEF DESCRIPTION OF DRAWINGS

[0067] The above described features, characteristics, and advantages of the present application as well as ways to realize the features, characteristics, and advantages will become more clearly and fully understood from the following description of the embodiments, considered in connection with the accompanying drawings. In this case, the drawings are diagrammatic and are not to precise scale but are presented for the purpose of illustration only:

[0068] Figure 1 A construction of a device for carrying out a metal production and / or a metal processing is shown;

[0069] Figure 2 A construction of a device for controlling and / or monitoring a technical device with a computing system, other computing systems, sensor devices and human-machine interfaces is shown;

[0070] Figure 3 A diagram of a computer-implemented method for processing a query by a computing system is shown;

[0071] Figure 4 A method flow for processing other queries by means of a computing system is shown;

[0072] Figure 5 A construction of another device for controlling and / or monitoring a technical device with a computing system and other computing systems is shown;

[0073] Figure 6 A construction of another device for controlling and / or monitoring a technical device with a computing system and other computing systems is shown;

[0074] Figure 7 A construction of another device for controlling and / or monitoring a technical device with a computing system and other computing systems is shown, and

[0075] Figure 8 A construction of another device for controlling and / or monitoring a technical device with a computing system and other computing systems is shown. DETAILED DESCRIPTION

[0076] Figure 1A device 2 with the apparatus 1 for controlling and / or monitoring the technical plant 2 is shown in a schematic view. The plant 2 is configured for producing metal and / or further processing metal. For example, the plant 2 can be configured for steel production and / or steel processing. The plant 2 can have a blast furnace, a converter, an electric arc furnace, a machine group for secondary metallurgy, a continuous casting plant, a rolling mill and / or a strip processing line. For example, the plant can be configured as a plant for carrying out iron production, in particular steel production, and for example have a blast furnace, a Finex type, Corex type or Midrex type with or without a submerged arc furnace, a direct reduction plant of the rotary tube or rotary hearth type.

[0077] Furthermore, the plant 2 can have downstream devices in which steel is produced from pig iron. Examples of such plants are electric arc furnaces, converters and plants in which ladle processes are carried out, such as vacuum treatment plants. Furthermore, the plant can have devices downstream of the steel production in which shaping of the metal and reshaping of the shaped metal takes place. Examples of such plants are continuous casting plants and rolling mills. The rolling mill can for example be a rolling mill for rolling flat rolled material, such as a roughing mill train, a finishing mill train, a reversible furnace and strip hot rolling mill and others. Furthermore, the rolling mill can be a rolling mill for rolling an arbitrary cross section, for example a billet cross section. Alternatively, the rolling mill can be a rolling mill for hot rolling of the metal, a rolling mill for cold rolling of the metal or a combined rolling mill, in which the metal is first hot rolled and then cold rolled. Furthermore, the plant can also have a cooling section, if necessary in combination with the rolling mill. Furthermore, the plant can also have further plants upstream or downstream of the rolling mill, such as annealing furnaces or pickling devices.

[0078] The plant 2 can be divided into a plurality of sub-plants 21, 22, 23. The sub-plants 21, 22, 23 assume sub-tasks of the metal production and / or metal processing.

[0079] Furthermore, the apparatus 1 is provided for controlling and / or monitoring the plant 2. The apparatus 1 can have superordinate systems, such as production planning devices and / or logistics systems or control systems, which can for example be implemented by means of at least one or a plurality of the illustrated computing systems.

[0080] The apparatus 1 can comprise a first computing system 3 with a first data store 7, a second computing system 4 with a second data store 8, a third computing system 5 with a third data store 9 and a fourth computing system 6 with a fourth data store 10. Depending on the selected embodiment, more or fewer computing systems can also be provided.

[0081] The first computing system 3 can be set up as a central computing system which is connected with the other computing systems 4, 5, 6. Furthermore, each computing system 3, 4, 5, 6 can be connected with a first, second or third sensor device 11, 12, 13. The sensor devices can be configured in the form of smart sensors. Smart sensors are sensors which combine signal pre-processing and signal processing in one housing in addition to the actual measurement quantity acquisition. Such complex sensors usually comprise, inter alia, a microprocessor or microcontroller, if necessary additionally with DSP functionality, etc., such as complex logic units, such as FPGAs, etc., and provide standardized interfaces for communication with superordinate systems, such as field bus systems, sensor networks, IO links, etc.

[0082] The sensor devices 11, 12, 13 acquire operating states and / or operating parameters of the plant 2 and further report them to the respective computing system 4, 5, 6. Depending on the selected embodiment, the sensor devices 11, 12, 13 can also be connected directly with the first computing system 3. Furthermore, the sensor devices 11, 12, 13 can have their own computing system with a computing unit and an interface and can be configured as smart sensors and connected with the first computing system 3. Furthermore, the plant 2 can have actuators 14, 15, 16 which are manipulated by at least one computing system 4, 5, 6 with control values. The actuators 14, 15, 16 serve to change the operating states of the plant 2.

[0083] Furthermore, at least one human-machine interface 17 is provided which is connected with the first computing system 3 and is configured for inputting queries, receiving queries and outputting information. The human-machine interface 17 can have at least one or more screens. Furthermore, the human-machine interface 17 can have input means in the form of a keyboard, gesture control or microphone and voice control. An operator 26 can query information about the plant 2 via the human-machine interface 17, wherein the information about the plant 2 is outputted via the human-machine interface 17. Furthermore, the human-machine interface 17 can obtain queries from the computing systems and respond to them. To this end, the human-machine interface 17 can have a computing unit and a data memory. An application for operating the human-machine interface can run on the computing unit.

[0084] Each of the computing systems 3, 4, 5, 6 has at least one computing unit, a first interface and at least one program, in particular an auxiliary program and / or an application program, with which data and / or sensor information and / or control values can be processed, received and / or output. The computing systems 4, 5, 6 can have a process automation for the plant 2 in the form of a circuit and / or in the form of an application program. The process automation can comprise a plurality of levels. Level 0 is constituted, for example, by a sensor system and an actuator system. Level 1 constitutes a basic automation for controlling and / or regulating the plant, which basic automation, in particular, implements a regulating loop. Level 2 contains a technical automation, which includes a process model and determines a setpoint for the regulating loop. Furthermore, further levels can be provided, which can comprise, for example, a production planning, a maintenance, a maintenance planning and / or a quality assessment.

[0085] The operation of the plant 2 is generally highly automated, although by means of the computing systems 4, 5, 6, but is not always completely and closed automated. In particular, there is the possibility for an operator 26 to intervene in the automatic control of the plant in the event of a special situation, such as an interference. The operator 26 intervenes in the at least one computing system 3, 4, 5, 6, for example via a human-machine interface 17, for example with the purpose of maintaining a safe operation of the plant and / or of avoiding negative consequences for the operation of the plant itself, the productivity of the plant and / or the product quality as far as possible.

[0086] Figure 2 The basic concept for the cooperation of the first computing system 3 with the second, third and fourth and fifth computing systems 4, 5, 6, 18 is shown in a schematic diagram. The first, second, third, fourth and fifth application programs 31, 32, 33, 34, 35 can be run in the second, third, fourth and fifth computing systems 4, 5, 6, 18, respectively. For example, the second computing system 4 and the third computing system 5 can perform a process automation of the plant 2. Here, the second computing system 4 can perform level 1 of the process automation and the third computing system 5 can perform level 2. The fourth computing system 6 can perform, for example, a condition monitoring of the plant 2. The fifth computing system 18 can perform a process optimization for the entire process, wherein expert knowledge is stored in the form of expert rules in a data store of the fifth computing system 18 and is taken into account by the fifth computing system 18. Furthermore, the computing systems can perform an application program for a quality assurance system. The quality assurance system can be provided for continuously monitoring and controlling the quality in all production processes along the entire production chain.

[0087] In another embodiment, one of the mentioned computing systems can execute an application for a maintenance method. In the context of the maintenance method, expert knowledge is taken into account, which is stored, for example, in a data storage of the computing system, in order to convert non-productive and time-consuming maintenance routines into an intelligent asset management program with which maintenance decisions can be made strategically and in a dynamic manner. The maintenance program can be configured to call up past data for a particular device component. The data can include information about how frequently repair work is required, when the component of the device was last replaced, what improvements were made, etc. Through data analysis, maintenance can be planned more smoothly, more estimable and better.

[0088] The data communication between the first computing system 3 and the second, third, fourth and fifth computing systems 4, 5, 6, 18 takes place, for example, via a first interface 24. The first interface 24 can be configured, for example, in the form of a defined communication protocol with defined messages. For example, the first interface 24 can be implemented in the form of a message broker, which is used to send messages between the first and / or second interfaces of the computing systems and / or the sensor devices. The message broker can implement one of a plurality of possible communication protocols for exchanging messages.

[0089] The first computing system 3 can have an auxiliary program 30, which implements the data communication with the second, third, fourth, fifth computing systems 4, 5, 6, 18, at least one human-machine interface 17 and / or the sensor devices. The human-machine interface 17 is connected to the first computing system 3 or the auxiliary program 30 via a further interface 41, for example a programming interface. The sensor devices can be connected to the first computing system, for example, via the second, third and fourth computing systems, or can be connected directly to the first computing system.

[0090] In addition, the first computing system 3 has a second interface 25 via which further applications 35, 36, 37, 38 are connected to the auxiliary program 30. The further applications can be running on the first computing system 3 and / or on one of the other computing systems 4, 5, 6, 18, and / or the further applications can be implemented as a cloud solution and connected to the computing system 3 via a network connection, for example an internet connection. Each of the further applications 35, 36, 37, 38 can perform one or more functions with which defined information about the device 2 can be determined. The second interface 25 can be configured in the form of a message broker, as is the first interface 24. Depending on the embodiment chosen, other types of data communication can also be used, in particular with defined messages and / or defined communication protocols. Further auxiliary programs 35, 36, 37, 38 can be provided, for example, for finding and / or storing information about the device 2.

[0091] For example, the fifth application program 35 can manage a database of images relating to the device 2. Current and / or past recorded images and / or video sequences of specific sections of the device can be stored in the image database. The sixth application program 36 can have a database of abbreviations, functions, data, operating states, control data, etc. relating to components of the device. The seventh application program 37 can be configured, for example, in the form of a document search program with which documents relating to components, parts and / or sub-devices of the device 2 can be searched. The eighth application program 38 can be configured as a program for noting information relating to operating states and / or control data and / or operating parameters of the device 2.

[0092] The first computing system 3 furthermore has, for example, a speech recognition program 40 with which spoken language can be analyzed and evaluated and information of the spoken language can be determined and / or further processed. For example, a Rasa-NLU program can be used as speech recognition program. Instead of the first and second interfaces, only one interface can also be provided in order to connect the computing system with other computing systems, sensor devices and / or human-machine interfaces and to exchange data.

[0093] The assistance program 30 implements a digital assistance system which can access application programs in the form of other application programs 35, 36, 37, 38 or application programs of other computing systems and / or application programs of sensor devices. Furthermore, the digital assistance system in the form of the assistance program 30 can access data and information of the computing systems 4, 5, 6, 18, sensor devices and / or human-machine interfaces and exchange queries and responses with the computing systems, sensor devices and / or human-machine interfaces.

[0094] The operating personnel 26 thus has the possibility to access all data of the computing systems and sensor devices in a simple manner via the human-machine interface 17 by means of the other application programs 35, 36, 37, 38 and the assistance program 30. Furthermore, the second, third, fourth and fifth computing systems 4, 5, 6, 18 and the sensor devices can access data of other computing systems, other sensor devices and human-machine interfaces by means of the assistance program, wherein the other application programs 35, 36, 37, 38 can also be used.

[0095] Figure 3 A computer-implemented method with which the operating personnel can acquire information relating to the device 2 via the human-machine interface 17 by means of the first computing system 3 is illustrated in a schematic diagram. To this end, the operating personnel enters a query, for example, via a keyboard into an input field 50 of the screen of the human-machine interface 17. The query reads: "Show me some examples". Depending on the selected embodiment, the query can also be entered in the form of spoken language via a microphone. The query is represented in the input field 50 of the screen of the human-machine interface 17.

[0096] In a subsequent step, the query (Search Request Message) is handed over to the programmatic interface 41 of the first computing system 3. If the query is present in the form of a written text, the query is handed over to a recognition program 40. In the recognition program 40, a function (Intent) is filtered out of the text and, if present, a content (Entity) of the function, which specifies the function more precisely. The function can be, for example, a search function for information about a device, a control function for controlling a device, a monitoring function for monitoring a device. The content can specify, for example, which information about a device is searched for. Furthermore, the content can specify which function of a device should be controlled and, in particular, how the function of a device should be controlled. Furthermore, the content can specify which part of a device should be monitored with respect to which parameter. To this end, the recognition program has a data model of pre-given texts assigned to pre-given functions. The inputted text is compared to the stored pre-given texts. From the comparison, at least one pre-given text is recognized as being in agreement with the inputted text. Furthermore, the function assigned to the agreed pre-given text is determined. The data model for determining the function and the content can have been determined beforehand from experiments.

[0097] The data model can be configured by means of artificial intelligence, in particular as a trained neural network. The artificial intelligence is trained and configured for determining the function and / or the content of the function. Furthermore, the artificial intelligence can be configured for determining other computing systems and / or sensor devices and / or human-machine interfaces and / or applications, which can provide the searched function and the searched content, i.e. the information searched for with the query.

[0098] Furthermore, the artificial intelligence can be configured for determining the function of an application and, if present, the content that should be determined with the function, so that the information searched for with the query can be provided. The artificial intelligence can be realized in the form of hardware and / or software. For example, the RASA NLU program can be used as a recognition program.

[0099] If the query is present in the form of a spoken language, the query is handed over to a recognition program 40, with which the speech input can be processed. In the recognition program 40, the function (intention) and, if present, the content (entity) specifying the function more precisely are filtered out of the spoken query. As set out above, the recognition program can also be configured in the form of artificial intelligence, in particular in the form of a trained neural network. The function can be, for example, a search function, a control function, a monitoring function. The content can specify, for example, which information about the device is to be searched. Furthermore, the content can specify how a function of the device is to be controlled. Furthermore, the content can specify which part of the device is to be monitored with regard to which parameter. To this end, the speech recognition has a data model of predefined speech inputs assigned to predefined functions. The input speech input is compared with the stored predefined speech inputs. From the comparison, at least one predefined speech input is recognized as being in agreement with the input speech input. Furthermore, the function assigned to the agreed predefined speech input is determined. The data model of the speech input for determining the function and the content can already have been determined beforehand from experiments. For example, the RASA NLU program can be used as speech recognition.

[0100] Furthermore, the recognition program can be configured to specify, in addition to the determined function (intention), also a probability that the determined application program and / or the determined function can be correctly assigned to the query. When using artificial intelligence, as in the case of the RASA NLU program, the probability that the assignment is correct is additionally specified for the determined result. Thus, a plurality of further application programs and / or functions can be determined for the query with different probabilities. In the case of other methods, first the one or more application programs and / or functions with a higher probability are executed in order to obtain the searched information as quickly as possible and with as little outlay as possible. Thus, a plurality of application programs and / or functions can be executed in order to obtain the searched information.

[0101] At a subsequent program point, the determined function (intent) and, if present, at least one determined content (entity) of the function are handed over to the first computing system 3. The content can also represent at least one parameter of the function, for example. The parameter can represent a control value of a control program of the device or a parameter of the desired information, for example. The determined content can be a parameter of the determined function, for example. In the present example, the "search example" is transmitted to the first computing system 3 as the determined function. The first computing system 3 compares the identified function "search example" with a pre-given data model. The data model contains pre-given assignments between functions and pre-given applications. In this example, the first computing system 3 determines, on the basis of the pre-given data model, an eighth application 38 which is capable of searching examples. The first computing system 3 hands over the query "search example" to the eighth application 38 via the second interface 25. The second interface 25 uses a message broker, for example, for communication with the eighth application 38.

[0102] The message broker of the second interface transmits the query as a TApp request message to the eighth application 38. The eighth application 38 searches a corresponding database for a corresponding example in accordance with the received query, which the eighth application can access. The eighth application 38 provides the found example, for example in the form of data or an image, as a response back to the second interface 25 of the computing system 3 via the message broker 25. In addition or instead of the found example itself, a memory address, such as an internet address, for the found example can also be transmitted to the computing system. By means of the message broker, the found example and / or the memory address are sent back to the human-machine interface 17 in the form of a response via the programming interface 41. The human-machine interface 17 represents the transmitted example, in particular the image of the device and / or the memory address and / or the memory address in a display window 58. Depending on the selected embodiment, the human-machine interface 17 can retrieve and represent the example itself from the memory address.

[0103] Figure 4 The program flow for determining information about the device 2 and / or the control of the device 2 by means of a query via the human-machine interface 17 is illustrated in a schematic diagram. Here, the query is entered into the human-machine interface 17 by the operating personnel via a microphone. The query content is: show me an image of the rolling mill. The spoken query 51 is handed over from the human-machine interface 17 to the computing system 3 via the programming interface 41. The programming interface 41 forwards the query 51 to the recognition program 40. The recognition program has speech recognition and is constructed as an artificial intelligence, in particular as a trained neural network, for example. The artificial intelligence is implemented in the form of a Rasa-NLU program, for example.

[0104] Instead of a query in the form of a speech signal, the query can also be present in the form of text and / or graphics and / or images.

[0105] As described above, the recognition program 40 determines a function from the query and, if possible, determines what should be searched with the function. The function corresponds to an application program which is available to the computing system, the human-machine interface and / or the sensor device or is available via a network connection. The function can also only represent a part of an application program. In particular, an application program can have different functions. With each function different content can be searched for, for example.

[0106] The recognition program uses, for example, a first data model which is set up for a plurality of other computing systems and / or sensor devices and / or human-machine interfaces and / or other application programs. The recognition program determines, from the query and by means of the first data model, at least one other computing system and / or at least one sensor device and / or at least one human-machine interface and / or at least one application program which can provide information searched with the query.

[0107] The first data model has prescribed examples, in particular example sentences, for the query with which information can be successfully searched in the other computing systems and / or in the sensor devices and / or in the human-machine interfaces and / or in the other application programs.

[0108] The first data model can be constructed by means of artificial intelligence, in particular as a trained neural network. The artificial intelligence is constructed for determining, from the query, other computing systems and / or sensor devices and / or human-machine interfaces and / or application programs which can provide information searched with the query.

[0109] For example, the first data model can be constructed as a trained neural network which has been trained with the following example sentences for the application program "displaying images of devices" and / or for the application program "searching for documents of devices".

[0110] The following example sentences have been used for training the neural network for the application program "displaying images of devices":

[0111] - Show me a photo of the rolling mill train.

[0112] - I would like to see an image of the hot strip mill train.

[0113] - Is there a picture of the blast furnace

[0114] - I am interested in a video about the finishing mill train.

[0115] The following example sentences have been used for training the neural network for the application program "searching for documents of devices":

[0116] - I need a mechanical drawing of the finishing mill stand.

[0117] - show me the data sheet of the motor.

[0118] In response 52, the recognition program delivers an application program with which the query 51 can be processed.

[0119] In this case, the fifth application program "Plant Visuals" is handed over to the third computing system 3 as response 52.

[0120] The third computing system 3 then creates a second query 52 with the query "show me an image of the rolling mill" using the fifth application program "Plant Visuals". The second query 52 is in turn handed over to the recognition program 40. The second query contains the same query "show me an image of the rolling mill" as the first query and additionally contains the information that this query should be executed by means of the fifth application program "Plant Visuals".

[0121] The recognition program has a second data model, which is set up for the other computing systems or sensor devices or human-machine interfaces or other application programs respectively. It is thus possible to set up a second data model for each other computing system and / or for each sensor device and / or for each human-machine interface and / or for each other application program respectively.

[0122] The recognition program determines from the second data model of the fifth application program "Plant Visuals" and the query "show me an image of the rolling mill" which function of the fifth application program and with which content the searched information can be determined.

[0123] The second data model is for example constructed as an artificial intelligence, in particular as a trained neural network, which is constructed for determining the application program and / or the function and in particular the content of the function of the other computing systems and / or sensor devices and / or human-machine interfaces and / or other application programs which can provide the searched information. The second data model can be realized by means of the program RASA NLU program. The second data model has for the fifth application program prescribed examples, in particular example sentences, for the query with which the information can be successfully searched in the other computing systems and / or in the sensor devices and / or in the human-machine interfaces and / or other application programs by means of the function of the fifth application program and preferably the content of the function of the fifth application program.

[0124] The second data model can be constructed by means of a trained neural network. The second data model for the application program "show me an image of the device" can already be trained with the following example sentences for the function "show image" and the function "show video":

[0125] For the function (intent) "show image" the following example is used for training the neural network:

[0126] Show me a picture of the [rolling mill train] (asset)

[0127] I want to see an image of the [hot strip mill train] (asset)

[0128] Is there a picture of the [blast furnace] (asset)

[0129] The content (entity) of the function is specified in square brackets and the type of the content (entity) is specified in round brackets. By means of the type, the content can be distinguished or classified again.

[0130] For the function (intent) "show video" the following example is used for training the neural network:

[0131] I am interested in a video of the [finishing mill train] (asset)

[0132] The content (entity) of the function is specified in square brackets and the type of the content is specified in round brackets.

[0133] A second data model can be constructed by means of the trained neural network. The second data model for the application "search for documents of a device" can have been trained with the following example:

[0134] For the function (intent) "show document" the following example is used for training the neural network:

[0135] I need a [mechanical drawing] (document type) of the [finishing mill train] (asset)

[0136] Show me a [data sheet] (document type) of the [electric motor] (asset)

[0137] The content (entity) of the function is specified in square brackets and the type of the content is specified in round brackets. In the mentioned example, two contents with different types are specified, respectively.

[0138] In a similar manner, the neural network can also be trained for other applications, in particular performing a control or monitoring of a device.

[0139] The recognition program 40 determines as function "search for image" and as content "rolling device" on the basis of the second query 52 and the second data model.

[0140] The information is returned as a second response 54 to the assistant program 30 of the first computing system 3. The assistant program of the computing system 3 creates a third query 55 from the transmitted function "search image" and the content "rolling mill" which is transmitted to the fifth application program 35 having the function of searching for images of rolling mills by means of the message broker.

[0141] The fifth application program 35 searches for images of rolling mills in the connected data storage after the query has been obtained. If images of rolling mills are found by the fifth application program 35, these images are sent back by the fifth application program 35 to the first computing system 3 as a third response 56. On the basis of the third response 56, the first computing system 3 creates a fourth response 57 which is transmitted to the human-machine interface 17. The third response 57 contains at least one image of a rolling mill. The human-machine interface 17 receives the third response 56 and represents the at least one transmitted image on the screen.

[0142] According to the selected embodiment, according to Figure 2 and Figure 3 Not only the human-machine interface 17 can send a query to the first computing system 3, but also, for example, the second, third, fourth and / or fifth computing systems 4, 5, 6, 18 and / or sensor devices and / or application programs can transmit a query to the first computing system 3. The first computing system 3 processes the queries of the other computing systems 4, 5, 6, 18 in the same way as the queries of the human-machine interface and / or the application programs and returns a response to the query accordingly. In this way, information can be exchanged in a simple manner between the computing systems, sensor devices and application programs connected to the first computing system 3.

[0143] Furthermore, the described computer-implemented method can also be used to respond to a query by means of different application programs and / or computing systems and / or sensor devices and / or human-machine interfaces, in particular to control a device.

[0144] For example, for a query which leads to a control of a device, the models, in particular trained neural networks, can be used in a similar way in order to find an application program which can execute the query. After the application program has been determined which can execute the query, the control instructions contained in the query are handed over to the determined application program in order to control the device accordingly.

[0145] Here are some examples for queries which execute a control of a device, wherein the first and second data models are used to determine a suitable application program which can execute the query. The first and / or second data model is, for example, constructed as a trained neural network.

[0146] A first example content for a control-oriented query is: reduce the bending force of the frame F3 of the plant by 100 kN. By means of the first data model, a function control of the plant (in particular a level 2 control) is determined from the query. Thus, the application program that can execute the query is the control program of the plant. The second data model for the control program of the plant determines from the query the function "reduce" (an execution element of the plant) and the content: bending force (= specific execution element), frame F3 (asset), 100 kN (value). The type of the content is indicated in brackets. Thus, the task of manipulating the respective execution element in the frame F3 of the plant in such a way that the bending force is reduced by 100 kN is handed over to the control program.

[0147] A second example content for a control-oriented query is: mark the current product, in particular the strip, in the finishing train, in particular in the rolling plant, for quality analysis.

[0148] The first data model identifies from the query as a function the "quality monitoring", which represents the supervising application program.

[0149] The second data model for the quality monitoring application program identifies from the query the function "inspect" with the content "current" (time: NOW (now)) and the content "strip" (asset). The type of the content is indicated in brackets. The quality monitoring application program will query another application program, for example the plant tracking program, in order to mark the current product. The marking serves to identify the current product later on from the marking and then to be able to analyze the marked product.

[0150] A third example content for a control-oriented query is: give me information about the current strip of the plant. The first data model identifies from the query as a function the plant tracking, which represents the supervising application program. The second data model for the plant tracking application program identifies from the query as a function the "strip information" with the content: current (time "NOW"). Thus, the information about the current strip can be queried from a respective data storage by means of the plant tracking application program or determined from the current sensor data of the sensors of the plant.

[0151] For example, the interrogated application or the interrogated computing system and / or the interrogated sensor device and / or the interrogated human-machine interface can send further interrogations for further information about the device back to the first computing system 3 from which the interrogation was transmitted in response to the interrogation of the information. The first computing system then determines, for example by means of the recognition program 40, from the further interrogations which further computing system 4, 5, 6, 18 and / or which sensor device 11, 12, 13 and / or which human-machine interface 17 can respond to a further interrogation. The first computing system 3 then acquires the interrogated further information from at least one determined further computing system 4, 5, 6, 18 and / or from at least one determined sensor device 11, 12, 13 and / or from at least one determined human-machine interface 17 via the interfaces 24, 25. The first computing system 3 forwards the acquired further information to the interrogating further computing system 4, 5, 6, 18 and / or to the interrogating sensor device 11, 12, 13 of the device and / or to the interrogating human-machine interface 17 via the interfaces 24, 25. The interrogating further computing system 4, 5, 6, 18 and / or the interrogating sensor device 11, 12, 13 of the device and / or the interrogating human-machine interface 17 processes the further information in order to determine, for example, the interrogated information and then forwards the interrogated information to the first computing system. Thereby, the interrogated information can be created by means of a plurality of further computing systems 4, 5, 6, 18 and / or sensor devices 11, 12, 13 and / or human-machine interfaces 17 of the device.

[0152] Figure 4 The example described in the middle shows a free field search input, in which the recognition program 40 is called in order to find out which application (App) is suitable for performing the interrogation. The recognition program calls a first data model, in which example sentences for at least a part, in particular for all available applications, are contained.

[0153] The interrogation is compared to the prescribed examples, in particular example sentences, of the first data model. An example, in particular an example sentence, of the first data model is selected which has the best agreement with the interrogation according to a predefined rule. The predefined rule can select the example, in particular the example sentence, which has the best agreement, in particular the greatest number of consistent words, with the interrogation. The application belonging to the selected example, in particular example sentence, is reported back to the first computing system 3. According to the selected embodiment, more than one application with a quality of agreement can also be transmitted to the first computing system with an indication of the quality of agreement.

[0154] The examples, in particular the example sentences, for the respective application can have been determined by means of artificial intelligence, in particular by means of a trained neural network. Furthermore, the comparison of the query with the example sentences of the first data model and the determination of one or more examples, in particular example sentences, which are best consistent with the query can be determined by means of artificial intelligence, in particular by means of a trained neural network. The artificial intelligence can additionally determine a probability of the consistency of the examples with the query. The probability is handed over to the first computing unit together with the selected examples, in particular example sentences, and the assigned application.

[0155] Furthermore, for the query one or more of the other applications can be determined by means of a trained neural network, with which the query can be executed or responded to with the greatest probability. The artificial intelligence can additionally output a probability of how well the application determined by means of the neural network is suited to execute the query.

[0156] Preferably, only the application with the highest probability is used for the second step with the second data model. Furthermore, it is also possible to use the applications whose probability exceeds a pre-given value for the second step with the second data model.

[0157] The application determined by means of the first data model and preferably the probability belonging to the determined application, which states how well the application fits the search query, i.e. the query, is returned as a list to the first computing system 3, in particular to the assistant program 30.

[0158] In the described example, the recognition program recognizes that an image has been queried and, therefore, the facility visual material application can be used for the search. A specific example of the NLU program is the Rasa-NLU program, which is implemented using artificial intelligence, in particular using a trained neural network.

[0159] For each application of the returned list, the recognition program is called again with the same query and with the second data model set for the respective application. The second data model has other specified examples, in particular example sentences, which are specified in particular for the respective application. Each example, in particular each example sentence, of the second data model can be assigned to a specific function of the application. The application can also have only one function.

[0160] The second data model is constructed, for example, as artificial intelligence, particularly as a trained neural network, wherein the AI ​​is configured to determine at least one function that can perform queries or provide information using queries for searching. Furthermore, the neural network can output probabilities for the determined function, indicating how well the function is suited to perform queries. The second data model can be implemented, for example, using a RASA NLU procedure.

[0161] It is now possible to determine, with the aid of artificial intelligence, particularly trained neural networks, examples, especially example sentences, for each data model set up only for an application. Furthermore, with the aid of artificial intelligence, particularly trained neural networks, it is now possible to determine the comparison between the query and the specified examples, especially example sentences, of the second data model, and to determine one or more examples, especially example sentences, that best match the query. The artificial intelligence additionally outputs the probability that the example sentence matches the query. This probability, along with the application assigned to the selected example sentence, especially example sentence, is transferred to the first computing unit.

[0162] For each application, the recognition program can have its own trained second data model, specifically a trained neural network. For example, for each function available in the application, the second data model can have example sentences with different grammar and vocabulary. Thus, a mapping can be created between queries and the respective functions of each application.

[0163] The identification program provides the first calculation unit with the name of the function to be invoked in the application, or a list of functions, along with their probabilities and, preferably more accurately, additional information describing the function's execution, such as content and parameters. If this information is missing, it can be retrieved, for example, from the human-machine interface and thus from the operator, to request the missing content and parameters.

[0164] exist Figure 4 In the example, the recognition program, using a data model trained on facility visual data, determines that the function "Search Image" should be invoked. Here, the additional information is "rolling equipment".

[0165] The auxiliary program 30 sends a message via a message broker to the fifth application facility visual data containing the extracted information. In the facility visual data application, the function "Search Image" with the additional information "For rolling mill" is invoked. In response, a list of images of the rolling mill is transmitted to the auxiliary program. The auxiliary program receives the response via the message broker, in this case, the list of images of the rolling mill. In response to the query, the auxiliary program transmits the list of images of the rolling mill to the human-machine interface 17.

[0166] Alternatively, the search query can also be transmitted to the assistant from an application, in particular from an application of another computing system. For example, the query can also be sent directly to the assistant via the message broker by another application without having to use the programming interface. Here, three possibilities can arise for the search query. For example, when the search text is free text, it can not be known which application should be queried. Here, the first application is determined by means of the first data model as described above and as described above. Furthermore, it can already be known which application should be queried, but it is not known which function within the application should be called and which additional information is present. Accordingly, only the above-described steps of the second data model for the known application are carried out. Furthermore, it can already be known which application should be queried, which function should be called and which additional information should be taken into account. In the context of such a search query, the individual applications can also again query additional information at other applications, as has already been described above. However, it is further ensured that the chaining of the search query is terminated.

[0167] Figure 5 The structure of the computing system for the device 2 is shown in a schematic view. Here, the first computing system 3, in particular the assistant 30 of the first computing system, is connected with a device automation 39, which also has a computing system. Furthermore, the first computing system 3, in particular the assistant 30 of the first computing system, is connected with a human-machine interface 17. Furthermore, a sensor device in the form of a camera 44 is connected not only with the first computing system 3, in particular with the assistant 30 of the first computing system, but also, for example, with the device automation 39. Furthermore, a display system 43, for example in the form of a monitor, is provided, which is directly connected with the camera 44. Furthermore, an application in the form of an interactive operator guidance 42 is provided, which is connected with the first computing system 3, in particular with the assistant 30 of the first computing system. Furthermore, the interactive operator guidance 42 can also be directly connected with the camera 44.

[0168] Figure 5 The program structure has the advantage that the device 2 can be controlled with less personnel and from a control room which cannot look directly at the device. In this way, even several consoles from several devices can be combined in one control room. Several devices can thus be operated simultaneously from one control room. Furthermore, the control center and the control room can also be arranged remotely and independently of the location of the device.

[0169] For enabling the operator to manipulate the plant, i.e. to operate the plant, a screen in the form of a human-machine interface 17 is provided, which enables looking through the camera 44 into a section of interest of the plant. In addition, the human-machine interface 17 provides an operating screen in order to have an overview about the plant automation. In addition, changes in the system automation, i.e. in the control of the plant, can be made through the operating screen of the human-machine interface. However, the following problem can arise: the number of screens required, which can be up to twelve monitors, exceeds the understanding of the operator and must therefore be reduced in display. It is therefore advantageous to switch the display on the monitors in terms of scenarios onto the respective section of the plant depending on the plant status and the operating status. The switching can be made automatically by means of an application of the human-machine interface according to the method described above. The plant status and the operating status can occur through information from the plant automation, through an evaluation of the camera images, for example in the case of the use of artificial intelligence, or through other evaluations of the application of the first computing system. The other evaluations can be connected to the aid program 30 of the first computing system 3 in the interactive operator guidance application 42 together with a knowledge-based, if necessary jointly learned, decision algorithm, which switches the monitors. The aid program 30 represents a digital assistant for monitoring and / or controlling the plant.

[0170] The plant automation is configured for using the images of the camera system for the plant automation.

[0171] By means of the described method, the plant automation can independently acquire, evaluate images of the plant and, depending on the evaluation, perform a change in the control value of a function of the plant in order to achieve or maintain a desired manner of functioning. Due to the direct connection, a low latency between the transmission of the images and the reaction of the system automation is achieved. The display system of the human-machine interface, which can comprise monitors, a video wall, etc., can also be directly connected to the aid program 30, i.e. to the core of the first computing system 3.

[0172] An operator can make and parameterize scenarios via the interactive operator guidance of the human-machine interface. It can be specified here which images of which camera are represented on which monitors and in the case of which events. Furthermore, it can be specified which events lead to a predefined scenario. By means of the scenarios it can be specified when which camera image is represented on which monitor of the display system of the human-machine interface. Furthermore, the control of the scenarios can be specified by voice input at the human-machine interface. Here, fixed commands can be deposited in order to enable voice recognition. Furthermore, the control of the individual monitors can be specified by means of a digital assistant, i.e. an assistant program 30. Here, fixed commands can be deposited in order to represent the individual monitors on the screen purposefully beyond the defined scenarios. For example, in the case of a sudden disturbance, the display monitors can be equipped with specified information. It can be specified here that, depending on the type of disturbance, specified images of specified cameras or device states are represented on the monitors of the display system.

[0173] Figure 6 A further embodiment of the program structure of the device 2 is shown. In the case of this embodiment, the assistant program 30 of the first computing system 3 is connected with the device automation 39, the condition monitoring program 33 and further applications 34. Furthermore, the human-machine interface 17 is connected to the assistant program 30 of the first computing system 3. Furthermore, a diagnostic application 45 is connected to the assistant program 30 of the first computing system 3.

[0174] The status of a device for metal production and metal processing can be acquired in a complex and difficult manner. However, for many applications it is convenient to obtain as complete a picture as possible about the status of the device. For example, an error analysis, a quality analysis, a maintenance planning and / or a further development to achieve a device improvement can be desired. Since the assistant program connects different computing systems and sensor devices of sub-devices of the device with one another, the overall status of the device can be aggregated at the desired point in time / period. For example, the device status can be determined from the status of a condition monitoring system, the status of a device automation, the status of a production quality from a quality management system, a maintenance status from a maintenance system and / or a production plan of a production management system, etc.

[0175] Furthermore, additional information can be automatically shaped from the above-mentioned and aggregated data by means of technical knowledge. Thereby, also action recommendations, such as a control intervention of the control of the plant or other more complex conclusions, can be given. A faster status analysis is achieved by the proposed improvement. A better knowledge of the functional capability of the plant or production is given. Furthermore, the following possibilities are provided thereby: a more accurate execution of the further control and analysis of the plant. Since the knowledge shaped technically in the assistance program, a constant execution of the status analysis, which is independent of the operating personnel, is possible. It is also possible to more simply compare repeated status analyses. Furthermore, the diagnostic program can already be used during the start-up of the plant to identify plant parts which have not been completely started up or to check and document a correct start-up thereof.

[0176] Figure 7 A further embodiment of the program structure of the plant 2 is shown, which substantially corresponds to the embodiment according to Figure 6 However, instead of or in addition to the diagnostic program 45, a documentation program 46 is connected to the assistance program 30 of the first computing system 3.

[0177] Plants for carrying out metal production or metal processing are usually composed of many different individual units, which also partly come from different manufacturers. For each individual unit, documentation material is stored in a data store. It is also meaningful to be able to access the documentation material of the individual units of the plant from different systems, such as a plant automation, a condition monitoring, a maintenance system, a spare parts catalog, etc., depending on the operating state. However, advantageously, the documentation for the entire plant should only be present in one location, i.e. in the data store, in order to avoid different versions or high maintenance costs. If different systems, such as a plant automation, a condition monitoring, a maintenance system, a spare parts catalog, etc., are connected to the assistance program and a documentation management device with a search function is connected to the systems, this function can be simply provided. Thus, in addition to the automatic access by one of the above-mentioned systems, access can also be made by the operating personnel via a man-machine interface. The systems can thus be expanded with documentation material with a clear objective. The search function for the documentation can in particular make use of the artificial intelligence provided by the first computing system 3.

[0178] Figure 8 A further program structure of the plant 3 is shown, which substantially corresponds to the embodiment according to Figure 6 However, instead of or in addition to the diagnostic program 45, a documentation program 46 is connected to the assistance program 30 of the first computing system 3.

[0179] A maintenance and operation assistant can be connected to the aid, which can give information or instructions to the operating personnel via a human-machine interface and can document the executed steps. Furthermore, the maintenance / operation aid can also intervene directly in the device operation. The maintenance / operation program can be based on stored knowledge or can take into account stored information and rules as well as independently learned knowledge, for example, by artificial intelligence. Here, too, an evaluation can be carried out with the aid of data analysis methods (deep learning, machine learning, support vector machines, etc.). Furthermore, a special maintenance / operation aid can be used for each application case.

[0180] The use of an aid offers the following advantages: it is possible to aggregate further applications in a simple manner. Thus, it is possible to expand the device with further single units or to remove individual units from the device and to change the respective documentation or program flexibly independently of the functionality of the respective single unit or independently of the functionality of the respective application.

[0181] The proposed system with a modular structure consisting of an aid and sub-devices, so-called applications, which can be run internally in a first computing system or externally on other computing systems, offers improved flexibility. The communication between the sub-devices and the aid takes place preferably via standardized messages and standardized communication protocols. For this, for example, a message broker can be used, which serves for secure message transmission. Buffering messages and storing them in a row makes it possible to check whether a message has actually been delivered to this.

[0182] Furthermore, it can be advantageous to automatically resume the connection establishment after a connection interruption. The basic establishment can be in accordance with the Apache Artemis program. With the aid of the proposed system, defined and free search queries can be made to information and / or knowledge without knowing specifically which other sub-device can provide and will provide which information in which way. Free search inquiries (for example voice, text, image or video input) can be converted into defined search queries with the aid of an NLU program (Natural Language Unit). Here, the NLU can use artificial intelligence to improve the conversion incrementally. In the prior art, there are NLU's that already partly enable the possibility of using different languages.

[0183] The described system can be implemented in a distributed computer network, if necessary also via the Internet. Information, knowledge and its associations can be stored in the applications.

[0184] By means of the described system, the communication of the sub-devices can be implemented for the exchange of information and / or knowledge. Defined as well as free search queries to the information and / or knowledge can be performed here without the specific knowledge of which other sub-device can provide and will provide which type of information. For the free search queries, different languages are available to the operator, which are also scalable. Thereby, information and knowledge can be extracted from the different languages.

[0185] Furthermore, even if it is not known a priori which sub-device of the device is present at a point in time, the modular structure of the system enables the functioning of the assistance program. Explicitly, the sub-devices can be exchanged, extended, modified, added or removed. In particular, the performance range of the sub-devices and thereby the provided information can change. Thereby, it can be achieved that a changed performance range of one of the sub-devices does not influence the general functioning manner of the other sub-devices in terms of communication and functionality adaptation. However, by changing the information of the sub-devices, the performance range of the other sub-devices can be enlarged or reduced here. Furthermore, the knowledge, which is scalable by the operator in the respective application, can be shaped and provided in the sub-devices. Furthermore, the knowledge can be applied to the information in order to generate added value for the use of the digital assistant. Furthermore, the communication between the sub-devices and between the sub-devices and the operator is shaped by means of free expressed queries to the information in the communication (see the aggregation of the NLU in the assistance program), which enables a simple operation without interface-specific know-how.

[0186] List of reference signs

[0187] 1 device

[0188] 2 apparatus

[0189] 3 first computing system

[0190] 4 second computing system

[0191] 5 third computing system

[0192] 6 fourth computing system

[0193] 7 first data storage

[0194] 8 second data storage

[0195] 9 third data storage

[0196] 10 fourth data storage

[0197] 11 first sensor device

[0198] 12 second sensor device

[0199] 13 third sensor device

[0200] 14 first actuator

[0201] 15 second actuator

[0202] 16 third actuator

[0203] 17 human-machine interface

[0204] 18 fifth computing system

[0205] 21 first sub-device

[0206] 22 second sub-device

[0207] 23 third sub-device

[0208] 24 first interface

[0209] 25 second interface

[0210] 26 operator

[0211] 30 assistance program

[0212] 31 first application program

[0213] 32 second application program

[0214] 33 third application program

[0215] 34 fourth application program

[0216] 35 fifth application program

[0217] 36 sixth application program

[0218] 37 seventh application program

[0219] 38 eighth application program

[0220] 39 device automation device

[0221] 40 identification program

[0222] 41 further interface

[0223] 42 interactive operator interface

[0224] 43 display system

[0225] 44 camera

[0226] 45 diagnostic program

[0227] 46 documentation program

[0228] 47 first maintenance / operation program

[0229] 48 second maintenance / operation program

[0230] 49 third maintenance / operation program

[0231] 50 input field

[0232] 51 query

[0233] 52 response

[0234] 53 second query

[0235] 54 second response

[0236] 55 third query

[0237] 56 third response

[0238] 57 fourth response

[0239] 58 display window

Claims

1. An apparatus (1) for controlling and / or monitoring a technical device (2) for producing and / or processing metal, having a computing system (3) with at least one interface (24, 25) for connecting with at least one other computing system (4, 5, 6, 18) of the device and / or with at least one sensor device (11, 12, 13) of the device (2) and / or with at least one human-machine interface (17) and / or with an application (31, 32, 33, 34, 35, 36, 37, 38), wherein the computing system (3) is configured to obtain, via the interface (24, 25), a query for information about the device (2) from at least one querying other computing system (4, 5, 6, 18) and / or from a querying sensor device (11, 12, 13) of the device and / or from a querying human-machine interface (17) and / or from a querying application (31, 32, 33, 34, 35, 36, 37, 38), wherein the computing system (3) has access to a data model via the other computing system (4, 5, 6, 18) and / or the sensor device (11, 12, 13) and / or the human-machine interface (17) and / or the other application (31, 32, 33, 34, 35, 36, 37, 38), wherein the data model gives a hint as to which information the other computing system (4, 5, 6, 18) and / or the sensor device (11, 12, 13) and / or the human-machine interface (17) and / or the other application (31, 32, 33, 34, 35, 36, 37, 38) has access to, wherein the computing system (3) is configured to determine the other computing system (4, 5, 6, 18) and / or the sensor device (11, 12, 13) and / or the human-machine interface (17) and / or the other application (31, 32, 33, 34, 35, 36, 37, 38) that can provide the searched information from the data model, wherein the computing system (3) is configured to obtain the queried information via the interface (24, 25) from at least one determined other computing system (4, 5, 6, 18) and / or from at least one determined sensor device (11, 12, 13) and / or from at least one determined human-machine interface (17) and / or from at least one other application (31, 32, 33, 34, 35, 36, 37, 38), wherein the computing system (3) is configured to output the obtained information via the interface (24, 25) to the querying other computing system (4, 5, 6, 18) and / or to the querying sensor device (11, 12, 13) of the device and / or to the querying human-machine interface (17) and / or to the querying application (31, 32, 33, 34, 35, 36, 37, 38). wherein the data model has prescribed examples for search queries for the other computing systems (4, 5, 6, 18) and / or sensor devices (11, 12, 13) and / or human-machine interfaces (17) and / or the other applications (31, 32, 33, 34, 35, 36, 37, 38), with which examples information can be searched in the other computing systems (4, 5, 6, 18) and / or the sensor devices (11, 12, 13) and / or the human-machine interfaces (17), wherein the computing system (3) is configured to determine, by comparing the prescribed examples with the query, which other computing system (4, 5, 6, 18) and / or which sensor device (11, 12, 13) and / or which human-machine interface (17) and / or which other application (31, 32, 33, 34, 35, 36, 37, 38) can provide the searched information, wherein the computing system (3) is configured to determine from the data model at which probability the other computing systems (4, 5, 6, 18) and / or the sensor devices (11, 12, 13) and / or the human-machine interfaces (17) and / or the other applications (31, 32, 33, 34, 35, 36, 37, 38) can provide the searched information, wherein the computing system forwards the query to the determined other computing system and / or the determined sensor device and / or the determined human-machine interface and / or the determined application with the highest probability, wherein the data model is configured as a trained neural network, wherein the neural network is configured to determine other computing systems and / or sensor devices and / or human-machine interfaces and / or applications that can provide information searched with a query, wherein the neural network has been trained by means of a supervised learning method with previously given examples in order to determine from a query other computing systems and / or sensor devices and / or human-machine interfaces and / or applications that can provide information searched with the query, wherein the examples are configured in a similar or identical manner to the query.

2. The apparatus according to claim 1, wherein the computing system is configured to determine from an existing query the information searched by the query via a natural language unit, NLU, procedure.

3. The apparatus according to claim 1 or 2, wherein the computing system (3) is configured to determine, by comparing the prescribed examples with the information searched with the query, which other computing system (4, 5, 6, 18) and / or which sensor device (11, 12, 13) and / or which human-machine interface (17) and / or which other application (31, 32, 33, 34, 35, 36, 37, 38) can provide the searched information.

4. The apparatus according to claim 1 or 2, wherein the computing system (3) is configured to obtain further inquiries for further information about the device from at least one determined further computing system (4, 5, 6, 18) and / or from at least one determined sensor apparatus (11, 12, 13) and / or from at least one determined human-machine interface (17) and / or at least one further application program (31, 32, 33, 34, 35, 36, 37, 38) in response to the inquiry for information, wherein the computing system (3) is configured to determine, by means of the data model, which further computing system (4, 5, 6, 18) and / or which sensor apparatus (11, 12, 13) and / or which human-machine interface (17) and / or which further application program (31, 32, 33, 34, 35, 36, 37, 38) is able to respond to the further inquiries depending on the further inquiries, wherein the computing system (3) is configured to acquire the inquired further information from the at least one determined further computing system (4, 5, 6, 18) and / or from the at least one determined sensor apparatus (11, 12, 13) and / or from the at least one determined human-machine interface (17) and / or from the at least one further application program (31, 32, 33, 34, 35, 36, 37, 38) via the interface (24, 25), wherein the computing system (3) is configured to output the acquired further information to the inquiring further computing system (4, 5, 6, 18) and / or to the inquiring sensor apparatus (11, 12, 13) and / or to the inquiring human-machine interface (17) and / or to the inquiring application program (31, 32, 33, 34, 35, 36, 37, 38) of the device via the interface (24, 25).

5. The apparatus according to claim 1 or 2, wherein the computing system (3) is configured to acquire the specified information from at least one further computing system (4, 5, 6, 18) and / or sensor apparatus (11, 12, 13) and / or human-machine interface and / or further application program (31, 32, 33, 34, 35, 36, 37, 38) via the interface (24) depending on at least one predefined operating state of the device (2) and / or depending on a predefined control value for the device (2), wherein the computing system (3) is configured to output the independently acquired information to the further computing system (4, 5, 6, 18) and / or to the sensor apparatus (11, 12, 13) and / or to the human-machine interface (17) and / or to the further application program (31, 32, 33, 34, 35, 36, 37, 38) via the interface (24).

6. The device according to claim 1 or 2, wherein further computing systems (4, 5, 6, 18) for sub-devices (21, 22, 23) of the plant (2) are connectable via the interface (24), wherein the further computing systems (4, 5, 6, 18) execute a plant control and / or status monitoring and / or maintenance system for at least one sub-device (21, 22, 23) of the plant (2) and / or have a spare parts catalog for at least one sub-device (21, 22, 23) of the plant, wherein each sub-device has a data storage with documentation about the sub-device (21, 22, 23), wherein the computing system (3) is configured for accessing the documentation of the sub-devices (21, 22, 23) of the plant (2) via the interface (24), searching for information in the documentation and outputting via the interface to the further computing systems (4, 5, 6, 18) and / or sensor devices (11, 12, 13) and / or human-machine interfaces (17) and / or applications (31, 32, 33, 34, 35, 36, 37, 38), and / or wherein the computing system (3) is configured for determining at least one information about the plant (2) by means of a maintenance / operation program, outputting instructions via the interface, documenting the performed steps and / or intervening in the control of the plant (2), wherein the maintenance / operation program takes into account stored knowledge and / or learned knowledge, wherein the maintenance / operation program performs an evaluation using data analysis methods.

7. The device according to claim 6, wherein the maintenance / operation program takes into account learned knowledge using artificial intelligence.

8. The apparatus according to claim 1 or 2, wherein the data model has a first data model which is set for a plurality of further computing systems (4, 5, 6, 18) and / or sensor apparatuses (11, 12, 13) and / or human-machine interfaces (17) and / or further applications (31, 32, 33, 34, 35, 36, 37, 38), wherein the computing unit (3) is configured to determine, from the query and by means of the first data model, at least one further computing system (4, 5, 6, 18) and / or at least one sensor apparatus (11, 12, 13) and / or at least one human-machine interface (17) and / or at least one further application (31, 32, 33, 34, 35, 36, 37, 38) which can provide the information searched with the query, wherein the data model has at least one second data model for a further computing system (4, 5, 6, 18) or a sensor apparatus (11, 12, 13) or a human-machine interface (17) or a further application (31, 32, 33, 34, 35, 36, 37, 38), wherein the computing unit (3) is configured to determine, from the second data model, which application (31, 32, 33, 34, 35, 36, 37, 38) of the determined further computing system (4, 5, 6, 18) and / or the determined sensor apparatus (11, 12, 13) and / or the determined human-machine interface (17) and / or which function of the determined further application (31, 32, 33, 34, 35, 36, 37, 38) can determine the information searched with the query, wherein the computing unit (3) is configured to hand over the query to the determined function or the determined application (31, 32, 33, 34, 35, 36, 37, 38) after determining the further application (31, 32, 33, 34, 35, 36, 37, 38) or the function of the further application (31, 32, 33, 34, 35, 36, 37, 38) which can determine the searched information, and subsequently to obtain the information from the determined function or the determined application (31, 32, 33, 34, 35, 36, 37, 38).

9. The apparatus according to claim 1 or 2, wherein the data model is configured by means of artificial intelligence, wherein the artificial intelligence is configured to determine further computing systems (4, 5, 6, 18) and / or sensor apparatuses (11, 12, 13) and / or human-machine interfaces (17) and / or applications (31, 32, 33, 34, 35, 36, 37, 38) which can provide the information searched with the query.

10. The apparatus according to claim 8, wherein the first and / or the second data model is structured by means of artificial intelligence, wherein the artificial intelligence is structured for determining further computing systems (4, 5, 6, 18) and / or sensor devices (11, 12, 13) and / or human-machine interfaces (17) and / or application programs (31, 32, 33, 34, 35, 36, 37, 38) which are capable of providing information searched with the query.

11. The apparatus according to claim 1 or 2, wherein the computing system (3) has a helper program (30), wherein at least one further computing system (4, 5, 6, 18) and / or sensor device (11, 12, 13) and / or human-machine interface (17) has at least one application program, wherein the helper program (30) is structured for receiving a query for information via the interface (24, 25), wherein the helper program (30) is structured for searching at least one application program capable of providing the searched information according to the data model according to the information searched by the query, wherein the helper program (30) is structured for forwarding the query to the application program capable of providing the information, wherein the helper program (30) receives the searched information from the application program, wherein the helper program (30) is structured for outputting the information received from the application program to the further computing system (4, 5, 6, 18) inquiring and / or to the sensor device (11, 12, 13) inquiring and / or to the human-machine interface (17) inquiring and / or to the application program inquiring via the interface (24, 25).

12. A method for controlling and / or monitoring technical devices for carrying out metal production and / or metal processing by means of an apparatus according to any one of claims 1 to 11 with a helper program having at least one interface for connecting with application programs, wherein the helper program obtains a query for information about the devices from at least one inquiring application program via the interface, wherein the helper program has access to a data model, wherein the data model gives a hint which information at least one further application program is capable of providing, wherein the helper program determines which application program is capable of providing the queried information according to the query and according to the data model, wherein the helper program transmits the query to at least one determined application program capable of providing the queried information, wherein the helper program receives a response from the determined application program, and wherein the helper program outputs the received response to the inquiring application program.

13. The method according to claim 12, wherein the data model has prescribed examples for queries with which the prescribed information can be found in further application programs, wherein it is determined which at least one further application program is capable of providing the searched information by comparing the prescribed examples with the query and / or the searched information of the query.

14. The method according to claim 12 or 13, wherein the data model has a first data model set for a plurality of other applications, wherein the assistant first determines the one or more other applications capable of providing information searched with the query according to the query and by means of the first data model, wherein the data model has a second data model for each other application, wherein the assistant determines a function of the determined other application capable of providing the searched information according to at least one second data model of at least one other application determined in the first step, wherein the assistant then hands over the query to the determined function of the determined other application, and wherein the assistant then receives the information from the determined function of the determined other application.

15. The method according to claim 12 or 13, wherein the data model is constructed as an artificial intelligence, wherein the artificial intelligence is constructed for determining the one or more other applications and / or functions of the other applications capable of providing the searched information.

16. The method according to claim 14, wherein the first and / or the second data model is constructed as an artificial intelligence, wherein the artificial intelligence is constructed for determining the one or more other applications and / or functions of the other applications capable of providing the searched information.

17. The method according to claim 12 or 13, wherein the assistant receives a further query for further information about the device from the queried application in response to a query for information, wherein the assistant determines which application is capable of responding to the further query according to the further query by means of a data model, wherein the assistant acquires the further information from at least one determined application, wherein the assistant outputs the acquired further information to the querying application.

Images