Method for readjusting a manufacturing process

Abstract

The invention relates to a method (100) for readjusting a separating process. As part of the method (100), a workpiece (14) and workpiece parameters are provided, wherein the workpiece parameters represent properties of the workpiece (14) that are relevant to the separating process. A manufacturing means (18) for separating the workpiece (14) and parameter values of control parameters for controlling the manufacturing means (18) are also provided. While carrying out a separation step from a group of specified separation steps, the occurring separating forces and / or travel, in particular of the manufacturing means (18), are measured. Actual parameter values of the workpiece parameters are ascertained from the measured separating forces and / or travel using an evaluation model. The actual parameter values of the workpiece parameters are converted into a prediction model (44). The prediction model (44) determines readjustment values of the control parameters in order to minimize the difference between the actual parameter values and target values of the workpiece parameters at separation of the workpiece (14). The control parameters are set to the readjustment values. The method steps from controlling the manufacturing means (18) so as to separate the workpiece (14) to readjusting the control parameters are repeated until the specified separation steps have been carried out.

Description

[0001] Method for readjusting a manufacturing process

[0002] The invention relates to a method for readjusting a cutting process and / or forming process, wherein “cutting” according to the invention is understood to mean in particular a partial cutting of a workpiece, in particular the at least partial, preferably complete, cutting through of a workpiece.

[0003] A readjustment method is known from DE 10 2019 216 315 A1 in connection with a laser processing system for processing a workpiece. The workpiece is processed by the laser processing system under a laser processing condition (e.g., a movement of the processing head). A machine learning device of the laser processing system is trained to predict a suitable laser processing condition for desired temperature distributions in the workpiece. Simulation results and measured data on various temperature distributions in the workpiece during laser processing are used for this purpose.

[0004] DE 10 2017 109 713 A1 relates to a roll forming stand with a sensor for measuring the force exerted by the roll forming stand on the sheet metal during forming. A detector algorithm is configured to detect the condition (in particular wear) of the roll forming stand based on the sensor signal in order to display the condition to a user of the roll forming stand.

[0005] For the adjustment of separation processes, it is known to change the settings of the production equipment used for such a separation process by comparing target values ​​with measured values ​​of positions or reference variables of the production equipment. It is also known that such adjustments can also be carried out by adapting the parameters of the separation process to the separation process itself within the framework of adaptive control.

[0006] With the known methods, the effects of deviations of a specific workpiece to be separated and / or of the production equipment from specified standards on the separation or forming process often cannot be taken into account with sufficient accuracy.

[0007] It is therefore an object of the invention to provide a method for adjusting a cutting process in which the effect of the production means on the workpiece can be precisely taken into account during the cutting process.

[0008] This object is achieved according to the invention by a method for readjusting a separation process, which comprises the following steps:

[0009] I. Providing a workpiece;

[0010] II. Providing a tool for separating the workpiece;

[0011] III. Providing separation steps for separating the workpiece;

[0012] IV. Providing parameter values ​​of control parameters for controlling the production tool;

[0013] V. Providing workpiece parameters to characterise the workpiece and specifying target values ​​of the workpiece parameters that the workpiece should have after cutting;

[0014] VI. Storing the parameter values ​​of the control parameters in a machine control system;

[0015] VII. Controlling the production tool by the machine control system using the control parameters to carry out a separation step from the provided separation steps;

[0016] VIII. Measuring the separation forces and / or strokes occurring during the separation step;

[0017] IX. Evaluation of the values ​​of the separation forces and / or strokes measured in step VIII by an evaluation model in order to determine actual parameter values ​​of the workpiece parameters for characterizing the workpiece; X. Determination of readjustment values ​​of the control parameters with the aid of the actual parameter values ​​of the workpiece parameters determined in step IX by a forecast model, so that, as part of the calculation by the forecast model with the readjustment values ​​of the control parameters when separating the workpiece, a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters of the workpiece is minimized, wherein the difference is determined after carrying out the provided separation steps;

[0018] XI. Readjustment of the control parameters to the readjustment values;

[0019] XII. Repeat steps VI to XI until the provided separation steps are completed.

[0020] By evaluating the forces and strokes occurring during the respective cutting step, precise statements can be made about the actual parameter values ​​of the workpiece parameters of the respective workpiece being cut. The forecast model can then calculate precise parameter values ​​of the control parameters with which the desired separation of the workpiece is achieved using the specific manufacturing tool and workpiece used. This allows the manipulated variables of the cutting process to be precisely adjusted depending on the workpiece and manufacturing tool. The two-stage controller in the form of the evaluation model and the forecast model allows the cutting process to be flexibly adapted to different workpieces and manufacturing tools.The cutting process can be adapted if the workpiece on which the cutting process is carried out and / or the production tool deviate relatively significantly from specified characteristics (e.g. a specified sheet thickness). The calculation of the parameter values ​​of the control parameters adapted to the specific workpiece leads to a high product quality of the manufactured workpiece. Adapting the parameter values ​​of the control parameters to the respective workpiece also reduces the risk of damage to the production tool in an automated cutting process. By evaluating the forces and stroke distances occurring during each cutting step, continuous monitoring of the cutting process is possible. The production tool is understood to mean in particular the tool or semi-finished product that acts on the workpiece in order to carry out the cutting process. Deviations from standards can arise, for example, due to wear and tear on the production tool.

[0021] Statements which, in the context of the application, refer to a separation of the workpiece may also generally apply to a manufacturing process, in particular to a forming process, of the workpiece in the context of the process according to the invention.

[0022] According to an advantageous embodiment of the method, the following steps are carried out after step XII:

[0023] XIII. Measuring the actual parameter values ​​of the workpiece parameters of the workpiece;

[0024] XIV. Determining differences between the measured actual parameter values ​​of the workpiece parameters and the target values ​​of the workpiece parameters;

[0025] XV. Determining readjustment values ​​of the control parameters with the aid of the differences determined in step XIV by the forecast model, so that, as part of the calculation by the forecast model with the readjustment values ​​of the control parameters during separation according to the provided separation steps, a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters is minimized, wherein the difference is determined after carrying out the provided separation steps;

[0026] XVI. Readjustment of the control parameters to the readjustment values;

[0027] XVII. Carrying out steps I to XII, wherein in step I a further workpiece is provided and in step IV the control parameters adjusted to the adjustment values ​​in step XVI are provided.

[0028] It is best to check the finished workpiece again after cutting for any deviations from the desired result of the cutting process. This can further improve the cutting accuracy of subsequent workpieces. This particularly applies to manufacturing defects that the workpiece exhibits as a whole after cutting (global manufacturing defects).

[0029] In an advantageous embodiment, the workpiece parameters comprise flatness parameters to characterize the flatness of a workpiece, wherein the target values ​​of the flatness parameters comprise desired parameter values ​​of the flatness parameters to achieve a desired flatness of the workpiece after performing the provided separation steps. The flatness is determined in particular according to the standard DIN EN ISO 1101. Flatness parameters can, for example, be parameters that specify the curvature of a surface of the workpiece and / or the deviations from a plane through the surface of the workpiece.

[0030] In a preferred embodiment of the method, the prediction model comprises a neural network, in particular a multi-layer perceptron. A multi-layer perceptron (MLP) is particularly suitable for making predictions from data sets, especially in the case of nonlinear relationships between the workpiece parameters and the control parameters.

[0031] In a further development of the aforementioned embodiment, the prediction model is trained in the following steps: i. Determining first parameter values ​​of the workpiece parameters of a test workpiece and determining parameter values ​​of the control parameters before carrying out the subsequent step ii; ii. Initializing weights of the prediction model; iii. Separating the test workpiece with the manufacturing tool while changing the first parameter values ​​to second parameter values ​​of the workpiece parameters; iv. Measuring the parameter values ​​of the control parameters after carrying out step iii; v. Calculating predicted parameter values ​​of the control parameters for separating the test workpiece while changing the first parameter values ​​to the second parameter values ​​of the workpiece parameters by the prediction model; vi. Calculating a difference between the predicted parameter values ​​of the control parameters and the parameter values ​​of the control parameters measured in step iv; vii.Updating the weights of the forecast model to minimize the difference according to step vi; viii. Repeating steps v to vii until the error according to step vi reaches a predetermined minimization limit.

[0032] An advantageous embodiment of the method is one in which the evaluation model has an algorithm for evaluating the forces and / or strokes measured during separation of the workpiece, a decision tree, a decision forest, which can also be referred to as a random forest, and / or a neural network in order to classify the workpiece into a workpiece class that belongs to predefined workpiece classes, and / or to classify the production tool into a production tool class that belongs to predefined production tool classes. The neural network can interact with the decision tree or the decision forest, for example, in particular in identifying relevant branches of a decision tree for classifying the workpiece. The neural network can be designed as the algorithm with which the aforementioned measured forces and / or strokes are evaluated or interact with this algorithm.

[0033] According to a preferred embodiment of the method, the evaluation model is selected from a group of evaluation models, in particular based on the parameter values ​​of the control parameters provided in step III and / or based on the parameter values ​​of the workpiece parameters provided in step IV. This allows the evaluation model to be precisely tailored to the type of workpiece and / or production tool in order to accelerate and / or refine the evaluation. Different materials or geometries of the workpieces and / or production tools may require different algorithms to evaluate the forces and / or strokes occurring during separation. In particular, this can reduce the uncertainty of probabilistic calculations of the evaluation model.

[0034] In an advantageous embodiment of the method, the prediction model is selected from a group of prediction models before step X, in particular based on the parameter values ​​of the control parameters provided in step III and / or based on the parameter values ​​of the workpiece parameters provided in step IV. This can accelerate and / or refine the prediction of the readjustment values ​​of the control parameters. Thus, different workpieces and / or production tools may require the use of different neural networks.

[0035] An advantageous embodiment of the process is one in which the machine control compares the parameter values ​​determined in step IX with the target values ​​of the control parameters and / or workpiece parameters before step X. This enables a time-saving check to determine whether the desired workpieces and / or production tools are being used during the cutting process. The target values ​​of the workpiece parameters can fully or partially match the target values ​​of the workpiece parameters.

[0036] In an advantageous embodiment of the method, the control parameters include a force with which the manufacturing means acts on the workpiece, geometric parameters of the manufacturing means, and / or path parameters of a path on which the manufacturing means is guided.

[0037] In a preferred embodiment of the method, the workpiece parameters comprise material parameters of a material of the workpiece, geometric parameters of the workpiece, in particular a thickness of the workpiece and / or a shear strength of the workpiece.A further embodiment of the method is characterized in that the workpiece parameters have cutting surface parameters in order to characterize the quality of a cutting surface, wherein the target values ​​have desired parameter values ​​of the cutting surface parameters for a desired quality of the cutting surface, wherein in step X, readjustment values ​​of the control parameters are determined with the aid of the parameter values ​​of the workpiece parameters determined in step IX by the prediction model or a further prediction model in such a way that a difference between the target values ​​of the cutting surface parameters and the parameter values ​​of the cutting surface parameters of the workpiece is minimized after carrying out the provided separation steps. The predictions of the prediction model and / or the further prediction model enable a particularly precise design of the cutting surface.The further prediction model can be implemented as a neural network, particularly a multi-layer perceptron. The training of the further neural network can be carried out following the steps described above for training the neural network.

[0038] In a preferred embodiment of the method, the cutting and / or forming process is designed as a punching process. This method enables particularly precise punching of a workpiece.

[0039] Further advantages of the invention will become apparent from the description and the drawings. Likewise, the above-mentioned and further-described features can be used individually or in combination in any desired manner. The embodiments shown and described are not intended to be exhaustive, but rather are exemplary in nature for describing the invention.

[0040] Detailed description of the invention and drawing

[0041] Fig. 1 shows a schematic side view of a separation system; Fig. 2 shows the separation system in a schematic side view, with a predetermined separation step being carried out;

[0042] Fig. 3 shows the separation system in a schematic side view after carrying out the specified separation step;

[0043] Fig. 4 shows schematically the sequence of the method according to the invention.

[0044] Fig. 1 shows a schematic side view of a cutting system 10, which is positioned on a floor 12. A workpiece 14 in the form of a sheet metal is moved on transport rollers 16 of the cutting system 10. Alternatively or in addition to the transport rollers 16, brush tables of the cutting system 10 can be used. In the embodiment shown, the cutting system 10 is designed as a punching machine. The cutting system 10 has a production means 18, which is designed in particular as a tool or a semi-finished product. In the embodiment shown, the production means 18 has a punch 20, with which through-holes 22 (see Figures 2 and 3) are punched into the workpiece 14. A frame 40 of the cutting system 10 is formed with two vertically aligned supports 24 and one horizontal support 26, wherein the horizontal support 26 is located between the vertical supports 24 and is fastened to the vertical supports 26.A lifting rod 28, which can be moved vertically, is mounted in the horizontal support 26. The lifting rod 28 holds the production tool 18, which can be moved together with the lifting rod 28 to punch the workpiece 14. Alternatively, the frame 40 can also be C-shaped according to the invention.

[0045] A sensor 30 is arranged in the cutting system 10, which is configured to measure forces that occur during the cutting process according to the invention and / or to measure paths traveled during the cutting process, in particular stroke paths of the tool 18. The sensor 30 can also have another position suitable for measuring the forces and / or paths. A die 32 of the cutting system 10 is formed with a recess 34 through which the punched-out parts of the workpiece 14 are removed. A machine control system 36 of the cutting system 10 serves to control a cutting process according to the invention.

[0046] The cutting process consists of a sequence of cutting steps that are defined prior to cutting and, in particular, stored in the machine control system 36. Control parameters are used to control the tool 18, with parameter values ​​of these control parameters being stored in the machine control system 36. These parameter values ​​determine the actions of the tool 18 during the individual cutting steps. The control parameters can include a force with which the tool 18 acts on the workpiece 14, geometric parameters of the tool 18, and / or path parameters of a path along which the tool 18 is guided.

[0047] The properties of the workpiece 14 are encoded in parameter values ​​of workpiece parameters, wherein the workpiece parameters are determined before the workpiece 14 is separated. The workpiece parameters can include material parameters of a material of the workpiece 14, geometric parameters of the workpiece 14, in particular a thickness of the workpiece 14 and / or a shear strength of the workpiece 14. At the beginning of the method according to the invention, values ​​of the workpiece parameters are specified as parameter values ​​that are typical for the type of workpiece 14 to be separated. Furthermore, target values ​​of the workpiece parameters that the workpiece 14 is to have after separation by the specified separation steps are stored in the machine control 36.

[0048] Fig. 2 shows the cutting system 10 in a schematic side view, with one of the specified cutting steps being performed. The tool 18 is moved vertically downward on the lifting rod 28 and punches a through-hole 22 (see Fig. 3) into the workpiece 14. The sensor 30 in the tool 18 measures the cutting forces that occur during the cutting process and / or the strokes that the tool 18 travels during the cutting process.

[0049] Fig. 3 shows the cutting system 10 in a schematic side view after the specified cutting step has been performed (cf. Fig. 2). The workpiece 14 has the through-hole 22 punched by the production means 18. The forces and / or strokes measured by the sensor 30 during the cutting step are evaluated by an evaluation model 42 stored in the machine control system 36. Alternatively or in addition to the use of a machine control system 36, the models mentioned in the application and / or the data used can, according to the invention, be stored in an external processing unit (not shown). The evaluation determines actual parameter values ​​of the workpiece parameters, which define the properties of the specific workpiece 14 on which the cutting step was performed. In the event of deviations, the specified parameter values ​​of the workpiece parameters of the workpiece 14 are replaced by the actual parameter values.The deviations of the actual parameter values ​​of the workpiece parameters from the specified parameter values ​​can arise, among other things, from fluctuations in the thickness of the workpiece 14 or from differences between the actual material of the workpiece 14 and the desired material.

[0050] A forecast model 44 stored in the machine control system 36 determines readjustment values ​​for the control parameters based on the actual parameter values ​​of the workpiece parameters. The readjustment values ​​of the control parameters are determined in such a way that, when processing the further cutting steps, a difference between the target values ​​of the workpiece parameters and the parameter values ​​of the workpiece parameters of the workpiece 14 after the provided cutting steps have been carried out is minimized. The control parameters are set to the readjustment values. This makes it possible, for example, to take wear of the production tool 18 into account. A further cutting step is then carried out and the control parameters are readjusted in the manner described above until all specified cutting steps have been carried out. The workpiece parameters can, in particular, have cutting surface parameters in order to characterize the quality of a cutting surface 46 that is produced during cutting.The prediction model 44 and / or another prediction model 48 stored in the machine control 36 can be used to determine readjustment values ​​of the control parameters such that a difference between target values ​​of the cutting surface parameters and actual parameter values ​​of the cutting surface parameters of the workpiece 14 is minimized after performing the provided separation steps.

[0051] Fig. 4 schematically shows the separation of the workpiece 14 (see Fig. 1 ) and readjustment of the control parameters within the scope of the method 100 according to the invention. In a first sub-step 102, the workpiece 14 is provided. In a second sub-step 104, the production means 18 for separating the workpiece 14 is provided. In a third sub-step 106, separation steps for separating the workpiece 14 are specified. In a fourth sub-step 108, parameter values ​​of the control parameters for controlling the production means 18 during the execution of the individual separation steps are made available. In a fifth sub-step 110, the workpiece parameters for characterizing the workpiece 14 are defined and target values ​​of these workpiece parameters are specified, which the workpiece parameters should have after the separation of the workpiece 14. The aforementioned sub-steps can take place one after the other or in parallel.

[0052] In a sixth sub-step 112, the parameter values ​​of the control parameters are stored in the machine control system 36. In a seventh step 114, the production tool 18 is controlled within the scope of a predetermined separation step by the control parameters stored in the machine control system 36 in order to separate the workpiece 14. In an eighth sub-step 116, the stroke distances and forces occurring during this separation are measured by the sensor 30 in the production tool 18. In an optional ninth sub-step 118, an evaluation model 42 is selected from a group of evaluation models 42 (see Fig. 3) in order to evaluate the measurement signals measured in the eighth sub-step 116. The selection of the evaluation model 42 is made in particular based on the parameter values ​​of the workpiece parameters, the control parameters, the type of workpiece 14 and / or the type of production tool 18.

[0053] In a tenth sub-step 120, the measurement signals measured in the eighth sub-step 116 are evaluated by the evaluation model 42, which may have previously been selected from a group of evaluation models 42 in the optional ninth sub-step 118. Actual parameter values ​​of the workpiece parameters are determined. During the evaluation by the evaluation model 42, noise components of the measurement signals from the eighth sub-step 116 are preferably removed. The evaluation model 42 has, in particular, an algorithm for evaluating the measurement signals and / or a decision tree for classifying the workpiece and / or the production tool. Alternatively or in addition to the decision tree, the evaluation model 42 may have a decision forest. Alternatively or in addition to a decision tree and / or decision forest, other suitable evaluation models may be used according to the invention.Preferably, the evaluation model 42 comprises a neural network, which can be combined in particular with a decision tree, e.g. in order to accelerate classifications.

[0054] In an optional eleventh sub-step 122, the actual parameter values ​​of the workpiece parameters determined using the evaluation model 42 are compared with the target values ​​of the workpiece parameters, whereby these target values ​​can, in particular, correspond to the target values ​​of the workpiece parameters. This comparison can be used to determine whether the desired workpiece 14 and / or the desired production tool 18 is being used.

[0055] In an optional twelfth sub-step 124, a forecast model 44 is selected from a group of forecast models 44 (see Fig. 3) in order to determine adjustment values ​​for the control parameters based on the actual parameter values ​​of the workpiece parameters determined in the tenth sub-step (see below). The forecast model 44 is selected, in particular, based on the parameter values ​​of the workpiece parameters, the control parameters, the actual parameter values ​​of the workpiece parameters, the type of workpiece 14, and / or the type of tool 18.

[0056] In a thirteenth sub-step 126, the actual parameter values ​​of the workpiece parameters determined with the evaluation model 42 are made available to the prediction model 44. The prediction model 44 is configured to determine readjustment values ​​of the control parameters such that, with the control parameters set to the readjustment values, a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters of the workpiece 14 is minimized. This difference is determined after the separation process has been carried out, in particular after the specified separation steps have been carried out. For this purpose, the prediction model 44 preferably comprises a neural network, in particular a multi-layer perceptron.

[0057] In particular, within the scope of step 126, a determination of the quality of the cutting surface 46 (see Fig. 3) generated during separation according to the seventh sub-step 114 can also be made. This determination can be made by the prediction model 44 or the further prediction model 48 (see Fig. 3), wherein the further prediction model 48 particularly preferably also comprises a neural network that was trained according to steps analogous to the prediction model 44, wherein relevant parameters for training the further prediction model are used to determine the quality of the cutting surface. In particular, within the scope of step 126, feedback on the quality of the cutting surface 46 can be provided, for example, to the machine control system 36.

[0058] Preferably, the machine control system 36 provides the feedback to a user of the method. In a fourteenth sub-step 128, the control parameters are adjusted to the readjustment values. The aforementioned sub-steps from the sixth sub-step 112 to the fourteenth sub-step 128 are repeated for each specified separation step until all specified separation steps have been completed. In a fifteenth sub-step 130, the separated finished part is made available for further processing or for its intended use after the specified separation steps have been performed.

[0059] After performing the predetermined separation steps, the following additional sub-steps can optionally be performed to achieve a separation of additional workpieces that better matches the target specifications (in the form of the target values ​​of the workpiece parameters): The parameter values ​​of the workpiece parameters of the workpiece 14 are determined according to a sixteenth sub-step 132 after the predetermined separation steps have been performed on the workpiece. Subsequently, according to a seventeenth sub-step 134, the difference between the parameter values ​​of the workpiece parameters determined in the sixteenth sub-step 132 and the target values ​​of the workpiece parameters is calculated.

[0060] Thereafter, according to an eighteenth sub-step 136, readjustment values ​​of the control parameters of the production means 18 are determined by the forecast model 44 in such a way that, within the scope of the calculation by the forecast model 44, a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters of the workpiece 14 is minimized.

[0061] When separating another workpiece (not shown), the above-mentioned sub-steps from the sixth sub-step 112 to the fourteenth sub-step 128 are carried out for each predetermined separation step (cf. the third sub-step 106). In this case, the additional workpiece is used instead of the workpiece 14. At the beginning of the processing of the predetermined separation steps in the sixth sub-step 112, the readjustment values ​​of the control parameters are stored in the machine control 36, which were calculated as described above in the eighteenth sub-step 136 after the separation of the workpiece 14 by the prediction model 44. The prediction model 44 can be trained by the following steps (not shown), in particular before carrying out the predetermined separation steps for separating the workpiece 14: First, a test workpiece (not shown) is provided, wherein first parameter values ​​of the workpiece parameters of the test workpiece are determined.In addition, initial parameter values ​​of the control parameters of the tool 18 are determined before the subsequent separation of the workpiece 14. Weights of the prediction model 44 in the form of a neural network are initialized. The test workpiece is then separated in such a way that the first parameter values ​​of the workpiece parameters are changed to second parameter values ​​of the workpiece parameters. The second parameter values ​​can be predefined. Alternatively or additionally, the second parameter values ​​of the workpiece parameters can be measured after separation. After the test workpiece is separated, the parameter values ​​of the control parameters are measured.

[0062] The prediction model 44 predicts parameter values ​​of the control parameters, which are used, within the scope of the calculation by the prediction model, to separate the test workpiece by changing the first parameter values ​​to the second parameter values ​​of the workpiece parameters. In particular, the initial parameter values ​​of the control parameters of the production tool 18 are included in the prediction. A difference is then determined between the parameter values ​​of the control parameters predicted by the prediction model 44 and the parameter values ​​of the control parameters measured after the separation of the workpiece 14. The weights of the prediction model are updated to minimize this difference.The steps of predicting the control parameter values, calculating the difference between the predicted and measured control parameter values, and updating the weights of the prediction model are repeated until the difference reaches or falls below a specified minimization limit. If necessary, further separation steps can be performed on the test workpiece, and the weights of the prediction model can be updated according to the aforementioned procedure after each separation step.

[0063] The substeps from the sixth substep 112 to the thirteenth substep 126 form a two-stage control system 50 of the separation process, with neural networks being used in particular for this control system. The substeps from the sixth substep 112 to the fourteenth substep 128 form an adaptive part 52 of the separation process. The substeps from the sixth substep 112 to the eighteenth substep 136 form a quality control loop 54 of the separation process.

[0064] Alternatively or in addition to at least some of the aforementioned steps for training the prediction model 44, in particular the cutting of the test workpiece (see step iii above) and the measuring of the parameter values ​​of the control parameters (see step iv above), a simulation can be carried out which simulates the cutting of the test workpiece and mathematically determines the associated parameter values. In such simulations, disruptive factors in the cutting process can be taken into account, e.g. sheet thickness deviations, strength fluctuations of the material, wear conditions of the production equipment used. The simulations can be carried out in particular as part of a sensitivity analysis in order to preferably create a broad database and to predict the optimal control parameters for possible disruptive factors in the cutting process.Such simulations can also be used to determine different characteristics of the cutting surfaces as a result of disturbance factors in the process that are generated during the process.

[0065] Taking a summary of all figures of the drawing, the invention relates to a method 100 for readjusting a cutting process. Within the scope of the method 100, a workpiece 14 and workpiece parameters are provided, wherein the workpiece parameters represent properties of the workpiece 14 relevant to the cutting process. In addition, a production tool 18 for cutting the workpiece 14 and parameter values ​​of control parameters for controlling the production tool 18 are provided. During the execution of a cutting step from a group of predetermined cutting steps, the cutting forces and / or strokes occurring, in particular of the production tool 18, are measured. From the measured cutting forces and / or strokes, actual parameter values ​​of the workpiece parameters are determined by an evaluation model.

[0066] Actual parameter values ​​of the workpiece parameters are transferred to a forecast model 44. The forecast model 44 determines readjustment values ​​of the control parameters in order to minimize the difference between the actual parameter values ​​and the target values ​​of the workpiece parameters when cutting the workpiece 14. The control parameters are set to the readjustment values. The process steps from controlling the tool 18 with the readjusted control parameters for cutting the workpiece 14 to readjusting the control parameters are repeated until the specified cutting steps have been completed.

[0067] List of reference symbols

[0068] Separation system

[0069] Floor

[0070] workpiece

[0071] Transport rollers

[0072] Manufacturing equipment

[0073] punching stamp

[0074] Through recess vertically aligned beams horizontal beams

[0075] lifting rod

[0076] sensor

[0077] die

[0078] recess

[0079] Machine control

[0080] Frame

[0081] Evaluation model

[0082] Forecast model

[0083] Interface further forecast model two-stage control adaptive part

[0084] Quality control loop of the inventive method - 136 sub-steps of the inventive method

Claims

Patent claims 1 . Method (100) for readjusting a cutting process and / or forming process, comprising the steps: I. Providing a workpiece (14); II. Providing a manufacturing means (18) for separating the workpiece (14); III. Providing separation steps for separating the workpiece (14); IV. Providing parameter values ​​of control parameters for controlling the production means (18); V. Providing workpiece parameters for characterizing the workpiece (14) and specifying target values ​​of the workpiece parameters that the workpiece (14) should have after separation; VI. Storing the parameter values ​​of the control parameters in a machine control system (36); VII. Controlling the production means (18) by the machine control (36) with the aid of the control parameters to carry out a separation step from the provided separation steps; VIII. Measuring the separation forces and / or strokes occurring during the separation step; IX. Evaluation of the values ​​of the separation forces and / or strokes measured in step VIII by an evaluation model (42) in order to determine actual parameter values ​​of the workpiece parameters for characterizing the workpiece (14); X. Determining readjustment values ​​of the control parameters with the aid of the actual parameter values ​​of the workpiece parameters determined in step IX by a prognosis model (44), so that within the scope of the calculation by the prognosis model (44) with the readjustment values ​​of the control parameters when separating the workpiece (14) a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters of the workpiece (14) are minimized, the difference being determined after performing the provided separation steps; XI. Readjustment of the control parameters to the readjustment values; XII. Repeat steps VI to XI until the provided separation steps are completed.

2. The method according to claim 1, wherein after step XII the following steps are carried out: XIII. Measuring the actual parameter values ​​of the workpiece parameters of the workpiece (14); XIV. Determining differences between the measured actual parameter values ​​of the workpiece parameters and the target values ​​of the workpiece parameters; XV. Determining readjustment values ​​of the control parameters with the aid of the differences determined in step XIV by the prognostic model, so that, as part of the calculation by the prognostic model with the readjustment values ​​of the control parameters during separation according to the provided separation steps, a difference between the target values ​​of the workpiece parameters and the actual parameter values ​​of the workpiece parameters is minimized, wherein the difference is determined after the provided separation steps have been carried out; XVI. Readjustment of the control parameters to the readjustment values; XVII. Carrying out steps I to XII, wherein in step I a further workpiece is provided and in step IV the control parameters adjusted to the adjustment values ​​in step XVI are provided.

3. The method according to any one of the preceding claims, wherein the workpiece parameters comprise flatness parameters to characterize the flatness of the workpiece (14), wherein the target values ​​of the flatness parameters comprise desired parameter values ​​of the flatness parameters to to achieve the desired flatness of the workpiece (14) after performing the provided separation steps.

4. Method according to one of the preceding claims, wherein the forecast model (44) comprises a neural network, in particular a multi-layer perceptron.

5. The method according to claim 4, wherein the prediction model is trained in the following steps: i. determining first parameter values ​​of the workpiece parameters of a test workpiece and determining parameter values ​​of the control parameters before performing the subsequent step ii; ii. initializing weights of the prediction model (44); iii. separating the test workpiece with the production tool (18) while changing the first parameter values ​​to second parameter values ​​of the workpiece parameters; iv. measuring the parameter values ​​of the control parameters after performing step iii; v. calculating predicted parameter values ​​of the control parameters for separating the test workpiece while changing the first parameter values ​​to the second parameter values ​​of the workpiece parameters by the prediction model (44); vi. calculating a difference between the predicted parameter values ​​of the control parameters and the parameter values ​​of the control parameters measured in step iv; vii.Updating the weights of the forecast model (44) to minimize the difference according to step vi; viii. Repeating steps v to vii until the error according to step vi reaches a predetermined minimization limit.

6. Method according to one of the preceding claims, wherein the Evaluation model (42) an algorithm for evaluating the separation of the workpiece, a decision tree, a decision forest and / or a neural network in order to classify the workpiece (14) into a workpiece class which belongs to predetermined workpiece classes and / or to classify the production means (18) into a production means class which belongs to predetermined production means classes.

7. Method according to one of the preceding claims, wherein the evaluation model (42) is selected from a group of evaluation models (42), in particular on the basis of the parameter values ​​of the control parameters provided in step III and / or on the basis of the parameter values ​​of the workpiece parameters provided in step IV.

8. Method according to one of the preceding claims, wherein before step X the prognosis model (44) is selected from a group of prognosis models (44), in particular on the basis of the parameter values ​​of the control parameters provided in step III and / or on the basis of the parameter values ​​of the workpiece parameters provided in step IV.

9. Method according to one of the preceding claims, wherein the machine control (36) compares the parameter values ​​determined in step IX with target values ​​of the control parameters and / or workpiece parameters before step X.

10. Method according to one of the preceding claims, wherein the control parameters comprise a force with which the production means (18) acts on the workpiece (14), geometric parameters of the production means (18), and / or path parameters of a path on which the production means (18) is guided.

11. Method according to one of the preceding claims, wherein the workpiece parameters are material parameters of a material of the workpiece (14), Geometric parameters of the workpiece (14), in particular a thickness of the workpiece (14) and / or a shear strength of the workpiece (14).

12. The method according to any one of the preceding claims, wherein the workpiece parameters comprise cutting surface parameters to characterize the quality of a cutting surface (46), wherein the target values ​​comprise desired parameter values ​​of the cutting surface parameters for a desired quality of the cutting surface (46), wherein in step X, readjustment values ​​of the control parameters are determined with the aid of the parameter values ​​of the workpiece parameters determined in step IX by the prognosis model or a further prognosis model in such a way that a difference between the target values ​​of the cutting surface parameters and the parameter values ​​of the cutting surface parameters of the workpiece (14) is minimized after carrying out the provided separation steps.

13. Method according to one of the preceding claims, wherein the Separation process and / or forming process is designed as a punching process.

Images