Method and apparatus for monitoring and / or controlling a machine by means of digital twinning
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- KRONES AG
- Filing Date
- 2024-12-03
- Publication Date
- 2026-08-04
Smart Images
Figure CN122514737A_ABST
Abstract
Description
[0001] This invention relates to a machine and a method and apparatus for monitoring and / or controlling the machine, particularly for use in a machine production line for processing food and / or beverages. The processing may include filling and packaging of food and / or beverages.
[0002] In modern industry, particularly in the filling and packaging sector, machinery is evolving into increasingly complex systems. With the integration of advanced technologies such as the Internet of Things (IoT), artificial intelligence (AI), and automated control systems, the performance and complexity of these devices are steadily increasing. This development has led to a dramatic increase in the costs associated with recording, monitoring, and describing the dynamic state of these devices.
[0003] To ensure a detailed and dynamic state description of as many machines and systems as possible within such equipment, data must be recorded, synchronized, and analyzed from various sources. This includes sensor measurements, operational data, condition monitoring, and many other system parameters. The cost of collecting this data increases with the number of components being monitored and the frequency of data collection. In highly complex systems, this can quickly lead to data overload, which cannot be effectively processed without specialized systems and algorithms.
[0004] Accurate and dynamic condition descriptions are crucial for several reasons, including early fault detection, operational optimization, and extending the lifespan of individual components or the entire machine. Continuous monitoring of equipment condition allows for the identification of anomalies and abnormal operating conditions at an early stage, enabling preventative maintenance and minimizing unplanned downtime. Furthermore, an accurate understanding of equipment condition allows for fine-tuning of operating parameters, thereby improving efficiency, energy savings, and product quality.
[0005] In this context, providing a state description in near real-time or with minimal delay is crucial, as it allows for immediate responses to changes. This is especially important in critical processes, where delays can lead to significant risks or losses.
[0006] The current method for providing data and information is through the use of digital twin technology. A digital twin is a virtual representation of a physical machine or device that is constantly updated using data from the real operating environment to simulate, analyze, and predict processes.
[0007] In practice, machines continuously record vast amounts of sensor data and provide it to digital twins. This requires a robust and reliable infrastructure. In existing technologies, the need to transmit these data volumes to the cloud or central data centers leads to various problems, such as bandwidth limitations, latency, security risks, and transmission costs. Time-critical applications are affected by the latency introduced during transmission, which can cause delays in response and decision-making.
[0008] Therefore, a solution is needed to overcome these shortcomings in existing technologies, thereby providing an effective dynamic state description, especially for digital twin-based filling and packaging machines, production lines, and equipment.
[0009] According to the present invention, this objective is achieved by the method of claim 1 and the edge device of claim 9. Embodiments and improvements are included in the dependent claims.
[0010] One embodiment of the invention relates to a method for monitoring and / or controlling machines, particularly machines in a production line for processing food and / or beverages. In this case, a code-based model for simulating the physical processes of the machine and / or a cross-machine code-based model for simulating the physical behavior of the machine production line are implemented on an edge device. The code-based model is part of a digital twin of the machine. Machine state data is input into the digital twin of the machine on the edge device, and the machine's production process is simulated in real time by means of the code-based model of the digital twin. The input state data are input parameters for the simulation, and results from the simulation are assigned to the input parameters at the data level. A temporal static or dynamic state description of the machine is provided based on the input parameters for the simulation and the results from the simulation, and may optionally be transmitted from the edge device to a server.
[0011] Another embodiment of the present invention relates to a computer device for monitoring and / or controlling machines.
[0012] Exemplary aspects of the invention are illustrated in the accompanying drawings. In the drawings: Figure 1 A schematic diagram is shown, illustrating an overview of the basic elements and structure of the invention. Figure 2 An exemplary flowchart of a method for starting a machine under operator guidance is shown; Figure 3 An exemplary device configuration for PET containers and adhesive bundles is shown; Figure 4 An exemplary equipment configuration for PET containers and shrink packaging machines is shown; Figure 5 An exemplary device configuration for jars or glass bottles is shown; and Figure 6 An exemplary device configuration for a tank is shown.
[0013] This invention aims to provide a time-based static or dynamic state description of filling and packaging machines, production lines, and equipment based on digital twins. The literature typically mentions three or four levels at which data logging, data processing, analysis and diagnostics, fault prediction, and intelligent decision-making are performed. These levels include, for example, machine controllers, edge devices, and servers.
[0014] A machine's digital twin is typically a virtual image of the physical equipment, providing data in real time. This makes it possible to closely monitor and analyze the machine's condition and predict future operations. Data is usually recorded by sensors installed on the machine. These sensors can record a variety of information, such as temperature, pressure, vibration, humidity, and many other data points.
[0015] Based on data, digital twins can utilize machine learning and / or artificial intelligence algorithms to draw conclusions from machine data analysis. Therefore, potential problems can be diagnosed and future errors predicted. For example, "predictive maintenance" can also predict when parts are likely to fail and proactively replace them before downtime.
[0016] As mentioned in the discussion of existing technologies, digital twins, due to their use of machine learning algorithms, require a very large amount of data to create a similarly detailed picture of the processes running on machines and transport aircraft. These drawbacks can be overcome by the concepts described here.
[0017] Figure 1 A schematic diagram is shown, illustrating an overview of the basic elements and structure of the invention. For example, a machine production line 100 for filling beverages includes one or more machines 101, 102, and 103. Machines 101-103 can be functionally connected to each other, for example, via conveyor belts 110 and 112 that ensure the flow of materials. However, the invention is not limited to machine production line 100 or multiple connected machines 101-103. The invention can also be applied to a single machine.
[0018] The machine production line 100 is connected to the edge device 120, and via the edge device 120 is connected to the cloud or server 130. According to one implementation, the edge device 120 may be physically integrated into the server 130, or mapped by a virtual instance on the server 130.
[0019] Edge device 120 includes one or more computers that perform data processing tasks at or near the data generation source, i.e., directly at the network edge (edge). It is an important component that brings computing power closer to the required location, such as machines 101-103 and / or conveyor belts 110, 112, i.e., closer to the machine production line 100 that generates the data.
[0020] Edge device 120 typically includes several core components essential to the tasks of the edge device, such as processors, memory, networking components, interfaces, operating systems, and software. The hardware, particularly one or more processors in edge device 120, is designed to provide high computing power, enabling computational operations on machine data to occur in real time. Depending on the requirements of specific applications, these processors can range from traditional CPUs to dedicated microcontrollers to advanced GPUs.
[0021] According to embodiments of the invention, a digital twin of machine equipment 100 or a portion thereof is at least partially implemented on edge device 120. This means that at least a portion of the digital twin of equipment 100, or one of machines 101-103, or conveyor belts 110, 112, or (partial) components of equipment 100, can be implemented on edge device 120. Another portion of the digital twin can be implemented, for example, on server 130.
[0022] The following description uses the term "digital twin of the machine," but "machine" refers to one or more of machines 101-103, or (partial) components of conveyor belts 110, 112, or machine 100.
[0023] The digital twin portion implemented on edge device 120 includes a code-based model for simulating the physical processes of the machine.
[0024] This physical model can depict the most different parts of container transport or accurately simulate the most different functions of the machines in device 100. A major advantage of the physical model is that it is code-generated. The code can operate on edge device 120 (i.e., on computing units near the machines).
[0025] The physical model can receive information about the products currently being filled / produced from the production line management system (e.g., from server 130). Additional optional data sources could be real-time video (where possible) to capture information and status on a portion of the production line, or other measurement signals transmitted to the digital twin via industrial communication methods. This information can be processed in the simulation model, and the results can be transmitted to the cloud 130 via an interface based on the input data. This offers the advantage that only the manageable portion of the data needs to be transmitted to the cloud 130 to provide monitoring and / or control of the machines.
[0026] For example, a model located in a digital twin can dynamically and in real time simulate a production process by continuously recording or recording images and videos of the machine’s controller, drive or its FU, installed sensors (e.g., machine pressure, position, temperature and / or environmental characteristics of the machine) and transport equipment and / or operator movement / intervention for specific events or at specific intervals.
[0027] According to one embodiment, the data may be collected by an edge device 120 or a computer system (data collection and processing system) located near the production line. According to another embodiment, the data may also include information from a central production line management and / or production planning system to obtain a direct relationship between machine and conveyor status and the currently filled and / or packaged products.
[0028] According to some implementations, data collected in this way can be preprocessed and processed so that it can be used as input parameters for parallel physics simulations. This can also occur on edge devices, such as... Figure 1 As shown.
[0029] For example, results from simulations can be explicitly assigned to the input parameters of the simulation model at the data level and carried over to the next working level of the digital twin. Thus, the model used in the simulation can act as a virtual or soft sensor capable of describing states in a manufacturing or process engineering process that would be impossible to represent in measurement techniques without simulation, or would only be possible with considerable additional overhead. Examples of this include the temperature inside a component, the temperature distribution in a fluid-filled tank, and / or the load and damage to a component due to improper use, as well as a shortened lifespan.
[0030] The advantage of this approach is that only a limited amount of data must be transferred from the active plane (level 2) of the edge device 120 to the active plane (level 3) on the server 130, but the data can always present a peripheral picture of causality from machine data and simulation results.
[0031] For example, digital twins can make it possible to simulate the occupancy of transport equipment from current machine status data and / or conveyor belt speeds—equipment that cannot be seen by cameras. As another example, the heating process of preforms (i.e., PET preforms) can be represented by machine power and furnace formulation, for example, for different material parameters. Simulation results can show the behavior inside the object being inspected, compared to information from images or sensors, thus allowing the mapping of temperature gradients across the wall thickness of preforms or the heating of container contents in a pasture to be performed based on machine data and a physical model of parallel simulated heating.
[0032] According to one implementation, simulation on a physical model can be much faster than the actual process requires. Therefore, a digital twin can be used as a pilot controller, or regulatory algorithms can be developed and operated based on a digital twin. The data connection from the data source (i.e., from device 100) to the edge device 120 can be ideally configured so that data, information, and / or software code or software artifacts or parameters can be exchanged simultaneously or sequentially in both directions. The edge device 120 can be tailored to its configuration regarding the number of processors, clock speed, main memory, GPU, and memory, enabling it to act as a calculator providing raw or processed data to the model and to perform simulations in a simulation area (separate from) the data processing area.
[0033] According to one implementation, the physical model may reside in a so-called container containing the necessary software infrastructure for communication, execution, evaluation, and providing simulation results from the model. However, in an alternative implementation, it is conceivable that several or all models reside in a single container. According to an alternative implementation, the model may be provided as a "functional prototype unit" (FMU) and may be executed on the edge device 120 using suitable software. The container / FMU may also communicate with other provided container / FMUs.
[0034] In the case described here as an example, different types of models reside on edge device 120. Therefore, data-based models, logical models, and physical models that allow for real-time physical simulations based on the provided data can be used. In this case, from a tuning perspective, real-time can be considered in the range of 1-5 ms. However, defining real-time in terms of actual processing time (milliseconds, seconds, minutes) may also be sufficient. Depending on the application, the model on edge device 120 is always fast enough to determine the necessary information in the simulation and report the resulting measurements to the real system or higher levels.
[0035] According to a preferred embodiment, the model implemented on edge device 120 can be used for active conditioning of the connected real system 100. In some cases, this may lead to the model located on edge device 120 being used as part of a model prediction method (MPC). In this case, simulation based on the model used can be much faster than the actual process in device 100.
[0036] Therefore, digital twins can, to some extent, look to the future, predict the ideal next step, and report it to the real system100. Besides serving as a model predictor regulator for timeframes typically on the order of milliseconds, the described approach can also represent longer future timeframes. Specifically, physical models are used here; unlike purely data-based models (i.e., so-called predictor and / or regression models), physical models do not require the complex generation of a data pool for prediction but can represent new situations through simulations with corresponding new boundary conditions.
[0037] These simulation results can also be combined with statistics on failures that have occurred, allowing for very detailed projections into the future. For example, this can be used to plan product replacement scenarios, keep operators on standby, and provide raw materials.
[0038] The implementation can also be used for condition monitoring and predictive maintenance and / or self-healing because the impact of failures on the process can be directly assessed by using a physical model. For example, if a heating radiator in a preform heating furnace fails, the impact on the preform heating process can be immediately determined based on a physical simulation, and wiring optimization for the remaining radiators can be performed on the model and written back to actual control.
[0039] In the field of "predictive maintenance," the use of physical models no longer relies on the generation of complex fault data. Instead, it allows for comparison of currently running simulation results with a database or relevant model that establishes a direct relationship between the load in the current cycle and the maximum tolerable load cycle. For example, for particularly stressed components, stress in MPa can be calculated, caused by acceleration from the actuator and the mass of the product to be transported. By comparing this stress online with data from the Waller chart of the corresponding material, the remaining service life can be determined with great precision.
[0040] According to the implementation, Level 2 (edge device 120) and Level 3 (server / cloud 130) are expanded to allow for simulations using physical models based on available data. Therefore, a more detailed view of events occurring on and within the machine production line 100 can be generated using the same amount of data. The potential ranges from model-based occupancy determination to simulating the heating behavior of filled products.
[0041] Figure 2An example flowchart of a method 200 for monitoring and / or controlling a machine is shown. The method begins at step S202, where a code-based model for simulating the physical processes of the machine is implemented on an edge device 120. As previously described, the code-based model is part of a digital twin of the machine. For example, the code-generated model can be created and parameterized in an automated, partially automated, or manual workflow in the cloud 130 and can be brought to the edge device 120 via a data cable.
[0042] In step 204, the machine's status data is recorded by and / or provided to the digital twin of the machine on the edge device 120. The status data may include sensor data, but may also include other data sources, such as video streams as described above.
[0043] In optional step S206, the state data may be preprocessed. Preprocessing of the state data may include filtering sensor data based on its relevance to the desired output of the code-based model. Additionally or alternatively, preprocessing may also include discarding incomplete datasets.
[0044] In step S208, the machine's production process is simulated in real time using a code-based model of a digital twin. The input state data are the input parameters used for the simulation, and the results from the simulation are assigned to the input parameters at the data level. Then, in step S210, a time-static or dynamic state description of the machine is provided based on the input parameters used for the simulation and the results from the simulation, and (optionally) in step S212, it is transmitted from the edge device 120 to the server 130.
[0045] In a further step, the state description can be processed and used as an input variable for control and / or regulation on an actual PLC and / or industrial PC.
[0046] In this scenario, the simulation of the physical process on the edge device 120 can run faster than the actual process in the machine. In this case, in step S214, the digital twin on the edge device 120 can be used as a pilot controller for the machine. The edge device 120 can generate control signals based on a time-dependent static or dynamic state description and send them to the machine's control unit.
[0047] In the following Figures 3 to 6 The present invention or at least some aspects thereof is described herein for various bottle filling equipment. Figures 3 to 6 The description is intended to provide only a general overview of machines that can collect state data, upon which LLMs can process user requests.
[0048] Figure 3An exemplary device configuration 1000 for PET bottles or PET containers and adhesive bundles is shown. Figure 3 As shown, equipment configuration 1000 includes different modules that form a line, at the end of which fully filled PET containers are distributed onto pallets in bundles. Some of the modules and machines may be optional, and the invention is not limited to the specific form and arrangement of the equipment configuration.
[0049] Equipment configuration 1000 includes an oven 1002 for preforms, a preform sorter 1004 with a feeder, and a blow molding machine 1008. Modules 1002, 1004, and 1008 typically form a stretch blow molding machine in which PET containers are made and shaped from initial materials. The produced PET containers are then conveyed to a filler 1010, where bottles are filled. The filler may optionally include a rinsing device. During storage or transportation, various particles, such as dust, cardboard, or wooden pallet residue, may accumulate in the preforms. These particles can be removed using the rinsing device. A sealing machine may be arranged at the end of the filler to seal the PET containers after filling.
[0050] Optionally, the equipment configuration 1000 may include a rotating device after the filling machine 1010 for hot filling of PET containers. The filled PET containers are conveyed to a separator 1020 and then to a drying unit 1024 via one or more conveyor belts 1016 (which may also include a buffer 1018 for intermediate loading of filled containers), where the PET containers are dried.
[0051] After drying, the PET containers are conveyed to labeling machine 1026. Labeling machine 1026 can be designed for various labeling techniques, such as hot glue, cold glue, self-adhesive labels, or sleeve labels. After printing or labeling, the PET containers are conveyed to handle applicator via second drying unit 1028, production line distributor 1030, conveyor belt 1032, adhesive bundle production device 1034, and curing path. In adhesive bundle production device 1034, PET containers are grouped together in specific bundle sizes and packaged into bundles, such as "six-packs". In handle applicator, handles are attached to the containers, making it possible to comfortably carry the bundles. The finished bundles are then correspondingly arranged into layers by robot 1042 and packed on pallets by palletizer 1044.
[0052] In equipment configuration 1000, so-called format carriages or format racks can be arranged at various modules and machines to provide quickly changeable format kits for short changeover times and automated tool switching. Examples of format carriages are format carriage 1006 for blow molding machine 1008, format carriage 1012 for filling machine 1010, format carriage 1022 for labeling machine 1026, format carriage 1038 for adhesive bundling production device 1034, and format carriage 1046 for palletizer 1044.
[0053] Figure 4 Another exemplary device configuration 1100 for PET containers and shrink packaging machines is shown. Figure 4 Equipment 1100 includes Figure 3 The equipment configuration includes many modules and machines of 1000, but some differences exist. Therefore, for Figure 4 The already combined part is omitted. Figure 3 The description of the module being described.
[0054] A significant difference between the two exemplary device configurations 1000 and 1100 is that the labeling machine 1126, with labeling module 1127, can be installed after the blow molding machine 1008 and before the filling machine 1008. For this purpose, device configuration 1100 may include up to six transport tracks 1150 into which PET containers can be extruded. After the PET containers have been correspondingly extruded into one of the six tracks 1150, they are conveyed to a film wrapping module 1152 and then to a shrink tunnel 1154.
[0055] Figure 5 An exemplary device configuration 1200 for jars or glass bottles is shown. Figure 5 The exemplary device configuration 1200 is again with Figure 3 and Figure 4 The device configurations 1000 and 1100 have some similarities, so the description of the device configuration is limited to the differences in device configuration.
[0056] like Figure 5 As shown, an exemplary device configuration may include two separate supplies. Figure 5 The first feed on the left shows a branch of the can, or optionally a branch of a new reusable bottle. Here, the containers (i.e., cans or new bottles) are guided into the machine by the depalletizer 1302, where they are guided to the filler 1010 via a conveyor belt. Figure 5 The second feed on the right shows a portion of a reusable bottle being introduced into the device from a reusable sorting device (not shown).
[0057] In cases where used reusable bottles are introduced into device 1200 via a sub-branch for reusable bottles, the reusable bottles first pass through a cleaning or washing machine 1304. Another possible difference in the exemplary device configuration 1200 is the addition of a transshipment packaging machine 1306 after the labeling machine 1026. The transshipment packaging machine can sort bottles or cans into cardboard clip application devices or boxes, or both.
[0058] Figure 6 An exemplary device configuration 1300 for cans is shown, wherein elements already described in other device configurations are not described again. In device configuration 1300, cans are introduced from a can magazine 1402 containing cans into a depalletizer 1302. After the cans have passed through a filling machine and been filled, the cans are sealed by means of a sealing magazine 1404, and the cans are further conveyed along device 1400 via a conveyor belt, as described above.
[0059] If not required, the optional pasteurizer 1408 can be bypassed via bypass 1412. In pasteurizer 1408, freshly filled products can be pasteurized for preservation.
[0060] Compared to device configurations 1000, 1100, and 1200, exemplary device configuration 1300 shows different tanks for corresponding consumables, such as tank 1410 having rinsing liquid and / or filling product, and tank 1406 having lubricant. These tanks may also be included in the exemplary device configurations already described above. For example, chemical product 106 conveyed from mixer 110 to the machine may be stored in tanks 1406 and 1410.
Claims
1. A method for monitoring and / or controlling machines (101-103), particularly machines in a machine production line (100) for processing food and / or beverages, wherein, The method includes: Implement (S202) a code-based model on the edge device (120) for simulating the physical processes of the machine or multiple machines in the machine production line or part or all of the machine production line, wherein the code-based model is part of the digital twin of the machine. Provide the machine's status data to the digital twin of the machine on the edge device (S204); The production process of the machine is simulated in real time (S208) using a code-based model based on a digital twin, wherein the input state data are the input parameters for the simulation, and the results from the simulation at the data level are assigned to the input parameters of the simulation; and Based on the input parameters used for simulation and the results from the simulation, a time static or dynamic state description is provided to the machine (S210).
2. The method according to claim 1, wherein, The simulation of the physical processes on the edge device (120) occurs faster than the actual processes, and the method further includes: Operation (S214) The digital twin on the edge device serves as a pilot controller for the machine or multiple machines or part or the entire machine production line, wherein the edge device generates control signals based on a time-static or dynamic state description and sends them to the control device of the machine.
3. The method according to claim 1 or 2, wherein, The code-based model is implemented in a container or as an FMU on the edge device (120), which includes the necessary software infrastructure for communicating with the model, performing, evaluating and / or providing results from the simulation. and / or The code generation model is created and parameterized in an automated or partially automated or manual workflow on the server (130) and transmitted to the edge device via a data line.
4. The method according to any one of claims 1 to 3, wherein, Simulation using code-based models further forms the basis for model prediction tuning of the machine, and / or, The simulation results are also combined with statistical data on the faults that have occurred in order to predict the future state of the machine and / or to input the simulation results as virtual sensor signals into process regulation.
5. The method according to any one of claims 1 to 4, wherein, The static or dynamic state descriptions over time form the basis for the state monitoring and / or predictive maintenance of the machine.
6. The method according to any one of claims 1 to 5, wherein, Also includes: Preprocessing (S206) status data; And / or, The static or dynamic state description of the time is sent from the edge device (S212) to the server (130). The preprocessing of the state data includes: Filter sensor data based on its relevance to the desired output of the code-based model, and / or Discard incomplete datasets, and / or stationarity analysis, and / or Determine the static parameters.
7. The method according to any one of claims 1 to 6, wherein, Status data includes data from the machine's controller, data from the drive or its inverter, data from installed sensors, particularly data from sensors for the machine's pressure, position, and temperature, data about the machine's environmental characteristics, data from images and / or video recordings from the machine and / or transport equipment, and / or data on operator movement and intervention, and / or data from a central management and / or production planning system.
8. The method according to any one of claims 1 to 7, wherein, The edge device (120) is a near-machine computing unit and also includes a data link from the data source to the edge device, which is adapted to transmit data, information and / or software code or software artifacts or parameters simultaneously or sequentially in both directions.
9. A computer device (120) for monitoring and / or controlling machines, particularly machines in a production line for processing food and / or beverages, wherein, The computer device includes: Memory, used to store computer code; A network interface for receiving and sending data; and Processor for executing computer code, wherein the computer device is adapted to In the memory, a code-based model is implemented (S202) for simulating the physical processes of the machine or multiple machines in the machine production line or part or all of the machine production line, wherein the code-based model is part of the digital twin of the machine; The status data of the machine is received via a network interface (S204), and the status data is input into the digital twin of the machine; The machine's production process is simulated in real time (S208) using a code-based model based on a digital twin, wherein the input state data are the input parameters for the simulation, and results from the simulation at the data level are assigned to the simulation's input parameters; and Based on the input parameters used for simulation and the results from the simulation, a time static or dynamic state description is provided to the machine (S210).
10. The computer device (120) according to claim 9, wherein, The simulation of the physical process occurs faster than the actual process, and the computer device is also adapted to... Operation (S214) The digital twin acts as a pilot controller for the machine, wherein the computer device generates control signals based on a time-static or dynamic state description and sends them to the machine's control device; or Operation (S214) The digital twin serves as a virtual sensor for the machine and / or the processes running on the machine, wherein the simulation running in the digital twin provides data on the corresponding processes or states that would otherwise be impossible to generate or could only be generated by using complex and expensive measurement techniques.