Computer-implemented method for processing machine-related data of a component of a device

Abstract

The invention relates to a computer-implemented method for processing machine-related data in order to obtain at least one trained machine-learning model and to a computer-implemented method for detecting the operating state of a component of a device. The invention also relates to a device for processing data, comprising means which are designed to carry out a respective method of the computer-implemented methods or both computer-implemented methods, and to a data structure.

Description

[0001] Description

[0002] Title of the invention

[0003] Computer-implemented method for processing machine-related data of a component of a device

[0004] field of technology

[0005] The present invention relates to a computer-implemented method for processing machine-related data to obtain at least one trained machine learning model, as well as a computer-implemented method for detecting an operating state of a component of a device. The present invention also relates to a data processing device with means configured to execute one or both of the computer-implemented methods. Furthermore, the invention relates to a data structure.

[0006] State of the art

[0007] Computer-implemented methods for detecting operating states in a device are known from the prior art. For this purpose, data recorded during the operation of the device, for example using sensors, can be evaluated. By recognizing patterns in the data that were previously associated with a specific operating state of the device, a statement can be made about its current operating state. Even if such an approach has led to useful results in the past, it is nevertheless desirable to further improve the reliability and efficiency with which operating states in a device can be detected and to further reduce the effort required to implement the measures necessary for detection.

[0008] Summary of the invention

[0009] It is therefore an object of the present invention to overcome the described disadvantages of the prior art and in particular to provide means by which operating states in a device can be detected more reliably, efficiently and with less effort.

[0010] The object is achieved by the invention according to a first aspect in that a computer-implemented method for processing machine-related data, which in particular represent information about state variables and events assigned or assignable to device components or can be used to determine them, for obtaining at least one trained machine learning model (ML model), the computer-implemented method comprising: (i) components of at least two devices are selected as specific components and each is assigned to at least one of at least one defining component group and / or (ii) for each device of at least two devices, at least one component of the respective device is selected as a specific component and is assigned to at least one and / or exactly one component group of at least one defining component group in such a way thatso that after the assignment of all specific components of all devices (a) the specific components assigned to one and the same component group are all of the same type or identical and / or (b) the specific components of the same type or identical among all specific components of all devices are assigned to the same component group, that data is received which, with respect to each of the specific components, represents and / or makes ascertainable information about (i) one or more state variables assigned or ascribable to the respective specific component and (ii) represents and / or makes ascertainable one or more events relating to the respective specific component; and that for each of the at least one component group, a separate ML model is created on a separate basis, each at least partially formed on the basis of at least parts of the received data,database is trained, is suggested.,

[0011] The invention is therefore based on the surprising finding that with cross-device information on corresponding device components, a database can be formed on which an ML model can be particularly advantageously trained to recognize operating states of components of the respective component group.

[0012] In other words, multiple devices are categorized into their components, and component-specific information is collected for similar or identical components in groups. This advantageously allows for obtaining training data for training an ML model focused on a specific component type and for training the correspondingly specialized ML model.

[0013] With the proposed computer-implemented method, the database can also be based on data on corresponding components, especially from different devices. This inherently also contains information on a multitude of conditions from different application scenarios for the respective component type. This enables particularly high reliability in the detection of operating states when the ML model trained on the database is subsequently used. This is sometimes even possible for previously unknown operating states.

[0014] Therefore, an ML model trained in this way can subsequently be used very advantageously to detect the operating state of a corresponding, i.e., similar or identical, component in a device with great accuracy and reliability. Data on this component does not necessarily have to have been part of the database on which the ML model was trained, although it can be advantageous.

[0015] In contrast to conventional computer-implemented methods, the proposed computer-implemented method no longer considers the device as a whole and no longer only includes data relating to the entire device. Instead, the similar or identical components of multiple devices are also or exclusively considered, and a database is created for each of these, which in turn forms the basis for training an ML model focused on the corresponding component group.

[0016] The data with information on state variables can then be used at least partially as input data during training and / or the data with information on events can then be used at least partially as truth data during training.

[0017] The events can represent operating states and, for example, have been manually marked with reference to state variables. Using the individual ML models, operating states for the individual components can then be recognized based on input data for the corresponding state variables. Based on this, the operating state of the device can also be recognized as a result.

[0018] An advantageous approach against the background of the proposed computer-implemented method can therefore be as follows:

[0019] Several devices are selected.

[0020] Each of these devices is (by definition) divided into several components. For each device, several of these components are selected (which are then referred to as the specific components).

[0021] All specific components of all devices are then grouped (by definition) in such a way that the similar or identical components among the specific components are always assigned to a common component group (X). For each component within a component group (X), the data of the individual components (i.e., data representing information on previously defined state variables as well as on detected events of the individual components) are combined into a database, so that each component group (X) has its own database. Each database can, for example, contain input data and truth data with which an ML model can be trained.

[0022] On each database obtained in this way, an individual ML model is trained for each component group (X).

[0023] A model trained in this way on a database for a component group (X) can then be used to recognize operating states for components that are similar or identical to the components of the component group (X).

[0024] It goes without saying that the assignment of a specific component to a component group is a purely definitional process. Likewise, a component group itself is a purely definitional measure. Group assignment (i.e., the assignment of a specific component to a component group) is carried out for the purpose of appropriately merging the information of the individual components, in particular, merging the information of the individual components into a database according to their group affiliation. Therefore, the components of a component group are generally not actually (i.e., physically) grouped within the framework of the proposed computer-implemented method, and they do not need to be.

[0025] In the proposed computer-implemented method, each specific component is advantageously assigned to a component group. Even if several specific components have been selected for a device, the assignment to a component group is advantageously carried out individually for each specific component. Thus, if N specific components have been selected for a particular device, these N specific components can be assigned to 1 to N different component groups. For example, a device can have several similar or identical components, so that two or more than two specific components of this device are assigned to one and the same component group.

[0026] At least the information on some of the state variables and / or events can, for example, be represented and / or determined by partial data of the received data, preferably exclusively.

[0027] Each database on which a separate ML model is trained for a component group can, for example, be formed from data that represents information on the state variables and / or events of those components or that makes it possible to determine which components are assigned to the respective component group.

[0028] For example, the received data may be received from a remote computing device, from a data network, from a storage medium, via a data connection, and / or from a sensor.

[0029] The processed machine-related data therefore represent, at least in part, information on state variables and events assigned or assignable to device components or can be used to determine them.

[0030] In this application, the term "ML model" is used as an abbreviation for "(data) model of machine learning." Such an ML model can, for example, be a convolutional neural network (CNN), particularly from the field of deep learning.

[0031] For the purposes of the present application, two components are preferably of the same type if (i) they each perform the same function in the respective devices, (ii) they are each used for corresponding purposes in the respective devices and / or (iii) they are of identical construction.

[0032] Alternatively or additionally, it can also be provided that for each device of at least two devices, at least one component of the respective device is selected as a specific component and is assigned to at least one and / or exactly one component group of at least one definitional component group in such a way that after all specific components of all devices have been assigned (i) the specific components assigned to one and the same component group are all of the same type or identical and / or (ii) the specific components of the same type or identical type among all specific components of all devices are assigned to the same component group.

[0033] For each device, at least one component is selected as a specific component. Each specific component is assigned to at least one component group. After all specific components have been assigned, the constraints (i) and / or (ii) must be met. By organizing the assignments accordingly, a particularly reliable database can be obtained on which an ML model can be trained that is tailored to components of the respective component group.

[0034] Alternatively or additionally, it can also be provided that at least two, preferably all, of the at least two devices are of different types.

[0035] This has proven to be particularly advantageous because it allows information from different application scenarios resulting from the different device types to be incorporated into the database.

[0036] For example, a component X can be provided in a corresponding manner in two devices (U, V) that are completely different in terms of their intended use, i.e. a component X in device U, and a component X in device V. The data for component X in device U and the data for component X in device V then form a common database (if necessary with further data for component X or similar or identical components). An ML model trained on this database can also be used to recognize states of component X (or a similar component) provided in a device W (which may be different from devices U and V). The proposed computer-implemented method naturally also allows components to be provided in consistently identical or at least partially identical devices.

[0037] Alternatively or additionally, it can also be provided that of the at least two devices, at least two are selected differently from the group of devices comprising: dosing device, screening device, vibration device, mill, extruder, conveying device, weighing device, mixing device and / or test bench.

[0038] These different types of devices are particularly advantageous.

[0039] In one embodiment, all of the at least two devices can be used in the field of bulk material handling, bulk material processing and / or bulk material processing.

[0040] Alternatively or additionally, it can also be provided that each selected specific component is assigned to at least one and / or exactly one component group of at least two definitional component groups.

[0041] The definitional component groups are advantageously all different.

[0042] This also creates at least two databases and trains at least two ML models (one per database).

[0043] Alternatively or additionally, it can also be provided that at least one component group is selected from the group of component groups comprising: wheel, axle, bogie, carriage, gearbox, bearing (machine element), guide element, motor, discharge element, drive system, conveyor belt, agitator, weighing unit and / or unbalance drive.

[0044] A discharge device can be, for example, a screw, a spiral and / or a slide.

[0045] A weighing unit can be, for example, a measuring eye, a weighing disk and / or a weighing beam.

[0046] Preferably, at least two or all definitional component groups are selected from said group of component groups.

[0047] Alternatively or additionally, it can also be provided that for each device of the at least two devices, two or more than two specific components are selected and / or for each device of the at least two devices, the respectively selected specific components are assigned to different component groups.

[0048] By selecting multiple components per device, the resulting ML models can be used to perform particularly comprehensive monitoring of device components.

[0049] By assigning the components to different component groups, it is more reliably ensured that the databases of the individual component groups are formed from information from as many devices as possible.

[0050] Alternatively or additionally, it may also be provided that the specific components are mechanical components, electrical components and / or components of the devices that are subject to wear.

[0051] Preferably, this applies to all specific components. Alternatively or additionally, it can also be provided that each of the at least one component group is assigned at least one and / or exactly one specific component of each of the at least two devices.

[0052] This makes it possible to more reliably ensure that the databases of the respective component groups are formed from information from as many devices as possible.

[0053] Alternatively or additionally, it can also be provided that each specific component is assigned to exactly one component group or at least one of the specific components is assigned to several component groups.

[0054] Alternatively or additionally, it can also be provided that the received data represent or make it possible to determine information on state variables and / or events, in particular error events and / or good events, with regard to the specific components, wherein the received data preferably has individual information for at least one and / or each specific component.

[0055] Advantageously, the individual information represents information on component-specific state variables and / or events.

[0056] The received data can advantageously comprise multiple data subsets, and the information on each state variable of each component as well as the information on each event of each component are contained in separate data subsets. In one embodiment, the data subsets can at least partially overlap and / or the data subsets can be at least partially combined for each component (for example, into data representing information of the respective component and / or into data representing events of the respective component).

[0057] Alternatively or additionally, it can also be provided that the received data are at least partially raw data from sensors or data derived therefrom, and preferably at least some of the sensors are arranged on the specific components, and / or the received data originate at least partially from a PLC system.

[0058] Sensor data can be recorded efficiently using appropriate sensors and is therefore particularly suitable for creating a database.

[0059] In one embodiment, the received data is at least partially preprocessed sensor data, which preferably originate at least partially from sensors arranged on the respective specific components.

[0060] Alternatively or additionally, it can also be provided that the respective database for training the respective ML model of a component group contains at least the information on state variables and / or events in relation to the specific components assigned to the respective component group, in particular explicitly and / or implicitly.

[0061] Thus, each database (of a component group) is formed by at least those data that at least partially represent or make determinable information about the state variables and / or events of the specific components assigned to the respective component group. This makes it particularly advantageous to obtain a group-specific database and thus perform targeted training of an ML model for each component group.

[0062] Alternatively or additionally, it can also be provided that the information on each state variable is (i) represented by time series data, (ii) has a time-dependent course (iii) is represented or can be determined by at least part of the received data and / or (iv) is obtained by processing at least part of the received data and / or that (i) each state variable is assigned to exactly one and / or at least one specific component, (ii) several specific components are assigned corresponding state variables and / or (iii) within a component group, each specific component is assigned the same state variables.

[0063] For example, the received data is time series data (e.g. time-indexed sensor data).

[0064] In one embodiment, identical state variables are considered for the specific components assigned to a common component group.

[0065] In one embodiment, for at least one, preferably for each, specific component, the state variables are at least partially selected from the group of state variables comprising: direction of movement (for example in connection with the data of a vibration sensor), speed (for example in connection with the data of a speed sensor), torque (for example in connection with the data of a torque sensor), electrical voltage, such as an input voltage (for example in connection with the data of a voltage sensor) and / or electrical current (for example in connection with the data of a current sensor).

[0066] Alternatively or additionally, it can also be provided that the state variables of at least one and / or all specific components are selected, in particular manually and / or with a feature selection algorithm, from a selection of state variables, wherein the selection of the state variables for the specific components of a component group is preferably carried out jointly.

[0067] For example, data can be continuously provided and / or received for different state variables; however, only the data from state variables selected as described above are then actually used to form a database.

[0068] In one embodiment, the selection of state variables for the specific components is reviewed periodically and adjusted based on the review results. This allows for new state variables to be considered that become more meaningful over time.

[0069] Alternatively or additionally, it can also be provided that each event of a specific component is an assignment of a good or bad marking to an event time or an event period in relation to the course of one or more state variables, in particular assigned to the respective specific component, in particular by assigning a state category to the respective time or period.

[0070] This makes it advantageous to directly determine the value (at the time) or the course (during the period) of one or more state variables relevant to the event (or in this case the data underlying this state variable) based on an event (via the assignment linked to it).

[0071] Alternatively or additionally, it can also be provided that the processing of the machine-related data is carried out to obtain at least one trained ML model and / or that at least one trained ML model is obtained as a result of the processing of the data, in particular at least the ML model that is calculated to detect an operating state of a specific component of a device in a computer-implemented method according to the second aspect of the invention. The proposed computer-implemented method for processing machine-related data is thus advantageously a computer-implemented method for obtaining at least one trained ML model.

[0072] Alternatively or additionally, it can also be provided that the individual ML model trained with respect to a specific component group can be used to detect operating states, in particular error states, with respect to a component used in a device which is of the same type or identical to the specific components which are assigned or assignable to the specific component group, by calculating the respective ML model on a database which is obtained at least partially from data which represent and / or make ascertainable information on one or more state variables assigned or assignable to the component used, which correspond to the state variables considered during the training of the respective ML model.

[0073] The trained ML model can therefore be calculated using data from live operation of a device or its components, which data corresponds to that from the training, and thus recognize operating states of components of the device.

[0074] The object is achieved by the invention according to a second aspect in that a computer-implemented method for detecting an operating state, in particular a fault state, in a specific component of a device, the computer-implemented method comprising receiving data that represent and / or make ascertainable information about one or more state variables assigned or assignable to the specific component, and that an ML model, which was trained during a computer-implemented method according to the first aspect of the invention preceding the detection, is calculated at least partially with the received data as input data, wherein preferably the specific component is assignable to that component group,for which the ML model was trained in the preceding computer-implemented method and / or the database used in the preceding training of the ML model was at least partially formed with data that represents and / or makes it possible to determine information on state variables that correspond to the one or more state variables assigned or assignable to the specific component.

[0075] By means of an ML model trained according to a computer-implemented method according to the first aspect of the invention, the operating state of a device component, and thus advantageously also of a device as a whole, can be recognized by supplying the ML model with data corresponding to the data used during training (i.e., the model is calculated on this data).

[0076] If, during the training of the ML model, data concerning information on the state variables X and Y have at least partially formed the database, then the current operating states of the respective device component can be recognized by means of the trained ML model on the basis of current data concerning information on precisely these state variables X and Y for a device component to be monitored.

[0077] Due to the specially designed prior training, the trained data model is particularly well suited to reliably and precisely detecting the operating states of device components.

[0078] In this respect, all the advantages explained with reference to the computer-implemented method according to the first aspect of the invention also apply accordingly to the computer-implemented method according to the second aspect of the invention. Reference can therefore be made to the previous explanations at this point.

[0079] Alternatively or additionally, it can also be provided that the ML model used to calculate the trained ML model is that ML model of the ML models trained within the scope of the method according to the first aspect of the invention which was trained on data from specific components which are assigned or assignable to specific components of the component group to which the specific component is also assigned or assignable.

[0080] By selecting the ML model so that it has been trained on data from components that belong to the same component group as the specific component, a particularly reliable detection of the operating state of the specific component can be achieved.

[0081] Alternatively or additionally, it can also be provided that an operating state of the specific component is detected at least partially based on a result of the calculation of the trained ML model and / or that a control signal is generated in response to the detection of the operating state of the specific component and / or as a function of the detected operating state of the specific component.

[0082] The calculation makes it particularly reliable to detect the operating state of the specific component. For example, the control signal can be used to communicate information regarding the detected operating state of the specific component to an entity, which may also be different from the specific component and / or from the device comprising the specific component. This can cause the entity to perform an action in response to the control signal. For example, the control signal can represent the detected operating state of the specific component or make it determinable.

[0083] In one embodiment, the control signal is generated by a control signal generation unit. The control signal generation unit can be implemented in software, hardware, or a combination of both.

[0084] Alternatively or additionally, it can also be provided that the specific component and / or the device having the specific component is influenced by the control signal, wherein the influencing preferably comprises adapting a configuration of the component and / or the device, adapting an energy consumption of the component and / or the device and / or switching off the specific component and / or the device.

[0085] For example, by influencing, a fault condition of the specific component and / or the higher-level device can be avoided (e.g. before it occurs) and / or eliminated (e.g. if it has already occurred).

[0086] Resource-efficient operation of the specific component and / or the higher-level device can also be advantageously enabled or improved by influencing. For example, in an operating state representing inactivity of the specific component, another component of the same component and / or another device can be activated and / or yet another component of the same component and / or another device can be deactivated.

[0087] Adjusting a configuration may, for example, comprise adjusting one or more parameters of the specific component and / or the higher-level device. In this way, for example, a mode of operation of the specific component and / or the higher-level device can be achieved, such as avoiding and / or eliminating a fault condition as described above. Adjusting energy consumption may, for example, comprise activating and / or deactivating functions of the specific component and / or the higher-level device.

[0088] Alternatively or additionally, it can also be provided that the received data, which represent and / or make ascertainable information about one or more state variables assigned or assignable to the specific component, are at least partially raw data from sensors or data derived therefrom, and preferably at least some of the sensors are arranged on the specific component(s), and / or the received data originate at least partially from a PLC system.

[0089] Preferably, the data used for training the ML model (i.e. during the training process) and the data used for calculating the trained ML model (i.e. when later recognizing the operating state) originate at least partially from identical or the same sensors or are based on at least partially identical or the same sensor data.

[0090] The object is achieved by the invention according to a third aspect in that a device for data processing with means designed to carry out a computer-implemented method according to the first aspect of the invention and / or according to the second aspect of the invention is proposed.

[0091] All advantages explained with respect to the computer-implemented method according to the first and / or second aspects of the invention apply accordingly to the data processing device according to the third aspect of the invention. Reference can therefore be made to the previous explanations at this point.

[0092] The data processing device may, for example, comprise means for receiving the data received in the computer-implemented methods. The data processing device may also comprise a memory for at least temporarily storing the received data and / or the ML model.

[0093] The object is achieved by the invention according to a fourth aspect in that a data structure in which component objects and assignments are stored is proposed, wherein the assignments can be used to place the component objects in a relationship to one another and to display them in the form of a tree structure, wherein the component objects describe device components and a tree structure of the associated component objects can be displayed for the components of at least two devices, wherein at least two of the at least two tree structures have at least one component object identically.

[0094] The proposed data structure makes it particularly easy to manage the components of multiple devices, particularly within a data processing system, while representing the similar or identical components across devices using a single object. Furthermore, this allows the data relating to state variables and / or events of the respective similar or identical components to be efficiently assigned to the single component object. This allows subsequent training of an ML model intended for similar components on this data to be carried out particularly effectively and simply, particularly using a computer-implemented method according to the fifth aspect of the invention, described in more detail below.

[0095] If, for example, several devices (such as dosing devices) use a similar or identical component X (e.g., a discharge device), only a single object needs to be provided for this component in the data structure. Based on the assignments, the multiple devices that have this component X, the component X (or the associated component object) can then be placed in relation to the other components of the respective device in the tree structure of each device (which in turn allows the tree structure to be formed). The data that arises in connection with the individual components in the individual devices with regard to state variables and / or events can then be assigned to the respective (single) component object as state variable and / or event objects. This particularly simplifies subsequent training, since the data structure contains all or at least much of the information for training.

[0096] The data structure can be used particularly advantageously for training an ML model because the data (of the objects) within the data structure represent data that can fulfill a control function in a data processing device during the training of a data model. In this respect, the object data of the data structure preferably represent "instructions" that at least partially specify to a computer device used for training how the ML model is configured in the trained state. Thus, the object data represent "instructions" for a transformation process from an untrained ML model to a trained ML model.

[0097] The data structure can advantageously be implemented and / or stored in a database.

[0098] The data structure can advantageously be stored or storable on a data carrier and / or transmitted as a signal sequence over a data line. This allows the data structure to be easily deployed at different locations.

[0099] Alternatively or additionally, it can also be provided that state variable objects are stored in the data structure, each with at least one assignment to at least one component object.

[0100] This makes it particularly easy, for example, to make the associated state variables available for each component in the tree structure, especially to retrieve them. Several identical state variable objects can be provided, each assigned to different device components.

[0101] In one embodiment, assignments of several, in particular more than two, state variable objects to at least one component object are stored in the data structure of this component object.

[0102] Advantageously, a state variable object comprises data that describes the respective state variable or makes it determinable. The data may comprise features described in the computer-implemented method according to the first aspect of the invention with regard to state variable data, individually and in any combination.

[0103] Alternatively or additionally, it can also be provided that event objects are stored in the data structure, each with assignments to at least one component object and to at least one state variable object.

[0104] This makes it particularly easy, for example, to make the associated events and state variables available for each component in the tree structure, and in particular to retrieve them. Multiple event objects can be provided that are assigned to the same device components.

[0105] In one embodiment, a reference to an event time and / or event period is stored in the data structure for each event object in the respective state variable object. This allows the event to be directly linked to a specific course of the state variable. This allows events to be associated with anomalies in the course of the state variable. The event object and the state variable object are advantageously assigned to the same device component (and, by means of appropriate assignments, to the corresponding component object).

[0106] In one embodiment, assignments of several, in particular more than two, event objects to this component object as well as the associated state variable objects are stored in the data structure for at least one component object.

[0107] Advantageously, an event object comprises data that describes the respective event or makes it determinable. The data may comprise features described in the computer-implemented method according to the first aspect of the invention with regard to event data, individually and in any combination.

[0108] The object is achieved by the invention according to a fifth aspect in that a computer-implemented method for training an ML model which serves for the detection of operating states, in particular error states, with respect to a specific device component, wherein during the training of the ML model the state variable objects and event objects assigned to a component object stored in the data structure according to the fourth aspect of the invention, which describes the specific device component, are used as training data, in particular as input data and truth data.

[0109] The data organized in the data structure can therefore be used particularly easily for training.

[0110] Optionally, the computer-implemented method may also include providing the data structure. This may, for example, include receiving the data structure via a data line and / or retrieving the data structure from a memory.

[0111] Furthermore, the features of the computer-implemented method according to the first aspect of the invention can advantageously also be provided individually and in any combination in the computer-implemented method according to the fifth aspect of the invention. This particularly applies to features that describe the origin of the object data and / or the relationship between the object data.

[0112] The object is achieved by the invention according to a sixth aspect in that a use of a data structure according to the fourth aspect of the invention for training an ML model is proposed.

[0113] For the reasons already mentioned above, the data structure is particularly advantageous for use in training an ML model. Training can then be carried out, for example, using a computer-implemented method according to the first and / or fifth aspects of the invention.

[0114] In this respect, all the advantages explained with regard to the data structure according to the fourth aspect of the invention also apply to the use according to the sixth aspect of the invention. Reference can therefore be made to the previous explanations at this point. Brief description of the drawings

[0115] Further features and advantages of the invention will become apparent from the following description, in which preferred embodiments of the invention are explained with reference to schematic drawings.

[0116] Showing:

[0117] Fig. 1 is a schematic, highly simplified representation of a production environment for the

[0118] cement production;

[0119] Fig. 2 is a flowchart of a computer-implemented method according to the first

[0120] aspect of the invention;

[0121] Fig. 3 is a flowchart of a computer-implemented method according to the second

[0122] aspect of the invention;

[0123] Fig. 4 shows a data processing device according to the third aspect of the invention; and

[0124] Fig. 5 shows a data structure according to the fourth aspect of the invention.

[0125] Description of the embodiments

[0126] The invention can be explained particularly clearly using the example of cement production.

[0127] Fig. 1 shows a schematic, highly simplified representation of a production environment 1 for cement production.

[0128] The individual raw materials required for cement production, such as gypsum, additives, granulated blast furnace slag, and fly ash, are provided by dosing devices 3a, 3b, 3c, and 3d. Dosing devices 3a, 3b, and 3c can each be implemented, for example, as a belt scale, while dosing device 3d is designed as a loss-in-weight feeder.

[0129] The raw materials dosed by the individual dosing devices 3a-3d according to a predeterminable recipe are mixed, if necessary after further intermediate steps and the addition of additional raw materials, with a mixing device 5. After mixing, the finished cement product is weighed using a weighing device 7 in the form of a road vehicle scale or a weighing device 9 in the form of a rail scale, depending on whether it is transported further by road or rail.

[0130] Connecting lines are drawn in Fig. 1 between the individual devices 3a-3d, 5, 7 and 9 to illustrate the described process flow.

[0131] Each of the dosing devices 3a-3d has an identical drive system 11a-11d (occasionally also referred to as a gearbox). In the case of the weighfeeders 3a, 3b, and 3c, the drive system 11a-11c can operate a conveyor belt 13a, 13b, and 13c of the respective weighfeeder 3a, 3b, and 3c. In the case of the loss-in-weight feeder 3d, the drive system 11d can operate a discharge device 15, such as a screw.

[0132] The mixing device 5 also has a drive system Ile identical to the drive systems Ila-Ild. The drive system Ile can be used to operate an agitator 17 of the mixing device 5 for mixing the individual raw materials. Finally, both the road scale 7 and the track scale 9 each have a weighing unit 19a and 19b. The weighing units 19a and 19b can advantageously be designed as weighing beams.

[0133] The devices 3a-3d, 5, 7 and 9 of the production environment 1 have numerous components, of which, for the sake of clarity, only some have been described with the components 11a-11e, 13a-13c, 15, 17, 19a and 19b and illustrated in Fig. 1.

[0134] During operation of devices 3a-3d, 5, 7, and 9, the operating states of the individual components 11a-11e, 13a-13c, 15, 17, 19a, and 19b are to be continuously monitored. This allows for a rapid response, for example, in the event of a fault related to one of the components, and the initiation of corrective measures. This allows downtimes to be avoided or at least reduced. Monitoring the wear of components, especially mechanical components such as a discharge device, is also feasible in this way.

[0135] To do this, machine-related data is first processed in such a way that a separate database is created for each group of components and a separate machine learning model (ML model) is trained on it.

[0136] The machine-related data originates, on the one hand, from sensors assigned to the individual components lla-lle, 13a-13c, 15, 17, 19a, and 19b, and represents the information on the state variables of the respective components lla-lle, 13a-13c, 15, 17, 19a, and 19b. The following table shows the selected state variables and the sensors used as data sources for the individual components lla-lle, 13a-13c, 15, 17, 19a, and 19b.

[0137] Accordingly, for example, a speed sensor is provided for each conveyor belt 13a-13c, which continuously provides the current speed of a support roller of the conveyor belt as a measured value. This measured value can be received and further processed.

[0138] The individual components 11a-11b, 13a-13c, 15, 17, 19a, and 19b can be grouped into the aforementioned component groups. In this case, for example, components 11a-11b can be assigned to a "drive system" component group, components 13a-13c to a "conveyor belt" component group, component 15 to a "discharge device" component group, component 17 to a "stirrer" component group, and components 19a and 19b to a "weighing unit" component group. In Fig. 1, the rectangles of the components assigned to a common component group are filled with the same pattern for quicker orientation.

[0139] If certain states (in particular, good states and error states) occur in the individual components during operation of devices 3a-3d, 5, 7, and 9, these states can be logged as events, for example, along with a timestamp of their occurrence. These events represent further machine-related data that is processed.

[0140] For each component in a component group, the time-indexed sensor data associated with these components and the logged events form the aforementioned database. For example, the database for the "Drive System" component group is formed by the data from the vibration sensors assigned to the drive systems lla-lle and the events logged for the drive systems lla-lle within a defined period (e.g., the last seven days).

[0141] Subsequently, an individual ML model can be trained on the database of each component group. The sensor data from the database serves as input data, and the event data from the database serves as truth data. An ML model trained in this way can recognize the operating states of components in the respective component group for which it was trained. To do this, the trained ML model calculates state variables (for example, the direction of movement in the case of a component in the "drive system" component group) based on input data in the form of current (sensor) data that correspond to the state variables considered during training.

[0142] It is particularly noteworthy that the trained ML model can also detect operating states of components for which no data was included in the training database, as long as the component can be assigned to the respective component group for which the ML model was trained. This makes the proposed training of an ML model with a computer-implemented method according to the first aspect of the invention, which is summarized again below using an example, so advantageous.

[0143] Fig. 2 shows a flowchart 100 of a computer-implemented method according to the first aspect of the invention.

[0144] In 101, the above-described components 11a-11e, 13a-13c, 15, 17, 19a and 19b of the devices 3a-3d, 5, 7 and 9 of the production environment 1 are selected as specific components and assigned to the component groups "drive system", "conveyor belt", "discharge device", "agitator" and "weighing unit".

[0145] In 103, for the components lla-lle of the first component group "drive system," the sensor data on the direction of movement (through a time series with information on direction and absolute acceleration value at different times) of the drive systems lla-lle and the logged events for the drive systems lla-lle are received. For example, the data originates from an observation period of the last 7 days.

[0146] At 105, an ML model is trained on the database formed by the sensor data and event data. The sensor data forms the input data of the ML model, and the event data represents truth data. At 103 and 105, data is also received in parallel for the other component groups ("conveyor belt," "discharge device," "agitator," and "weighing unit"), and an ML model is trained on a database formed from the respective data. This subsequently results in a trained ML model for each component group.

[0147] During operation of the devices 3a-3d, 5, 7 and 9, operating states of the individual components of the devices can be detected using the trained ML models in a computer-implemented method according to the second aspect of the invention.

[0148] Fig. 3 shows a flowchart 200 of a computer-implemented method according to the second aspect of the invention.

[0149] For example, the operating state of the drive system 11a of the dosing device 3a should be monitored and detected.

[0150] For this purpose, current sensor data regarding the direction of movement of the drive system 11a is received in 201. This can be the data from the vibration sensor, which is arranged on the drive system 11a for this purpose.

[0151] In 203, the ML model trained as described with reference to Fig. 2 is calculated on the received data as input data.

[0152] In 205, event data relating to the current operating state of the drive system 11a is obtained as a result of the ML model calculation. This event data can indicate a good or bad state of the component 11a.

[0153] Similarly, operating states of the remaining components 11b-11e, 13a-13c, 15, 17, 19a and 19b can also be detected using the previously trained ML models and the respective current sensor data as input data.

[0154] Fig. 4 shows a device 21 for data processing according to the third aspect of the invention. The device has means designed to execute a computer-implemented method according to the first and / or second aspect of the invention.

[0155] Fig. 5 shows a data structure 23 according to the fourth aspect of the invention.

[0156] In the data structure, information about the components can be stored in component objects, information about state variables in state variable objects, and information about events in event objects. For similar or identical components of multiple devices (such as the drive systems 11a-11 of devices 3a-3d and 5), only a single component object needs to be provided, to which the information about state variables and events is assigned via assignments.

[0157] With such an organized data structure, training for each component group can be carried out particularly easily, since each component group is represented by a single component object and all necessary training data in the data structure can be accessed via the assignments.

[0158] Finally, it should be pointed out again that the illustration in Fig. 1 only shows excerpts of the stages of cement production and the devices shown therein are only very simplified.

[0159] Advantageous embodiments of the individual aspects of the invention are presented in the following examples. Example 1. Computer-implemented method for processing machine-related data to obtain at least one trained machine learning (ML) model, the computer-implemented method comprising: selecting components of at least two devices as specific components and each assigning them to at least one of at least one defined component group; receiving data which, with respect to each of the specific components, represents and / or makes ascertainable information about (i) one or more state variables assigned or assignable to the respective specific component and (ii) represents and / or makes ascertainable one or more events relating to the respective specific component;and that for each of the at least one component group, a separate ML model is trained on a separate database, each of which is at least partially formed on the basis of at least parts of the received data;

[0160] Example 2. Computer-implemented method according to example 1, wherein for each device of at least two devices, at least one component of the respective device is selected as a specific component and assigned to at least one and / or exactly one component group of at least one definitional component group in such a way that after all specific components of all devices have been assigned (i) the specific components assigned to one and the same component group are all of the same type or identical and / or (ii) the specific components of the same type or identical among all specific components of all devices are assigned to the same component group.

[0161] Example 3. Computer-implemented method according to one of the preceding examples, wherein at least two, preferably all, of the at least two devices are of different types and / or wherein at least two of the at least two devices are differently selected from the group of devices comprising: dosing device, screening device, vibration device, mill, extruder, conveying device, weighing device, mixing device and / or test bench.

[0162] Example 4. Computer-implemented method according to any one of the preceding examples, wherein each selected specific component is assigned to at least one and / or exactly one component group of at least two definitional component groups.

[0163] Example 5. Computer-implemented method according to one of the preceding examples, wherein at least one component group is selected from the group of component groups comprising: wheel, axle, bogie, carriage, gear, bearing (machine element), guide element, motor, discharge element, drive system, conveyor belt, agitator, weighing unit and / or unbalance drive.

[0164] Example 6. The computer-implemented method according to any one of the preceding examples, wherein for each device of the at least two devices, two or more than two specific components are selected and / or for each device of the at least two devices, the respectively selected specific components are assigned to different component groups. Example 7. The computer-implemented method according to any one of the preceding examples, wherein the specific components are mechanical components, electrical components, and / or components of the devices subject to wear.

[0165] Example 8. Computer-implemented method according to one of the preceding examples, wherein the received data represent or make ascertainable information on state variables and / or events, in particular error events and / or good events, with respect to the specific components, wherein preferably the received data comprises individual information for at least one and / or each specific component.

[0166] Example 9. Computer-implemented method according to one of the preceding examples, wherein the respective database for training the respective ML model of a component group comprises at least the information on state variables and / or events with respect to the specific components assigned to the respective component group, in particular explicitly and / or implicitly.

[0167] Example 10. Computer-implemented method according to one of the preceding examples, wherein the information on each state variable is (i) represented by time series data, (ii) has a time-dependent course (iii) is represented or ascertainable by at least part of the received data and / or (iv) is obtained by processing at least part of the received data and / or wherein (i) each state variable is assigned to exactly one and / or at least one specific component, (ii) a plurality of specific components are assigned corresponding state variables and / or (iii) within a component group, each specific component is assigned the same state variables.

[0168] Example 11. Computer-implemented method according to one of the preceding examples, wherein each event of a specific component is an assignment of a good or bad label to an event time or an event period with respect to the course of one or more state variables, in particular assigned to the respective specific component, in particular by assigning a state category to the respective time or period.

[0169] Example 12. Computer-implemented method for detecting an operating state, in particular an error state, in a specific component of a device, the computer-implemented method comprising:

[0170] - that data representing and / or making it possible to determine information on one or more state variables assigned or assignable to the specific component are received and

[0171] - that a trained ML model is calculated with the received data as input data, such that the output data of the trained ML model represents the operating state of the device.

[0172] Example 13. Apparatus for data processing comprising means configured to carry out a computer-implemented method according to any one of Examples 1 to 11 and / or Example 12.

[0173] Example 14. Data structure in which component objects and assignments are stored, wherein by means of the assignments the component objects can be placed in a relationship to one another and represented in the form of a tree structure, wherein the component objects describe device components and a tree structure of the associated component objects can be represented for the components of at least two devices, wherein at least two of the at least two tree structures have at least one component object identical.

[0174] Example 15. Data structure according to example 14, wherein state variable objects are stored in the data structure, each with at least one assignment to at least one component object, and / or wherein event objects are stored in the data structure, each with assignments to at least one component object and to at least one state variable object.

[0175] The features disclosed in the foregoing description, in the drawings and in the claims may be essential to the invention in its various embodiments, both individually and in any combination.

[0176] List of reference symbols

[0177] 1 production environment

[0178] 3a, 3b, 3c, 3d dosing device

[0179] 5 Mixing device

[0180] 7 Weighing device (road vehicle scale)

[0181] 9 Weighing device (track scale)

[0182] 11a, 11b, 11c, lld, Ile drive system

[0183] 13a, 13b, 13c Conveyor belt

[0184] 15 discharge organ

[0185] 17 Agitator

[0186] 19a, 19b Weighing unit

[0187] 21 Data processing device

[0188] 23 Data structure

[0189] 100 Flowchart

[0190] 101 Selection of specific components and assignment of these to component groups

[0191] 103 Receiving data and creating a database

[0192] 105 Training an ML model on the database

[0193] 200 Flowchart

[0194] 201 Receiving data on state variables of a device component

[0195] 203 Calculating the trained ML model on the received data

[0196] 205 Obtaining result data on the current operating state of the device component

Claims

Patent claims 1. A computer-implemented method for detecting an operating state of a specific component of a device, the computer-implemented method comprising receiving data that represents and / or makes ascertainable information about one or more state variables assigned or assignable to the specific component, and calculating a trained machine learning model (ML model) at least partially with the received data as input data, wherein the trained ML model is obtained as a training method during a computer-implemented method preceding the detection for processing machine-related data to obtain at least one trained ML model, the training method comprisingthat (i) components of at least two devices are selected as specific components and each is assigned to at least one of at least one definitional component group and / or (ii) for each device of at least two devices, at least one component of the respective device is selected as a specific component and is assigned to at least one and / or exactly one component group of at least one definitional component group in such a way that after all specific components of all devices have been assigned (a) the specific components assigned to one and the same component group are all of the same type or identical and / or (b) the specific components of the same type or identical among all specific components of all devices are assigned to the same component group, that data is received,which, with respect to each of the specific components, represent and / or make ascertainable information on (i) one or more state variables assigned or ascertainable to the respective specific component and (ii) represent and / or make ascertainable one or more events relating to the respective specific component; and that for each of the at least one component group, a separate ML model is trained on a separate database, each at least partially formed from at least parts of the received data.

2. Computer-implemented method according to claim 1, wherein the specific component can be assigned to the component group for which the ML model was trained in the preceding computer-implemented method and / or the database used in the preceding training of the ML model was formed at least partially with data that represents and / or makes ascertainable information about state variables that correspond to the one or more state variables assigned or assignable to the specific component.

3. Computer-implemented method according to one of the preceding claims, wherein the ML model used to calculate the trained ML model is that ML model from the ML models trained in the training method which was trained on data from specific components which are assigned or assignable to specific components of the component group to which the specific component is also assigned or assignable.

4. Computer-implemented method according to one of the preceding claims, wherein an operating state of the specific component is detected at least partially based on a result of the calculation of the trained ML model and / or wherein a control signal is generated in response to the detection of the operating state of the specific component and / or as a function of the detected operating state of the specific component.

5. Computer-implemented method according to claim 4, wherein the control signal is used to influence the specific component and / or the device having the specific component, wherein the influencing preferably comprises adapting a configuration of the component and / or the device, adapting a power consumption of the component and / or the device and / or switching off the specific component and / or the device.

6. Computer-implemented method according to one of the preceding claims, wherein the received data, which represent and / or make ascertainable information on one or more state variables assigned or assignable to the specific component, are at least partially raw data from sensors or data derived therefrom, and preferably at least some of the sensors are arranged on the specific component or components, and / or the received data originate at least partially from a PLC system.

7. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that for each device of at least two devices, at least one component of the respective device is selected as a specific component and is assigned to at least one and / or exactly one component group of at least one definitional component group in such a way that after all specific components of all devices have been assigned (i) the specific components assigned to one and the same component group are all of the same type or identical and / or (ii) the specific components of the same type or identical type among all specific components of all devices are assigned to the same component group.

8. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that at least two, preferably all, of the at least two devices are of different types and / or wherein the training method comprises that of the at least two devices at least two are differently selected from the group of devices comprising: dosing device, screening device, vibration device, mill, extruder, conveying device, weighing device, mixing device and / or test bench.

9. Computer-implemented method according to one of the preceding claims, wherein the training method comprises assigning each selected specific component to at least one and / or exactly one component group of at least two definitional component groups.

10. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that at least one component group is selected from the group of component groups comprising: wheel, axle, bogie, carriage, gear, bearing (machine element), guide element, motor, discharge element, drive system, conveyor belt, agitator, weighing unit and / or unbalance drive.

11. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that for each device of the at least two devices, two or more than two specific components are selected and / or for each device of the at least two devices, the respectively selected specific components are assigned to different component groups.

11. A computer-implemented method according to any one of the preceding claims, wherein the training method comprises the specific components being mechanical components, electrical components and / or components of the devices subject to wear.

12. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that each of the at least one component group is assigned at least one and / or exactly one specific component of each of the at least two devices 13. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that each specific component is assigned to exactly one component group or at least one of the specific components is assigned to several component groups.

14. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that the received data represent or make ascertainable information on state variables and / or events, in particular error events and / or good events, with respect to the specific components, wherein preferably the received data comprises individual information for at least one and / or each specific component.

15. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that the received data is at least partially raw data from sensors or data derived therefrom, and preferably at least some of the sensors are arranged on the specific components, and / or the received data originates at least partially from a PLC system.

16. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that the respective database for training the respective ML model of a component group comprises at least the information on state variables and / or events with respect to the specific components assigned to the respective component group, in particular explicitly and / or implicitly.

17. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that the information on each state variable is in each case (i) represented by time series data, (ii) has a time-dependent course (iii) is represented or ascertainable by at least part of the received data and / or (iv) is obtained by processing at least part of the received data and / or wherein the training method comprises that (i) each state variable is assigned to exactly one and / or at least one specific component, (ii) a plurality of specific components are assigned corresponding state variables and / or (iii) within a component group, each specific component is assigned the same state variables.

18. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that the state variables of at least one and / or all specific components, in particular manually and / or with a feature selection Algorithm, from a selection of state variables, wherein the selection of the state variables for the specific components of a component group is preferably carried out jointly.

19. Computer-implemented method according to one of the preceding claims, wherein the training method comprises that each event of a specific component is an assignment of a good or bad label to an event time or an event period in relation to the course of one or more state variables, in particular assigned to the respective specific component, in particular by assigning a state category to the respective time or period.

20. Computer-implemented method according to one of the preceding claims, wherein the training method comprises processing the machine-related data to obtain at least one trained ML model and / or obtaining at least one trained ML model as a result of processing the data, in particular at least the ML model that is calculated to detect the operating state of the specific component.

21. Computer-implemented method for processing machine-related data, which in particular represent information on state variables and events assigned or assignable to device components or can be used to determine them, for obtaining at least one trained machine learning model (ML model), the computer-implemented method comprising: (i) components of at least two devices are selected as specific components and each is assigned to at least one of at least one defining component group and / or (ii) for each device of at least two devices, at least one component of the respective device is selected as a specific component and assigned to at least one and / or exactly one component group of at least one defining component group in such a way,so that after the assignment of all specific components of all devices (a) the specific components assigned to one and the same component group are all of the same type or identical and / or (b) the specific components of the same type or identical among all specific components of all devices are assigned to the same component group, that data is received which, with respect to each of the specific components, represents and / or makes ascertainable information about (i) one or more state variables assigned or ascribable to the respective specific component and (ii) represents and / or makes ascertainable one or more events relating to the respective specific component; and that for each of the at least one component group, a separate ML model is created on a separate basis, each at least partially formed on the basis of at least parts of the received data,database is trained., 22. A computer-implemented method for processing machine-related data to obtain at least one trained machine learning (ML) model, comprising computer-implemented methods, that components of at least two devices are selected as specific components and are each assigned to at least one of at least one definitional component group, that data is received which, with respect to each of the specific components, represents and / or makes ascertainable information about (i) one or more state variables assigned or ascribable to the respective specific component and (ii) represents and / or makes ascertainable one or more events relating to the respective specific component; and that for each of the at least one component group, a separate ML model is trained on a separate database, each at least partially formed on the basis of at least parts of the received data.

23. Computer-implemented method for detecting an operating state, in particular an error state, in a specific component of a device, comprising the computer-implemented method, - that data representing and / or making it possible to determine information on one or more state variables assigned or assignable to the specific component are received and - that a trained ML model is calculated with the received data as input data, such that the output data of the trained ML model represents the operating state of the device.

24. A data processing device comprising means adapted to carry out a computer-implemented method according to any one of claims 1 to 22, claim 23 and / or claim 24.

25. Data structure in which component objects and assignments are stored, wherein by means of the assignments the component objects can be placed in a relationship to one another and represented in the form of a tree structure, wherein the component objects describe device components and a tree structure of the associated component objects can be represented for the components of at least two devices, wherein at least two of the at least two tree structures have at least one component object identical.

26. Data structure according to claim 26, wherein state variable objects are stored in the data structure, each with at least one assignment to at least one component object and / or wherein event objects are stored in the data structure, each with assignments to at least one component object and to at least one state variable object.

Images