Method and apparatus for simulating machining on a machine tool using a self-learning system

Abstract

The invention relates to a method and a device for simulating a machining process of a workpiece on a numerical control machine tool by means of a self-learning artificial neural network AAKI. Process parameters from a machining process on a real machine tool located in a manufacturing section FA and a digital machine model implemented in a simulation section SA are provided to the artificial neural network AAKI to learn the behavior of the machine tool including the used tools and workpieces and to be reformatted by means of mathematical transformations into input parameters E1-EN. By learning the behavior of the machining process, the artificial neural network AAKI can send output files F1 back to the simulation software of the simulation section SA in reverse, so that the behavior of the digital machine model is best adapted to the conditions of the real machine tool by adjusting the simulation parameters while making it more efficient in order to optimize the machining process on the machine tool.

Description

Technical Field

[0001] The present invention relates to an apparatus and method for simulating the machining process of a workpiece on a machine tool by means of a self-learning artificial neural network, wherein the artificial neural network is able to receive process and parameter data of the machining process from both the real machine tool and the digital machine model, and use them to optimize the simulated and / or real machining process. Background Technology

[0002] Due to the increasing complexity of modern workpiece machining processes, particularly in machine-executed or machine-supported machining, new machine tools often face significant demands for improved quality or cost. Increasingly demanding process mechanics requires increasingly robust and / or precise machine kinematics, accompanied by improvements in machine mechanics, drives, or controls. However, in most cases, this also leads to increased setup times and difficult, loss-involving, and particularly expensive commissioning.

[0003] Machine tool simulation preferably reproduces the machining process of a corresponding workpiece on a digital machine tool model. To this end, various mechanical models, such as multi-mass models, geometric kinematics, or finite element models, are used to describe the physical properties and interactions of machine components and workpieces, and are combined with control software to move the machine components. Furthermore, process simulation based on penetration calculations between the workpiece and the tool can also be advantageous.

[0004] EP1901149B1 illustrates a machine simulation for defining the sequence of machining of a workpiece on a machine tool, wherein, in particular, data structures are integrated into the simulation, which enables the integration of data or behavior of components recorded by sensors on a real machine tool, thereby further improving the control description of the implemented machine model.

[0005] In addition, WO2012 / 168427A1 illustrates machine simulation of the working process on a machine tool using a virtual machine, where CNC-controlled subprocesses can be distributed across different processor cores operating in parallel, thus enabling parallel computation to accelerate the simulation process.

[0006] However, when simulating machine tools in the machining process according to existing technology, there is always a problem: it is impossible to accurately specify all state parameters of the machine tool, the tools used and / or the workpiece, especially the temporal evolution of their physical properties, without a great deal of effort.

[0007] Therefore, one object of the present invention is to provide a method and apparatus for simulating the machining process of a workpiece on a machine tool, which solves the aforementioned problems of the prior art and particularly allows for adapting a digital machine model in process simulation to the conditions and properties of a real machine tool and / or improving it as efficiently, cost-effectively, and rapidly as possible. Furthermore, one object is to optimize the simulation and the adaptation of relevant simulation parameters so that the adaptation can be automated and therefore performed as independently as possible from human error. Summary of the Invention

[0008] The features of the independent claims are presented to achieve the above objectives. The dependent claims relate to preferred embodiments of the invention.

[0009] This invention specifically describes a method and apparatus for simulating the machining process of a workpiece on a machine tool, configured to use simulation data generated by simulating the machining process performed on a digital machine model in the simulation phase of the method, and recording machining data of the machining process performed on a real machine tool in the independent manufacturing phase of the method, to collect comparative data between the simulated and real machining processes and feed them to artificial intelligence (AI) implemented in the analysis phase of the method to improve the effectiveness of the simulation. Artificial neural networks are advantageously used as artificial intelligence, and the data model used to store the simulation data is particularly advantageously configured as a time-continuous data model.

[0010] The control of the digital machine model and the actual machine tool is preferably performed as a function of predefined NC and / or PLC data, and the analysis stage is configured to learn the behavior of the machine tool, at least one tool, and / or workpiece by feeding simulation data and machining data to the AI ​​machine, and, for example, output simulation variation parameters for changing and / or optimizing simulation properties as an output dataset (AI output for process control, prediction, and optimization). Furthermore, the state modeled by the AI ​​can be used to interpret and optimize the actual machine tool state.

[0011] This invention, particularly due to improvements in simulation, also known as machining process (preferably automatic) simulation, utilizes self-learning AI to create a more accurate and cost-effective simulation environment compared to existing technologies. This simulation environment preferably operates independently, thus eliminating the need for additional pauses or waiting times. Furthermore, evaluating and interpreting the generated simulation data simplifies and optimizes actual tool machining processes with equal precision, especially because machining process simulation can quickly identify potential inefficient settings within the machine tool without the aid of complex sensors, and adjustments can be made for the corresponding work steps.

[0012] Preferably, the digital machine model may already be a remanufactured machine tool in the machining process. The digital machine model is as accurate as possible even before optimization by AI, especially by digital twins (digital images of real machine tools), so that the changes in the simulated machining process in the simulation process can output the most accurate predictions about the results of the machining process on the machine tool (machining process analysis), as long as it undergoes the same changes.

[0013] Meanwhile, the AI ​​set in the analysis phase can preferably be configured to identify the differences between the machining process of the digital tool model and the machining process of the machine tool in the machining phase through the above learning process, and can be used for automatic improvement of the machining process simulation by simulating the output of changing parameters.

[0014] For example, in a preferred exemplary embodiment, the AI ​​can be configured to adapt the conditions of the simulated machining process to the machining process on the machine tool by simulating the output of varying parameters, thereby minimizing possible differences between the simulated machining processes and realizing the machining process on the machine tool. This has particular advantages: by precisely adjusting the simulation process, not only can the corresponding machining process be simulated and predicted more accurately, but it can also be used to optimize the machining process on the machine tool more precisely.

[0015] In another preferred exemplary embodiment, the AI ​​can also be configured to learn the behavior of machine tools, workpieces awaiting machining, and various tools, and in the simulation phase (e.g., by changing rotational speed, travel path, tool to be used, workpiece geometry, or process trajectory), additionally utilize the most effective setting variables for future (or currently executed, i.e., providing optimized parameters in real-time in parallel with the currently running actual machining process). In other words, preferably, the AI ​​can be used to improve machining process simulation not only by optimizing and / or adjusting the existing simulated process, but also by using predictive methods prior to the simulation process (or in parallel with the actual machining process). For this purpose, the simulation data can also preferably be described by integrating simulated variation parameters into the output dataset, thereby preserving the known data structure in each optimization method and thus using it effectively, while minimizing any break-in time in the simulated and / or actual machining process through AI-generated predictions.

[0016] Preferably, the output dataset of the analysis step, which includes the simulated variation parameters, can be fed back to the simulation step to optimize the machining process simulation, for example, to match the simulated machining process with the machining process of the machine tool and / or to predict effective process sequences, thereby preferably creating a closed program loop and optimally configuring the corresponding simulation.

[0017] Introducing the output dataset into the simulation phase can preferably be associated with the simulation software in such a way that at least one simulation parameter (but preferably, declared as all simulation parameters changed by the analysis phase) is adjusted by integrating the output dataset from the analysis phase into the simulation phase. For this purpose, the various simulation-changing parameters of the output dataset can, for example, preferably be equipped with digital simulation tags, which are preferably read within the simulation software and instruct the simulation software to set the simulation parameters associated with the simulation tags to values ​​stored as simulation-changing parameters. Artificial neural networks can also be advantageously configured to optimize the simulation parameters of the machining process, minimizing the possible differences between the selected process parameters of the machining data and the simulation data.

[0018] The simulation variation parameters of the output dataset can include various (preferably all) simulation parameters of the simulation software to be modified, such as the geometry of the tool or machine components, grinding or cutting conditions, the trajectory or physical properties (temperature, elasticity, coefficient of friction, etc.) of the machine or workpiece components, but may also include complete command chains or fundamental changes to the settings, such as selecting the appropriate machine model, thereby providing the artificial intelligence with the maximum number of degrees of freedom for optimizing the simulation. Similarly, the output data of the analysis phase can preferably have the same data format as the simulation data of the simulation phase and the processing data of the manufacturing phase, thereby optimizing the input and output speeds of each speed by omitting possible parsers.

[0019] In a particularly preferred embodiment, the program loop can be executed iteratively and preferably automatically, the program loop including at least simulating the machining process, feeding simulation data to the analysis stage, introducing the output dataset into the simulation stage, and setting simulation parameters based on the output dataset, the number of iteration steps n being at least n≥1. As described above, a new machining process, for example with varying workpieces and / or varying tools or tool settings, can preferably be executed within the simulation stage in each iteration, and can be taught to the AI ​​and thereby adjusted to generate an optimized method that evolves continuously in each iteration step.

[0020] In the iterations of the program loop and the basic process steps between them, a fluent approach can be understood, and these can preferably begin by transmitting an output dataset based on information learned by the AI ​​to the simulation stage, thereby optimizing the simulation parameters based on the simulation variations implemented in the output dataset before the actual simulation. In the next step, a simulation of the corresponding machining process can then be performed, and the resulting data obtained in this way can be fed as process parameters to teach the AI ​​to the analysis stage, which incorporates artificial intelligence. The latter can preferably be done in such a way that the process parameters of the simulation stage and the process parameters generated on the machine tool in the machining stage are recorded in the analysis stage, correlated, and then directly transmitted to the AI ​​as input parameters for teaching the AI. Therefore, such a program loop allows the system's improved optimization capabilities to be demonstrated through each further machining process simulation, as the AI ​​continuously receives and learns new information about the machine tool, workpiece, and tooling.

[0021] However, the order and mode of action of the process steps mentioned are not limited to the exemplary embodiments mentioned. For example, machining process simulation can preferably be performed as the first step, and the process parameters obtained in this way can be fed to the AI ​​as the "actual values" of the current simulation. The AI ​​can then interpret and optimize the process parameters by transmitting the output data to the simulation stage in sequence.

[0022] Similarly, the optimization process can preferably be performed independently of the machining phase or the machining process on the machine tool. In a particularly preferred exemplary embodiment, the AI ​​can, for example, use only the information / knowledge about the machine tool, workpiece, and tool acquired in the early learning phase, and thus optimize the corresponding simulation according to the operator / user's wishes. Additional process parameters for the machining phase can be generated, for example, by the machining process executed later or earlier on the machine tool, but can also be generated before or after the relevant simulation, so that the AI's learning can be completed before or after the actual simulation or can be extended at a later point in time, thereby maximizing the flexibility of the specified optimization process.

[0023] The simulated machining parameters and machining data can preferably differ from the simulation parameters used in simulating the machining process in that they only include information relating to the machining process, i.e., the properties of the actual (real) or simulated machine tool, the tools used, and the workpiece, but preferably, at least in the case of process parameters in the simulation data, do not involve higher-level simulation settings (e.g., information from the simulation model or the simulation functions used). Therefore, in this case, similar to a human participating in a manual optimization process, the AI ​​only has measurable parameters on the machine tool that can be used for teaching and optimizing the simulation learning process, thereby minimizing the occurrence of human error or erroneous decisions by the former.

[0024] Preferably, the simulation data can also be generated in such a way that, for at least each process parameter in the machining data, an equivalent process parameter exists, can be generated in, or can be derived from the various process parameters within the simulation data, thereby further maximizing the number of training parameters for the AI, allowing the AI ​​to be trained as variably as possible. In a particularly preferred embodiment of the invention, the datasets of the various process parameters in the simulation data and the machining data can also be time-dependent, wherein at least one corresponding process parameter in the simulation data preferably exists at any point in time of the process parameters in the machining data and can be assigned to the process parameters in the machining data.

[0025] The aforementioned correlation between process parameters in the simulation phase and process parameters in the machining phase, along with the general generation of simulation data through machining process simulation, offers the following advantages: these can not only be used as comparative data for AI learning, but also preferably allow for direct conclusions regarding the machining process on the machine tool (e.g., currently running). Using simulation-generated data, such as wear, temperature, or motion data of individual machine or workpiece components, users / employees can also identify errors or incorrect settings on the machine tool, thereby enabling actions to be taken to optimize the machining process with the help of (e.g., parallel-running) machining process simulations. In other words, the aforementioned optimization process of artificial intelligence can not only be used to analyze and / or simulate the machining process more accurately, but also to optimize the machine tool in the machining phase, thereby enabling the machining process to operate more efficiently and with higher quality.

[0026] In a particularly preferred example, these measures can also be taken fully automatically, preferably with the aid of feedback between the simulation and machining stages, and can therefore be considered as a further optimization process of the AI. Process parameters present in the machining and simulation stages (e.g., PLC and / or NC data, tool selection, or speed) can also be transferred in a time-dependent manner, for example, from the simulation stage to the machine tool, and can be used to improve control of the machine tool during machining. In another exemplary embodiment, these process parameters can also be transferred directly with the aid of the AI's output files. In other words, the AI ​​can also be preferably configured in such a way that it can control the machining process on the machine tool and the digital machine model, thereby simultaneously optimizing them by implementing the output files in the machining and simulation stages. Therefore, this approach not only independently improves the machining process based on previous workflows, minimizing the time and cost of optimization according to the training level of the AI, but also ensures that the digital machine model can provide a wealth of information important to the machining process.

[0027] To further improve the method, process parameters for both the simulation and machining stages can also be stored in a data storage provided for this purpose, preferably after the corresponding manufacturing or simulation stage is generated or before it is fed into the analysis stage. The corresponding simulation data is preferably added to a simulation database implemented in an iterative loop, and the process parameters for the manufacturing stage are added to a manufacturing database, which is separate from or independently arranged from the simulation sequence and can be declared as accessible to the respective operators. This has the particular advantage that operators can also review the results of previous AI decisions at later points in time, such as after further learning and / or optimization iterations, evaluate them if necessary, and use them to adopt certain machining process improvements or changes. Furthermore, the intermediate storage of separate process parameters in their respective databases allows the AI ​​to be reset to an earlier learning stage if the resulting parameters are incorrect or unsatisfactory.

[0028] To leverage AI for learning and optimization of machining processes, multiple process parameters from the simulation and manufacturing stages, preferably derived from simulation and manufacturing databases, can also be introduced into the analysis stage. These parameters are then transformed into individual input parameters suitable for the AI ​​through comparison and fed into the AI. The input parameters are preferably not limited to the process parameters themselves or a direct comparison between the machining process performed in the machining stage and the simulated machining process. Rather, they can represent, for example, combinations of multiple process parameters, mechanical, economic, or qualitative evaluations based on process parameters or related functions. This allows the AI ​​to be freely taught according to the operator's wishes and can be used to optimize the machining process simulation.

[0029] Furthermore, to generate these input parameters, the analysis phase can preferably include multiple processing phases, but at least one data linking phase and a data interpretation phase, which are preferably connected upstream or in parallel to the AI's input. The data linking phase can preferably be configured to receive simulation and machining data stored in manufacturing and simulation databases, search for corresponding comparable process parameters, and first link them into data packets and transmit them to the data interpretation phase, wherein linking the corresponding simulation and machining data can preferably be achieved through continuous data mapping. (In particular, a time-continuous mapping method can be used, for example advantageously by associating the performed operation / machining steps, NC lines, and / or axis positions.) Conversely, in a particularly preferred embodiment, the data interpretation phase can analyze the linked simulation and machining data in this regard, compare them with other data packets, and generate any number of input parameters for the AI ​​from them, which it also forwards to the artificial intelligence. With the aid of these processing phases, which preferably operate independently of each other, an optimized, and particularly rapid, method for iteratively feeding input parameters to the AI ​​can be achieved, which can be freely adapted by the relevant operator to control the processing phase, for example, via code to be implemented through a preferred additional input interface.

[0030] Furthermore, the artificial intelligence that ultimately obtains the input parameters can be directly connected to the output of the processing stage and is preferably configured as a clustering method, a support vector machine, or particularly preferably as an artificial neural network. The latter may also preferably have a typical network structure that can be modified within the analysis stage, such as a single-layer or multi-layer feedforward or recurrent network, so that the optimal improvement method can be selected based on the complexity of the machining process to be optimized. To optimize the simulated machining process, any learning algorithm of the artificial neural network with the desired activation function can be further implemented, wherein the algorithm is preferably configured at least in such a way that the simulation variation parameters or simulation parameters generated by the artificial neural network ultimately lead to the desired optimization of the machining process simulation, for example, by adapting the simulation to the conditions on the machine tool or setting predictions for the target of the new process simulation.

[0031] As already mentioned, the advantage of using AI to optimize simulated machining processes lies in its variability to adapt to individual problems or operator preferences, and the ability to operate during optimization without human intervention, which in particular minimizes time and labor costs. However, there are other objectives as well: in a particularly preferred embodiment, AI learning and AI optimization of the simulation process can also be performed in parallel and / or independently of the actual machining processes on the machine tool, further saving time. Similarly, the AI ​​training process can be configured to take place in a phase prior to the simulation and processing, for example, by creating multiple training datasets available before the start of machining. In particular, in this way, the already trained network can also be used for multi-process simulations, since the training data naturally does not depend on the structure of the corresponding machine tool or the corresponding machining process, thus providing an optimal basis for a more specific learning phase. Therefore, the artificial neural network can be trained with the training dataset during the training phase prior to the simulation and machining processes.

[0032] In a further preferred exemplary embodiment, the AI ​​can also store the output dataset generated in a scalable technical database, where the simulated variation parameters introduced in the database can be viewed and returned to the AI. This allows operators to understand the simulated data generated in each iteration based on the simulated variation parameters and continue to use it where appropriate; furthermore, by adding individual simulated variation parameters to the recurrent network system, the latter can also preferably be used to further improve the aforementioned learning and optimization process. Furthermore, at least the technical database, simulation database, and manufacturing database can preferably be located in an external system, such as the cloud or an external network, so that operators or manufacturers, as well as external employees, can effectively access and use the stored data, for example, for further machining or variation processes, preferably for multiple machine tools.

[0033] To ensure that the machining processes in the simulation and manufacturing phases are substantially identical, both process phases may include additional features. For example, the digital machine model implemented in the simulation phase may preferably create the entire geometry of the machine tool, the tools used, and the workpiece to be machined, but at least the parts required for the corresponding machining process are created based on their actual equivalents. Furthermore, in a highly preferred exemplary embodiment, NC data and / or PLC data for controlling the various machine elements of the machine tool may also be used in the same form to move the simulated elements of the digital machine model, so that the simulation phase provides an accurate digital copy of the real machine tool.

[0034] Previously described as digital twins, this type of machine model has a particular advantage in that it allows direct comparison of any characteristic or result of the simulated machining process with data from a real machine tool. Furthermore, the simulation of the machining process on such a model can preferably output physical parameters, such as the temperature, elasticity, and coefficient of friction of the machine tool or tool and / or the workpiece to be machined (parameters that are difficult or require considerable effort to determine in reality), thereby enabling the rapid and efficient identification of any problems within the machining process. For this reason, the digital machine model can also preferably output at least time-related data series of these physical parameters. For example, these physical parameters can be defined according to the time of the corresponding machining process and / or the corresponding work steps, through additional time and location markers associated with the data series, and can be output as process parameters.

[0035] Furthermore, to further enhance the comparability between the digital machine model and the actual machine tool, the analysis of the machine tool, workpiece, and / or the components of the tools used during the product analysis phase in each case can also be performed on the actual machine tool, preferably during and / or after the machining process on the actual machine tool. This can be implemented, for example, by sensors implemented on the machine tool or externally mounted analysis units. Here, the collected data is also preferably defined as process parameters and stored in the aforementioned manufacturing database.

[0036] Furthermore, the aforementioned NC and / or PLC and geometric data used to define the digital machine model and the actual machine tool can preferably be applied to process development in the input data preprocessing stage prior to the machining process. For example, the working steps of the machining process, defined at least by the NC and / or PLC data on the machine tool and the digital machine model, can preferably be initially generated on the upstream CAD / CAM system and can be transferred from the CAD / CAM system to the aforementioned input data preprocessing stage as an operation file. The operation file should be understood as a generic data packet containing information such as machine, tool, and workpiece geometry, the sequence of motion of individual components, and / or identification structures such as UUIDs. Depending on the data format of the operation file, the input data preprocessing stage can also include one or more parsers to convert the former into a suitable input format for further processing. The input data preprocessing stage itself can handle operation files of different formats and / or externally generated files.

[0037] Then, the operation files developed in the CAD / CAM system can be converted into NC data format during the input data preprocessing stage and can be transferred to the manufacturing stage, or transferred to the simulation stage along with the corresponding manufacturing and status data of the machine tools, tools, and / or workpieces to be simulated, which are also reformatted. This has the particular advantage that the operation files can be reformatted independently of the corresponding simulation cycle, thus not unnecessarily hindering or slowing down the corresponding simulation cycle.

[0038] In addition, as a result of reformatting the operation file into NC format to identify each work step, additional work step markers can preferably be inserted into the NC data so that the real machine tool and the digital machine tool model, for example with the help of the control model, can interpret the step markers in the machining process, thereby understanding the current work step of the (digital) machine tool during the machining process.

[0039] In another preferred embodiment, the work steps previously stored in the NC data can be additionally output as reformatted structured files, for example, as XML or STEP files from the input data preprocessing stage, for interpretation in other simulation systems. Therefore, the aforementioned formatting process can be used when changing the corresponding simulation engine, allowing for viewing independently of the simulation process. Similarly, the machine tool and digital machine model can also be preferably configured such that the work steps and / or process information can be obtained from other data formats, for example, through an additional parser module, and interpreted for transmission and execution of the corresponding work steps. This preferably enables communication with other simulation networks and even allows for parallel processing of different simulations.

[0040] In summary, based on the aforementioned characteristics, a method for simulating machining on machine tools using self-learning artificial intelligence can be implemented. The analysis and improvement of the machining process on the machine tool can be optimally adapted to the characteristics of each actual machining process using AI learning processes, based on comparisons between simulation data and machine tool data. This can thus be used to predict machining processes still under development. Here, the optimization process of the simulation preferably mainly includes performing a simulation of the machining process on a digital machine model implemented in the simulation phase, transmitting the simulation results as process parameters to the artificial intelligence arranged in the analysis phase, and the artificial intelligence learning based on comparisons between the simulation phase and machine tool process parameters, feeding back the simulation parameters to be changed to the simulation phase using output files generated by the AI. Necessary optimization steps can also be performed independently or in parallel on the actual machine tool with their respective machining processes, and can be implemented fully automatically after training the artificial intelligence, which can also be done before actual manufacturing. Furthermore, the advantage of this method is that the same NC data and geometric data under the simulation conditions are used to control the actual machine tool and the digital machine model, which can generate an accurate digital copy of the actual machining process and is supported by additional analysis of the machine tool, workpiece, and tools used.

[0041] Accordingly, a corresponding device can also be preferably implemented, which realizes the above features and can simulate the machining process of workpieces on machine tools through self-learning artificial intelligence.

[0042] Therefore, the device may preferably include at least one machine tool for machining the corresponding workpiece using the aforementioned NC data, which can be independent of the machine tool control simulation device, such as a host computer or server, for simulating the machining process on a digital machine model; and an analysis unit connected to the machine tool and the simulation device, configured by means of implemented artificial intelligence to adapt to the simulation parameters within the simulation device to understand the behavior of the machine tool, the workpiece used, and / or the tool, and output its results using an output dataset. Furthermore, the machine tool is configured to transmit machining data (equivalent to simulation data developed in the simulation device) generated during or after the machining process to the analysis unit, which may result in the aforementioned program or iterative loop between the analysis unit and the simulation device.

[0043] Similarly, for individual optimization of process simulation, at least the analysis unit and simulation device can preferably operate independently of the machine tool's machining process, allowing multiple optimization or simulation processes to be guided by the machine tool. The AI ​​can iteratively execute the aforementioned procedures, but these processes can also be extended at any time using the machine tool's process parameters, for example, by accessing relevant databases, and providing more information.

[0044] To facilitate information transfer between various equipment elements, machine tools, simulation devices, and analysis devices can be further preferably configured such that they can independently transfer data to each other. Specifically, the aforementioned simulation data and machining data, as well as output files generated by the analysis device, and hardware and / or program data can be transferred to each other, and these can preferably be stored in corresponding data storage devices or a centrally located storage server to optimally execute the simulation and optimization process. In a particularly preferred exemplary embodiment, this transfer can be performed, for example, by means of an intranet and / or the Internet. Attached Figure Description

[0045] Figure 1 : A network representation illustrating a first exemplary embodiment of the described method

[0046] Figure 2 : A network representation illustrating an exemplary embodiment of the analysis phase of the described method

[0047] Figure 3 A: Network representation illustrating an exemplary embodiment of the artificial neural network of this method

[0048] Figure 3 B: Showing Figure 3 A network representation of the learning process of artificial neurons in an artificial neural network.

[0049] Figure 4 : A flowchart illustrating the learning process of artificial intelligence (machine learning device)

[0050] Figure 5 : A flowchart showing the operation procedure of the simulation phase.

[0051] Figure 6 : A network representation illustrating a second exemplary embodiment of the described method

[0052] Figure 7 This shows an exemplary list of parameters obtained from virtual and real machine tools. Detailed Implementation

[0053] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Features of the exemplary embodiments may be combined in whole or in part, and the present invention is not limited to the described exemplary embodiments.

[0054] Figure 1A first simplified embodiment of the method according to the invention is shown, which is used to simulate the machining process of a workpiece on an NC-controlled machine tool in a network representation via self-learning artificial intelligence AAKI (especially an artificial neural network). First, the machine tool geometry to be reproduced in the digital machine model is defined by a CAD / CAM system, including the workpiece, the tools used, and the work steps required for the machining process, and transmitted in an operation file D1 to the input data preprocessing stage PRE. Here, the operation file D1 may already exist as an NC file, or the work steps involved may have already been converted to NC format in the CAD / CAM system, but any other file format can be read and interpreted in the input data preprocessing stage PRE, especially due to the parser implemented in the input data preprocessing stage PRE.

[0055] Within the input data preprocessing (PRE) stage, the operation file DL is then divided according to its functional components, such as the geometric data of the machine tool, the tools used, and the workpiece, as well as the geometric data of each working step of the machine tool-related components. This data is analyzed, and existing working steps are reformatted into NC file G1 if necessary. Furthermore, to better identify each subprocess, each registered working step in the NC file receives a designated UUID, which can be queried by the corresponding machine tool or digital machine model during machining, thus identifying the currently ongoing machining process.

[0056] In the next step, the required information files are then transferred to the simulation stage SA or manufacturing stage FA provided for this purpose, thereby simultaneously dividing the process into the real part (lower stage of the network) and the simulated part (upper stage of the network). To control the real machine tool used in the manufacturing stage FA, at least the NC data formatted in the input data preprocessing stage PRE, along with other program data for machine control, such as identifiers or backup files, is exported from the input data preprocessing stage PRE to the manufacturing stage FA within the machine file G1. In the manufacturing stage FA, after repeated checks by machine operators or automatic inspection algorithms (not shown), these are transferred to the machine tool. On the other hand, with the assistance of the input data preprocessing stage PRE, the simulation stage SA, which contains the simulation software, is provided with a separate simulation information file G2. In addition to the NC data in the machine file G1, the simulation information file G2 also contains at least the machine tool geometry data required to create the digital machine model, as well as further preliminary information, such as the physical characteristics of individual machine tool components, the workpiece or tool to be used, so that the simulation stage has at least enough information or parameters to initiate the corresponding simulation.

[0057] Furthermore, if necessary, the individual data processed within the input data preprocessing stage PRE can be reformatted into another data format, such as XML or STEP, for example, with the help of the implemented compiler, and these can be stored in separate data packets G3 for use by external simulation software or hardware, thereby allowing direct comparison of different simulation engines or structures, especially in parallel simulations on multiple simulation stages SA, thus accelerating the process of simulating multiple processor cores working simultaneously.

[0058] Furthermore, following the actual steps of the method, the actual machining process in the manufacturing stage FA can now begin by providing the NC file GL to the machine tool. However, the progress of this process is not time-dependent or otherwise dependent on the method steps related to the simulation of the machining process described above, but is used to generate reference or process parameters for teaching the artificial neural network AAKI. Similarly, the number of machining processes completed within the manufacturing stage FA is not fixedly defined, but can be manually specified by the operator at any time and / or can be increased at later points in time to more accurately verify the individual parameters. The latter can be obtained through multiple sensors mounted on the machine tool or externally, or through manual input of qualitative analysis processes, such as by experts evaluating the finished workpiece during or after machining in the so-called machine, process, and production analysis stage PA, where the process parameters ultimately obtained and bundled in this way are initially stored in a database DB1 for actual manufacturing and then forwarded as machining data R1 to the analysis stage AA, which is equipped with the artificial neural network AAKI.

[0059] Equivalent to the manufacturing phase of the method, when the simulation information file G2 is received in the method phase related to the simulation of the machining process, the machining process simulation begins, and the associated optimization and / or learning process can be initiated. First, within the simulation phase SA, using the implemented simulation software and with the aid of the simulation information file G2, a corresponding digital machine model is created, including tools and workpieces. The model is as similar as possible to the actual machine tool in the manufacturing phase and can be controlled through the machining process using the also received NC data. Parallel to the actual method phase, an analysis module integrated in the simulation phase SA is used to determine any process parameters to be used. These parameters are stored by the simulation phase SA in a separate simulation database DB2, preferably in the same format as the machining data, and sent to the analysis phase AA as simulation data R2. Here, the number and type of selected process parameters can also be manually selected according to the purpose and objectives of the optimization process, and / or adapted to the process parameters to be used in the actual method phase, so that the optimization and / or learning processes can be used as effectively as possible, depending on the machine tool used or the machining process to be adjusted.

[0060] Continuing the optimization and / or learning process, in the next step, the simulation data R2 and machining data R1 (or machine tool machining data) are processed within the analysis stage AA and sent as input parameters to the artificial intelligence AAKI for learning. Conversely, the artificial intelligence AAKI is configured to learn the behavior of the machining process on the manufacturing stage FA by comparing the simulation data R2 and the machining data R1 (or machine tool machining data), and, if necessary, output multiple simulation variation parameters (machining process analysis) in the form of an output file F1. This file, once generated, is output to an iterative loop declared with output 1A1 and stored in the technical database DB3 provided for this purpose, as well as fed back to the simulation stage SA to optimize the simulation of the digital machine model.

[0061] As already described, the learning process of the AI ​​AAKI and the generation of simulation data R2 can be performed independently of the machining process on the machine tool, and therefore can be performed before or after (or during) the manufacturing of a workpiece on the machine tool. Similarly, the AI ​​AAKI can also be trained using externally introduced input parameters, such as... Figure 1 As shown in the process “AI Learning” AL, this allows for maximum flexibility in the optimization process. Furthermore, the output file F1 generated by AI AAKI may contain not only individual parameters that the simulation software needs to change, which can be used in the simulation through additional markers added to the data, but also the entire setup (simulation model, simulation time, frame rate, etc.) or the functionality to be implemented (e.g., interactive features within the model), provided these have been previously declared in AI AAKI.

[0062] The implementation of simulation variation parameters in the output file Fl of the simulation stage SA can initiate various optimization processes based on previous steps and the designed process sequence: for example, if a machining process has already been simulated for teaching in the AI ​​AAKI, this process can be adjusted within the simulation stage SA by introducing simulation variation parameters and can be viewed for further analysis (e.g., for comparison with machining processes on machine tools). However, it is also possible, particularly using the AI ​​AAKI that has already been taught or trained with external data, to create predictions for the optimization of machining processes that have not yet been implemented, thereby generating entirely new machining process simulations that are as efficient as possible by forwarding the output file F1 to the simulation stage SA.

[0063] Further conclusions can be drawn from the subsequent optimization process of the simulation. For example, if the simulation optimization is successful, the decision results and system information (system variation parameters, weights, AI parameters, etc.) created by the AI ​​AAKI can be extracted from the final output (output 2A2) and used to gain a deeper understanding and improve the actual (i.e., real) machine tool machining process. The same applies to the simulation parameters and information created during the simulation phase, in which case this is... Figure 1 The process is shown in "Performance Output" A3.

[0064] Furthermore, by optimizing the machining process simulation, direct improvements can be made to the machining process on the machine tool and the learning process of the artificial intelligence AAKI. By directly connecting the simulation stage SA to the manufacturing stage FA or the machine tool, for example through the transmission of corresponding process data, the optimization of the simulated machining process can be easily implemented into the machine tool's machining process. In other words, the system shown here allows the real machining process to simultaneously adapt to the improved simulated process sequence, and due to these sequences, machine tool optimization can be significantly accelerated, making it more cost-effective and efficient. Moreover, the optimized simulation data can be re-input into the simulation database DB2 for the artificial intelligence AAKI to use for training and further improvement of the optimization process. Therefore, each optimization process can be selected to make the artificial intelligence AAKI more accurately adapt to the conditions on the real machine tool, thereby continuously improving the decisions and results of the artificial intelligence AAKI in an iterative manner, i.e., after completing an optimization or iterative cycle.

[0065] Figure 2 It also shows that it has already been shown Figure 1 A more detailed network description of the analysis stage AA of the illustrated exemplary embodiment allows for a further explanation of the implementation of the learning parameters of the artificial neural network AAKI. First, the process parameters contained in the simulation R2 and the machining data R1 are introduced into the data linking stage AADV implemented in the analysis stage AA, where they are first analyzed with additional labels, and the corresponding comparable process parameters appearing from the simulation data R2 and the machining data R1 are correlated with each other. The latter can be implemented in various ways, such as by copying and saving individual process parameters in an intermediate storage location provided for this purpose, or by labeling these parameters with their own ID numbers, and only involves connecting any kind of process parameters (also called input parameters) that lead to the generation of learning parameters for the artificial neural network AAKI. Furthermore, a trial-and-error comparison process, such as a digital mapping, can be used to link more complex structures (e.g., time-resolved datasets) to enable optimal comparison between the simulated process parameters and the parameters obtained on the machine tool.

[0066] The process parameter pairs linked in this way, along with the individual process parameters, are then passed from the data linking stage (AADV) to the data interpretation stage (AADI) of the analysis stage (AA), where they are again identified and converted into the desired input parameters E1-EN for the subsequent artificial neural network (AAKI), and finally introduced into the artificial neural network (AAKI). Any combination or mathematical processing of the process parameters and / or input parameters of the simulation data R2 and the machining data R1, such as NC data G1 created in the input data preprocessing stage (PRE), combined geometric data G2, or operation files D1 generated in the CAD / CAM system, can be understood as the transformation or generation of input parameters E1-EN.

[0067] also, Figure 3 A and 3B illustrate a more precise representation of the learning process of the artificial intelligence AAKI, which in this exemplary embodiment is represented as an artificial neural network in other network representations. Figure 3 A shows the typical inputs and outputs within the network. Figure 3 A shows the typical inputs and outputs within the network. Figure 3 B illustrates a detailed representation of the decision-making process occurring in the intermediate layers within the artificial neuron. Here, the plotted structure of the network shown should be understood as merely an exemplary representation.

[0068] As already described, the integration of the latent input parameters E1-EN is first generated through the accumulation of process parameters generated by the manufacturing stage FA and the simulation stage SA, as well as other information appearing in the previous methods. It is further combined through the data interpretation stage AADI, and can ultimately be introduced as a new combination element KOM or as one of the aforementioned parameters as input to the first learning layer of the network. Therefore, according to the rules of a self-learning system, within the relevant neurons of each network layer, a weighted weight W1-WN is assigned to each input parameter. Before the start of the learning phase, these parameters are assigned random values ​​and can be gradually changed to the desired decision weights W1-WN in each learning iteration through a "trial and error" algorithm. In a general sense, the interaction between the input parameters E1-EN and the corresponding weights W1-WN can be performed by a simple multiplication Ei×Wi of the i-th parameter and the i-th weight; however, other or more complex functions may be chosen in other exemplary embodiments.

[0069] In the next step, then by means of the transfer function F i The inputs are weighted and combined to form the network input. For example, according to the rules of artificial neural networks, ∑ can be used. i E i W i The network input is created by a simple summation in the form of , but this method can also vary depending on the problem and adaptability of each optimization attempt.

[0070] The network input is then processed through a predefined activation function A. i This is used to determine whether an artificial neuron activates ai upon receiving all weighted and aggregated inputs and is therefore allowed to pass information to the next layer. Typically, the activation functions A at the network input points are compared. i Whether a certain threshold is exceeded, which must also be learned (thus activating neuron ai), or whether the resulting value is insufficient and the neuron remains inactive. As before, the activation function is free to be chosen, but at least in this exemplary embodiment, the sigmoid function is favored due to its continuous and differentiable shape at every point.

[0071] By training an artificial neural network, the weights W1-WN obtained with the aid of the learning process can be used to transfer the selected input EN from one network layer to the next, and thus the outputs 1 and 2 in the final layer can be used as preferred simulation parameters. Depending on the network structure, these can also be returned to the original input integration, such as in a recurrent network system, or to specific neurons or layers to generate a feedback system created from the output file. The learning of the network, or the aforementioned weights W1-WN, and in special cases, the thresholds, can be performed, for example, using predefined training data before the actual input parameters E1-EN are provided, and can therefore be viewed independently of the actual optimization process.

[0072] Figure 4The structure of the learning process is illustrated again using a flowchart describing the learning process, where "start" is understood as the general operational concept for executing the AI ​​training process, and "end" is understood as successfully teaching the AI ​​the desired characteristics. Each learning process begins with a general decision by the system (or person) to conduct an AI AAKI learning process, which may involve various factors as the basis for these decisions. The learning process can be used, for example, to generate predictions for new tools, operating modes, or machine tools to be reinstated, and can also make existing machining process simulations largely adapt to the required environment, such as operating procedures, thereby improving the accuracy and efficiency of the simulation. If the decision is successful, the machine information SA02 and machining conditions SA03 required for the training process are collected, and the analysis step AA integrated with AI AAKI is prepared first. All physical information about the machine tool, such as geometry, dimensions, material properties, or physical characteristics of tools, workpieces, and / or individual machine tool components, which can be used to successfully describe the machining process to be learned, is considered machine information. Similarly, all machine tool settings or conditions that must be applied to the various (simulated) components of the machine tool to implement the above machining process, such as tool path, rotational speed, or machining speed, can be considered machining conditions. The data can be generated from externally stored datasets (e.g., servers or the cloud). Alternatively, it can be supplemented or overridden by the result parameters of the machining process simulation SA04 or the result parameters of the machining process SA05 performed in parallel or previously on the machine tool, so that current information about the tool machining can be continuously fed into the learning process of artificial intelligence at any time.

[0073] Furthermore, direct comparison of simulation results with actual machining conditions allows for further analysis and decision-making processes (SA06) in the next step. After creating the input parameters defined for the Artificial Intelligence AAKI, a decision can be made, for example, by comparing simulation data R2 and machining data R1, as already mentioned. However, a decision can also be made using the inputs or other parameters themselves to determine whether it is necessary to learn the Artificial Intelligence AAKI, or whether, for example, the current simulation settings already meet the required conditions. To this end, for example, the aforementioned parameters can be compared with well-defined limit values. If they are below the limit values, the learning process can be automatically continued using them, thus enabling individual start-ups for each simulation or learning process. As a result of a positive decision (“Yes”), the input parameters (SA07) can also be introduced into the analysis stage (AA) or the Artificial Intelligence AAKI according to the steps described above, and the learning process (SA08) can begin.

[0074] Figure 5 The structure of the machining process simulation and the operation process of continuing the process is further illustrated based on another flowchart. In this case, "start" should be regarded as the general operational concept of initiating the machining process simulation, while "end" should be regarded as the successful completion of the same simulation.

[0075] Similar to the graphical structure of a learning process, each simulation process begins with the decision to execute the learning process SB01, which can be made manually, for example by an employee, or automatically, for example through integrated program code or artificial intelligence AAKI. Similarly, if the input is positive (“Yes”), the (digital) machine information SB02 and machining conditions SB03 required for the simulation are retrieved from available servers, cloud services, or other types of databases and prepared for implementation in the corresponding simulation stage Sat. The former may include the aforementioned information data regarding the physical or kinematic conditions of the machining process, as well as general software settings related to the simulation (e.g., the engine to be used, the simulation model, and the parameters to be set in the software).

[0076] In the next optional step, the machine information and machining conditions to be introduced, or the simulation parameters to be obtained from them, can also be adjusted by introducing simulation variation parameters generated by the analysis step AA, and thus SB04 can be optimized using the decisions of the trained artificial intelligence AAKI. Depending on the machining process, the simulation variation parameters, or the output file F1 containing these parameters, can be directly output from the analysis step AA or obtained from an existing database (e.g., the technical database DB3), provided that the associated machine tool attributes correspond to the machine tool information and machining conditions of the machining process currently being simulated.

[0077] If the simulation parameters used in the simulation then correspond to the desired specifications, the machining process simulation begins and SB05 is subsequently evaluated, meaning the actual simulation process can be considered complete. However, in a continuous approach, the results or knowledge gained from the machining process simulation can be applied to the machining process on the machine tool after comparing certain conditions SB06 (e.g., whether the efficiency or results of the simulation meet certain requirements). For example, after ensuring the quality of the simulation results (SB06 – “Yes”), the latter can initially be output separately SB07 and transferred and / or reused for further use, such as analyzing potential inefficiencies within the machine tool. Furthermore, simulation results can (optionally) be transferred, for example, by classifying the simulation parameters used in the machining process simulation as efficient and transferring them directly to the machine tool SB08, thereby realizing the efficiency and process increases obtained through optimization simulation also on the actual machine tool.

[0078] Figure 6 It also shows something similar to Figure 5 A detailed view of the network representation of the exemplary embodiment shown. See also Figure 1The diagram illustrates, in particular, other elements and interactions between different process stages. For example, communication paths of the various exemplary elements of the input data preprocessing stage PRE are shown and linked to components of the simulation stage SA and the manufacturing stage FA. For instance, elements of the program structure, such as geometric data of the digital machine model or NC data already defined as G-code in this case, are fed to the manufacturing stage FA, which records data for manufacturing real workpieces, and to the simulation stage SA located in the simulation engine lo. Other elements not yet described are the G-code interpreter in the simulation stage SA, which first reads the G-code of the NC data generated in the input data preprocessing stage PRE and introduces a kinematics solver to realize motion data of individual machine elements into the model for simulation, as well as a virtual NC, which creates virtual copies of the NC data in parallel with the interpretation of the corresponding NC data G-code, thus making them available to the aforementioned kinematics solver. Furthermore, reference should be made to the various parameters stored in the aforementioned databases DB1, DB2, and DB3, especially the parameters stored in the simulation database DB2. For example, in addition to the process parameters of the simulation stage, the SA already described contains more simulation-related information, such as TCP, thus providing various parameters for teaching the artificial intelligence AAKI.

[0079] Figure 7 An exemplary comparison of various analytical parameters obtained using real machine tools and digital machine models is also shown, where x represents the probability of obtaining the corresponding parameter. This data can, for example, be used as training data for the artificial intelligence AAKI. In particular, it is evident from this example that a large number of other elements, such as the engagement depth or width of individual tools in the workpiece, can be readily obtained compared to real machine tools and used for analyzing and further improving machining processes. In this respect, especially during the complex, difficult-to-implement, and / or costly introduction phase of new machine tools, implementing machining processes in digital machine models provides a cost-effective method for improving process flows.

[0080] These features, components, and specific details can be exchanged and / or combined according to their intended use to produce further embodiments. Any modifications that are within the knowledge of those skilled in the art or of ordinary skill are implicitly disclosed in this specification.

Claims

1. A computer-implemented method for simulating a machining process of a workpiece on a machine tool using functions of numerically controlled NC data and / or PLC data, wherein a digital machine model of the machine tool is used to simulate the machining process, the method comprising: - Perform the digital machining process and store the simulation data by simulating the machining process on the digital machine model in the simulation phase based on the NC data and / or PLC data; - Record machining data of the machining process on the machine tool, wherein the machining process is executed using functions of the NC data and / or PLC data; - The simulation data of the digital machining process and the machining data of the machining process on the machine tool are fed into artificial intelligence executed in the analysis phase, and the simulation data and the machining data are linked, wherein the analysis phase learns the behavior of the machine tool, at least one tool and / or workpiece based on the simulation data of the digital machining process and the machining data of the machining process, and identifies the differences in the machining process between the digital machine model and the machine tool, and In order to link the simulation data to the machining data, continuous data mapping is used to associate the machine tool's sensor data with the corresponding analysis data from the simulation process. The output dataset contains simulation variation parameters used to change and / or optimize simulation properties for automatic improvement of machining process simulation. The output dataset is fed back to the simulation stage to optimize the simulation of the digital machine model; The optimized simulation results and / or simulation parameters are directly transmitted to the machine tool to control the machine tool during the machining process.

2. The computer-implemented method according to claim 1, The process of simulating the machining process, feeding the simulation data into the analysis stage, feeding the output dataset back into the simulation stage, and the changes in the simulation parameters in the simulation stage based on the output dataset form a program loop, which is used to continuously adjust the simulation of the machining process.

3. The computer-implemented method according to claim 1 or 2, At least one simulation parameter is changed based on the output dataset of the analysis step within the simulation step, and The simulation data generated by the simulation, or at least one process parameter of the simulation data, is stored in the simulation database before being fed into the analysis stage.

4. The computer-implemented method according to claim 1 or 2, in, For each process parameter in the machining data, at least one corresponding process parameter exists in the simulation data and / or is generated during the simulation, and each process parameter is associated with the machining data, and / or The analysis process compares the machining data with the simulation data, and defines the input parameters of the machine learning device by comparing the machining data and the simulation data.

5. The computer-implemented method according to claim 1 or 2, The digital machining process is executed in parallel or before the machining process on the machine tool, and the real-time output of the performance data of the current machining process is made possible by outputting the simulation data and machining data to the analysis stage.

6. The computer-implemented method according to claim 1 or 2, The machine learning device is an artificial neural network configured to optimize the simulation parameters of the machining process simulation so that the difference between the selected process parameters of the machining data and the simulation data is as small as possible.

7. The computer-implemented method according to claim 1 or 2, The optimization of the learning and simulation process of the machine learning device is performed in parallel with the machining process on the machine tool and / or independently of the machining process on the machine tool.

8. The computer-implemented method according to claim 1 or 2, The output dataset from the machine learning device is stored in a scalable technology database, and The machine learning device accesses the output dataset stored in the technology database to provide feedback on the learning process.

9. The computer-implemented method according to claim 1 or 2, in, The same NC data is used for the machining process on the machine tool and for the simulation of the machining process on the digital machine model in order to match the working steps between the machining process and the simulation process.

10. The computer-implemented method according to claim 1 or 2, in, The physical parameters of the machine tool, tool, and workpiece to be processed are output by simulating the machining process on the digital machine model, and the physical parameters of the machine tool, tool, and workpiece to be processed are defined according to the machining process and / or the time of each working step.

11. The computer-implemented method according to claim 1 or 2, The NC data used to identify each work step is equipped with additional tags, and The markers within the NC data are used to interpret the machine tool and digital machine tool model, thereby enabling the tracking of the working steps and / or positions of the machine tool and / or digital machine tool at determinable points in time.

12. The computer-implemented method according to claim 1 or 2, The machining process steps are additionally output as structural data, used to explain the working steps in other simulation devices, and The machine tool and the digital machine model extract and implement work steps and / or process information from other data formats through a parser.

13. The computer-implemented method according to claim 12, The structured data is output as an XML file or a STEP file to explain the working steps in other systems.

14. An apparatus for controlling a machining process of a workpiece using a machine tool by means of functions of numerical control NC data and / or PLC data, the apparatus comprising: - A machine tool for machining the workpiece using specified NC and / or PLC data. - A simulation device, controlled independently of the machine tool, for simulating the machining process on a digital machine model based on specified NC data and / or PLC data. - An analysis unit connected to the machine tool and the simulation device, used to adjust simulation parameters within the simulation device. in, The machine tool is configured to transmit machining data of the machining process on the machine tool to the analysis unit, and the simulation device is configured to transmit simulation data of the machining process simulated on the digital machine model to the analysis unit, wherein... The analysis unit includes artificial intelligence and is configured to link the simulation data with the machining data, wherein the artificial intelligence learns the behavior of the machine tool, at least one tool, and / or workpiece based on the simulation data and the machining data, and identifies differences in the machining process between the digital machine model and the machine tool. Through continuous data mapping, the sensor data of the machine tool is correlated with the corresponding analysis data in the simulation process, thereby linking the simulation data to the machining data. The analysis unit is configured to use transmitted machining data and simulation data to enable a machine learning device located within the analysis unit to learn the behavior of the machine tool, tool, and / or workpiece to be machined, and to output the analysis results of the machining process from the machine learning device; wherein The analysis unit is configured to output simulation variation parameters for changing and / or optimizing simulation properties as an output dataset for automatic improvement of machining process simulation; wherein The analysis unit is configured to feed the output dataset back to the simulation stage to optimize the simulation of the digital machine model; and wherein The simulation device is configured to directly transmit the optimized simulation results and / or simulation parameters to the machine tool to control the machine tool during the machining process.

15. The apparatus of claim 14, wherein the machine tool, the simulation device, and the analysis unit are configured to transmit data to each other. in, The data is transmitted via an intranet and / or the Internet.

16. The apparatus of claim 14 or claim 15, wherein the simulation device and the analysis unit are independent of the machining process on the machine tool, and The analysis unit is configured to continuously match the machining process on the digital machine model with the machining process of the machine tool by transmitting the output dataset of the machine learning device to the simulation device.

Images