Device and method for preprocessing state variables detected during stamping process

Abstract

The present invention relates to a device for determining the state of a stamping process, the device being designed to: obtain first data, second data and third data; remove the start-up data obtained in the start-up cycle of a stamping tool from the first data and the second data; interpolate the first data and the second data so that the second data obtains a uniform length and step size relative to the first data for multiple cycles of the stamping process; identify and remove anomalies in the first data and the second data; determine and compensate for thermal disturbances of the first data and the second data by the third data; and determine the state of the stamping process based on the first data and the second data compensated for the thermal disturbances.

Description

Technical Field

[0001] The invention relates to a device and a method for determining the state of a punching process, in particular by preprocessing state variables detected during the punching process. Background Art

[0002] With the help of piezoelectric sensors, temperature sensors and rotary encoders, it is possible to determine the state variables of the punching tool. However, it has been shown that the detected data have a very high variance. Due to the high variance, the direct use of the data to determine the state of the punching tool is only feasible to a limited extent. Summary of the invention

[0003] The object of the present invention is to achieve an improved determination of state variables of a punching process and an improved design for the state determination of a punching process, in which the above-mentioned disadvantages do not occur.

[0004] The invention is based on the idea of ​​using various data preprocessing steps when determining the state variables, including: 1. removing data generated during machine startup (Inbetriebnahme); 2. interpolating the data to ensure a uniform length of the time series; 3. automatically detecting and removing anomalies; and 4. compensating for disturbances.

[0005] With the aid of suitable sensors, state variables relevant for the implementation of "predictive maintenance" are recorded. State variables are, for example, the structure-borne noise of the punching tool. The temperature of the punching tool and information about the eccentric angle of an eccentric press or the stroke of a hydraulic press. For example, several piezoelectric sensors are placed in the punching tool to detect structure-borne noise. The temperature sensor is located in the immediate vicinity. The angle can be recorded with the aid of a rotary encoder. After the data is collected, all data generated during the "start of punching" or during the start-up of the punching tool can be discarded, as this data does not contain any information about the actual production.

[0006] Experiments have shown that the measurement of structure-borne sound has a strong temperature dependency. Anomalies are automatically identified and removed beforehand in order to be able to determine this dependency as accurately as possible. The cleaned data set can now serve as the basis for determining a family of compensation functions that describe the temperature dependency of the structure-borne sound signal. Thermal disturbances are so pronounced that wear characteristics are not visible. However, since the wear state of the punching tool is to be determined, thermal disturbances must be compensated. A family of compensation functions can serve as the basis for this.

[0007] According to the first aspect of the present invention, the above technical problem is solved by a device for determining the state of a stamping process, wherein in the stamping process, a stamping tool for stamping a material is periodically moved from a starting position to an ending position, and the device is designed to obtain first data, wherein the first data represents the force applied by the stamping tool to the material when the stamping tool moves from the starting position to the ending position within multiple cycles of the stamping process; obtain second data, wherein the second data represents the angle or stroke traversed by the stamping tool when it moves from the starting position to the ending position within multiple cycles of the stamping process; obtain third data, wherein the third data indicates the temperature of the stamping tool within multiple cycles of the stamping process; remove the startup data obtained during the startup cycle of the stamping tool from the first data and the second data; interpolate the first data and the second data so as to obtain a uniform length and step size of the second data relative to the first data for multiple cycles of the stamping process; identify and remove anomalies in the first data and the second data; determine and compensate for thermal disturbances of the third data to the first data and the second data; and determine the state of the stamping process based on the first data and the second data compensated for the thermal disturbances.

[0008] During the stamping process, the stamping tool moves from the starting position to the end position, and the end position corresponds to the starting position, because this is a cyclic process. Figure 4 The starting position, end position and various intermediate positions are further illustrated in the example of an eccentric press in FIG.

[0009] When the punching tool moves from the starting position to the end position, the force it exerts on the material is not measured directly but indirectly. The structure-borne sound signal is detected by a piezoelectric sensor, which provides information about the force and is referred to below as pseudo force.

[0010] The combination of the first and second data gives an indication about the mechanical work which is recorded in the form of vibrations and elastic deformations on the individual plates of the punching tool. Figure 3 An exemplary construction of a punching tool is shown. Figure 3 A more detailed description is given.

[0011] The apparatus disclosed herein can improve the determination of state variables of a stamping process and improve the determination of the state of a stamping process.

[0012] Due to data preprocessing, the device disclosed here can determine the state of the tool and thus create a basis for dynamically adjusting the maintenance intervals. The advantage here is that unintentional exceeding of load limits can be prevented. It is possible to optimize maintenance intervals and reduce the associated maintenance costs.

[0013] Furthermore, wear characteristics become visible and more in-depth cause studies are possible. Unknown design weaknesses, systematic errors or processing faults can then be investigated and conclusions drawn.

[0014] According to an exemplary embodiment of the device, the first data represents the force acting on the punching tool due to the punching process by means of a bending or deformation of the punching tool. This force is also referred to as a pseudo force below.

[0015] This offers the technical advantage that the punching force can be displayed in a simple manner, namely by measuring the bending or deformation on the punching tool, for example by means of a piezoelectric sensor connected to the punching tool.

[0016] According to an exemplary embodiment of the apparatus, the first data are sensor data from a piezoelectric sensor placed in the punching tool.

[0017] This achieves a technical advantage, namely that the forces referred to here as "pseudo forces" can be determined simply and effectively by means of piezoelectric sensors, which, for example, provide an indication of the mechanical work acting on the punching tool. In addition, many punching tools already have such piezoelectric sensors, so that the data generated with them can be used for further evaluation without great effort.

[0018] According to an exemplary embodiment of the device, the second data includes angle information of a rotary encoder of the punching tool, in particular angle information of the eccentricity of an eccentric press; or the second data includes stroke information of a stroke encoder of the punching tool, in particular stroke information of a piston of a hydraulic press.

[0019] This has the technical advantage that the mechanical work acting on the punching tool can be determined in a simple manner via the angle information or the stroke information and conclusions can be drawn about the wear behavior of the punching tool.

[0020] According to an exemplary embodiment of the device, the first data and the second data include a force-angle curve, and the device is designed to determine the start-up data based on the deviation between the shape of the force-angle curve during the punch tool start-up cycle and the shape of the force-angle curve after the punch tool start-up cycle.

[0021] This achieves the technical advantage that start-up data of a punching process which are unsuitable for evaluating the punching process can be filtered out in an efficient manner.

[0022] According to an exemplary embodiment of the device, the device is designed to determine the activation data as a function of a shift in the force direction of the force-angle curve during an activation cycle of the punching tool.

[0023] This achieves the technical advantage that such a shift in the force-angle curve provides a conclusion as to whether the data are startup data or data during normal production operation. Startup data can thus be identified and removed in an appropriate manner so as not to interfere with the state determination.

[0024] According to an exemplary embodiment of the device, the device is designed to remove from the first data and the second data a force-angle curve which is generated when the number of strokes of the punching process is lower than a fixed percentage of a rated number of strokes of the punching process.

[0025] This achieves the technical advantage that the start-up data can be easily distinguished from normal production data by means of the stroke count of such a punching process and can be filtered out so that it does not have a negative influence on the state determination.

[0026] According to an exemplary embodiment of the device, the device is designed to extend the first data and the second data by one or more zero terms for each detection cycle of the punching process until the second data has a uniform length and step size relative to the first data.

[0027] This achieves a technical advantage in that different series of measurements refer to a uniform basis or scale and can therefore be compared with each other.

[0028] According to an exemplary embodiment of the device, the first data and the second data include a force-angle curve, the device is designed to generate a reference curve from the force-angle curve by averaging, and determine anomalies based on deviations of the force-angle curve relative to the reference curve.

[0029] This achieves the technical advantage that anomalies in the measurement data can be easily determined and filtered out so that they do not negatively affect the status determination.

[0030] According to an exemplary embodiment of the device, the above-mentioned deviation is based on the product of a standard deviation of the force-angle curve and an adjustable factor.

[0031] This achieves the technical advantage that the identification of anomalies can be adapted to the corresponding database to avoid too many or too few anomalies being identified. Categorizing the anomalies enables an improved determination of the state of the punching process, since unusual values ​​are not taken into account.

[0032] According to an exemplary embodiment of the device, the first data and the second data are divided into different channels, wherein each channel comprises the first and second data recorded with a corresponding piezoelectric sensor on the punching tool.

[0033] The technical advantage of this is that, through the measurement of the various channels, a more precise statement about the punching process can be made, since more data is available. Therefore, the state determination can be made more precisely.

[0034] According to an exemplary embodiment of the device, the device is also designed to, when it is identified that the force-angle curve of one of the different channels is abnormal, also mark the corresponding force-angle curves of other channels as abnormal.

[0035] This provides the technical advantage that anomalies can be identified more accurately. For example, an anomaly may be obvious in one channel but less obvious in another channel. The anomaly in the less obvious channel can then be found more easily and the status determination can be made more accurately.

[0036] According to an exemplary embodiment of the device, the device is designed to determine a compensation function for a thermal disturbance of the third data to the first data and the second data.

[0037] This achieves a technical advantage that the compensation function can accurately describe the thermal disturbance of the third data to the first and second data, so that the first and second data without the thermal disturbance can more accurately determine the state of the stamping process.

[0038] According to an exemplary embodiment of the device, the compensation function is based on a model that specifies deviations of the first data depending on the third data, wherein the model is based on a mathematical approximation function and / or on a prediction using artificial intelligence. This achieves the technical advantage that thermal disturbances can be accurately determined and compensated.

[0039] According to an exemplary embodiment of the apparatus, the compensation function is a second order polynomial function or e function, which can be expressed as:

[0040]

[0041]

[0042] in, represents the temperature and the parameters a, b, c and d are predetermined.

[0043] This achieves the technical advantage that the compensation function can be easily calculated but at the same time accurately describes the thermal perturbations.

[0044] According to an exemplary embodiment of the device, the device is designed to determine a compensation function for each individual angle and to compensate the third data for thermal disturbances of the first data and the second data for each individual angle.

[0045] This achieves the technical advantage that thermal disturbances which are dependent on the angle of the punching tool can thereby be determined very accurately and the first and second data can be effectively freed of thermal disturbances.

[0046] According to the second aspect of the present invention, the above technical problem is solved by a method for determining the state of a stamping process, wherein a stamping tool used to stamp a material is periodically moved from a starting position to an ending position, and the method comprises the following steps: obtaining first data, the first data representing the force applied to the material when the stamping tool moves from the starting position to the ending position within multiple cycles of the stamping process; obtaining second data, the second data representing the angle or stroke traveled by the stamping tool when it moves from the starting position to the ending position within multiple cycles of the stamping process; obtaining third data, the third data representing the temperature of the stamping tool within multiple cycles of the stamping process; removing the startup data obtained during the startup cycle of the stamping tool from the first data and the second data; interpolating the first data and the second data so as to obtain a uniform length and step size of the second data relative to the first data for multiple cycles of the stamping process; identifying and removing anomalies in the first data and the second data; determining and compensating for thermal disturbances of the third data to the first data and the second data; and determining the state of the stamping process based on the first data and the second data compensated for the thermal disturbances.

[0047] By means of such a method, the state variables of the stamping process can be better determined, thereby more accurately determining the state of the stamping process.

[0048] Due to data preprocessing, the method described here makes it possible to determine the tool status and thus create a basis for dynamically adjusting the maintenance intervals. The advantage here is that unintentional exceeding of load limits can be prevented. As a result, it is possible to optimize the maintenance intervals and reduce the associated repair costs.

[0049] Furthermore, with the method described here, wear characteristics become visible and a more in-depth investigation of the causes is possible. Conclusions about unknown design weaknesses, systematic failures or process errors can then be investigated. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Hereinafter, the present invention will be described in more detail with reference to embodiments and accompanying drawings.

[0051] Figure 1 is a schematic diagram of a device 100 for determining a stamping process state according to the present invention.

[0052] Figure 2 a / b / c are three exemplary illustrations of detected force-angle curves 200a, 200b, 200c for three channels of a punching process.

[0053] Figure 3a / b / c are three exemplary illustrations of force-angle curves 300a, 300b, 300c with the actuation data removed for three passes of a punching process.

[0054] Figure 4 is an exemplary illustration of the variation of pseudo force 400 over time measured on two different dates without compensating for thermal disturbances.

[0055] Figure 5 a / b is an exemplary illustration of the variation of pseudo force (500a without smoothing filter or 500b with smoothing filter) over time measured over several months with compensation for thermal disturbances.

[0056] Figure 6 a / b are two exemplary illustrations of force-angle curves 600a, 600b after preprocessing the detected state variables according to the present invention.

[0057] Figure 7 An exemplary illustration of states 700a, 700b, 700c of a punching process is shown using an eccentric press as an example.

[0058] Figure 8 is an exemplary illustration of a punching tool 800 according to the present invention. And

[0059] Fig. 9 is a schematic diagram of a method 900 for determining a stamping process state according to the present invention. DETAILED DESCRIPTION

[0060] In the following detailed description, reference is made to the accompanying drawings which form a part of this article, in which specific embodiments in which the present invention can be implemented are shown by way of illustration. It is to be understood that other embodiments may be used without departing from the concept of the present invention, and structural or logical changes may also be made. Therefore, the following detailed description should not be construed as a restrictive sense. It is to be further understood that, unless otherwise specified, the features of the various embodiments described herein may be combined.

[0061] These aspects and embodiments will be described with reference to the accompanying drawings, wherein similar reference symbols generally refer to similar elements. In the following description, in order to provide a deeper understanding of one or more aspects of the present invention, many specific details are listed to achieve the purpose of explanation. However, for those skilled in the art, one or more aspects or embodiments may be embodied with fewer specific details. In other cases, known structures and elements are shown in schematic form to facilitate the description of one or more aspects or embodiments. It is understood that other embodiments may be used without departing from the concept of the present invention, and structural or logical changes may also be made.

[0062] In the following description, reference is made to the punching process and the punching tool. Punching is a cutting process that produces stamped parts from various materials such as sheet metal, cardboard, textiles, etc., using a punch or striking a cutting tool with a hammer. The punching tool consists of a punch that displays the internal shape and a die with a corresponding matching opening. The punch is usually set in an upper / lower part. Figure 8 An example of a punching tool is shown. In high-performance punching, processes such as welding, flanging, riveting and forming are integrated into special progressive dies. In addition to the punching station, such progressive dies often have bending stations and other processing stations. Some of these tools are very complex and must be used and protected efficiently.

[0063] Figure 1 1 shows a schematic diagram of a device 100 for determining a state of a stamping process according to the present invention. The device 100 is suitable for determining a state of a stamping process, wherein: Figure 7 and 8 As shown, the punching tool 730, 800 periodically moves from the starting position 710 to the ending position 710 (see Figure 7 ) in order to stamp the material.

[0064] The device 100 is designed to obtain first data 101, which represents the force applied by the punching tool to the material when the punching tool moves from a starting position to an end position in a plurality of cycles of the punching process. This force, referred to herein as a pseudo force, is the result of vibrations and elastic deflections acting on the punching tool during the punching process.

[0065] The device 100 is designed to obtain second data 102, which represents the angle or stroke covered by the punching tool when it moves from the starting position to the ending position during a plurality of cycles of the punching process.

[0066] The device 100 is designed to obtain third data 103 which represent the temperature of the punching tool during a plurality of cycles of the punching process.

[0067] The device 100 is designed to remove 110 the start-up data detected during a start-up cycle of the punching tool from the first data 101 and the second data 102 .

[0068] The apparatus 100 is designed to interpolate 112 the first data 101 and the second data 102 to obtain a uniform length and step size of the second data 102 relative to the first data 101 for a plurality of cycles of the punching process.

[0069] The apparatus 100 is designed for identifying and removing 114 anomalies in the first data 101 and the second data 102 .

[0070] The device 100 is designed to determine and compensate 116 thermal disturbances of the third data 103 to the first data 101 and the second data 102 .

[0071] The device 100 is designed to determine 116 the state of a stamping process based on first data 101 and second data 102 that compensate for thermal disturbances.

[0072] For example, the device may include an electronic circuit or a processor or a controller to perform the functions given above. Here, it may be a digital controller for controlling the punching process, or a separate processor or computer on which the above functions may be performed.

[0073] The determination of the punching process state may involve, for example, the wear state of the punching tool, based on which a maintenance point in time for the punching tool can be determined.

[0074] For example, the first data 101 represent the force acting on the punching tool 730 , 800 due to the punching process by means of a bending or deformation of the punching tool 730 , 800 .

[0075] For example, the first data 101 may be sensor data from a piezoelectric sensor placed in the punching tool 730 , 800 .

[0076] like Figure 6 and Figure 7 As shown, the second data 102 may include angle information 602, 701 of the rotary encoder of the punching tool, in particular, eccentric angle information 701 of the eccentric press 700a, 700b, 700c. Alternatively, the second data 102 may include stroke information 702 of the stroke sensor of the punching tool, in particular, stroke information of the piston of the hydraulic press.

[0077] like Figure 2 As shown, the first data 101 and the second data 102 may include force-angle curves, such as force-angle curves 200a, 200b, and 200c. Figure 2 As shown in a / b / c, the device 100 can be designed to determine the starting data based on the deviation between the shape of the force-angle curve 200a, 200b, 200c during the punch tool starting cycle and the shape of the force-angle curve 200a, 200b, 200c after the punch tool starting cycle.

[0078] The device 100 may be designed to determine the activation data based on the shift of the force-angle curves 200a, 200b, 200c in the direction of the force during the activation cycle of the punch tool, such as Figure 2 As shown in a / b / c.

[0079] The apparatus 100 may be designed to remove the force-angle curves 200a, 200b, 200c generated when the number of strokes of the punching process is lower than a fixed percentage of the rated number of strokes of the punching process from the first data 101 and the second data 102.

[0080] The apparatus 100 may be designed to extend the first data 101 and the second data 102 by one or more zero terms for each detection cycle of the punching process until the second data 102 has a uniform length and step size relative to the first data 101 .

[0081] The first data 101 and the second data 102 may include force-angle curves 300a, 300b, 300c, for example Figure 3 The device 100 may be designed to generate a reference curve from the force-angle curves 300a, 300b, 300c by averaging, and determine anomalies based on deviations of the force-angle curves 300a, 300b, 300c from the reference curve.

[0082] For example, the deviation may be based on the product of the standard deviation of the force-angle curves 300a, 300b, 300c and an adjustable factor, such as Figure 3 Described in more detail.

[0083] The first data 101 and the second data 102 may be divided into different channels, wherein each channel may include the first data 101 and the second data 102 recorded with a corresponding piezoelectric sensor on the punching tool.

[0084] like Figure 3 As shown in a / b / c, the device 100 can be designed so that when an abnormality is identified in the force-angle curve 300a, 300b, 300c of one of the different channels, the corresponding force-angle curve 300a, 300b, 300c of the other channels is also marked as abnormal.

[0085] The apparatus 100 may be designed to determine a compensation function for the thermal disturbance 400 of the third data 103 to the first data 101 and the second data 102, such as Figure 4 described in more detail.

[0086] The compensation function may be, for example, a second order polynomial function or an e-function, which may be expressed as:

[0087]

[0088]

[0089] If targeted Figure 4 Further described, wherein, represents the temperature and the parameters a, b, c and d are predetermined.

[0090] If targeted Figure 4 As further described, the apparatus 100 may be designed to determine a compensation function for each individual angle, and compensate the third data 103 for the thermal disturbance 400 of the first data 101 and the second data 102 for each individual angle.

[0091] Figure 2 a / b / c show three exemplary illustrations of detected force-angle curves 200a, 200b, 200c for three channels of a punching process. Figure 2 a / b / c show initially detected force-angle curves 200a, 200b, 200c, which correspond to Figure 1 The first data 101 and the second data 102. Figure 1 Description, the removal of the startup data 110 and Figure 2 The force-angle curves 200a, 200b, 200c shown as a / b / c are related.

[0092] exist Figure 2 In the example of a / b / c, three piezoelectric sensors are used. Therefore, the force-angle curve is provided through three channels. The force-angle curve 200a represents the first channel, the force-angle curve 200b represents the second channel, and the force-angle curve 200c represents the third channel.

[0093] As can be seen from the force-angle curves 200a, 200b, 200c shown, the raw data contains many curves that do not represent production defects or wear characteristics. During the start-up of the punching tool (start of the punching process), the machine operator manually operates the machine and can check from station to station whether the metal strip is in the correct position and whether there are any foreign objects in the tool. The measurement results recorded at this stage can be found in Figure 2 a by X-movement (with Figure 2 b and 200c) and different shapes. For example, the ridge or dome in the middle of measurement curve 200a is more obvious than in measurement curves 200b and 200c. Removing these data constitutes Figure 1 The data cleaning or removal of the startup data 110 is the first step.

[0094] Additionally, all curves generated when the number of punch tool strokes is less than a fixed percentage of the nominal number of punch tool strokes may be removed from the data set.

[0095] In order to be able to apply mathematical methods, the measurement sequences 200a, 200b, 200c can be interpreted as matrices. However, this is only possible if the measurement sequences have the same length. A remedy is to interpolate 112 the data, as described above for Figure 1 Description.

[0096] For this purpose, the measurement sequences 200a, 200b, 200c are first extended by several zero entries so that all measurement sequences can be interpolated between fixed defined limits. Subsequent interpolation 112 gives the measurement sequences a uniform length and step size. Thus, even without angle information, measurement sequences of structure-borne sound values ​​can be compared with one another.

[0097] Figure 3 a / b / c show three exemplary representations of force-angle curves 300a, 300b, 300c of three channels of a punching process after the start data have been removed.

[0098] In addition to machine learning algorithms such as local outlier factors or one-class support vector machines and deep learning algorithms such as autoencoders, simpler mathematical methods can also be used to automatically identify anomalies. For example, for the data sets 300a, 300b, 300c, several reference curves are generated by averaging. If the difference between the force-angle curve 300a, 300b, 300c and the respective reference curve is greater than a predetermined limit, an anomaly can be identified. For example, the limit is determined by the standard deviation and the coefficients to be determined This factor can be adjusted by an expert until the desired result is achieved and all production errors can be reliably detected.

[0099] For example, the median is a better reference value than the arithmetic mean because anomalies have a much smaller impact on the median.

[0100] If multiple structure-borne noise signals are considered, they can be represented by a matrix as follows:

[0101]

[0102] in,

[0103] and

[0104] The number of structure-borne sound signals is denoted by m and the number of measuring points is denoted by n. The reference signal is also represented by a row vector whose entries can be calculated as the median of the rows of X.

[0105] The formula for identifying anomalies can be given as follows, for example:

[0106]

[0107]

[0108] For example, when three piezoelectric sensors are used, a stroke process can be defined by three force-angle curves. Therefore, once the force-angle curve of one channel is determined to be abnormal, the relevant force-angle curves of other channels can also be marked as abnormal.

[0109] Figure 4 An exemplary graph of a pseudo force 400 over time measured 401 , 402 without compensation for thermal disturbances on two different dates is shown.

[0110] As above Figure 2 and Figure 3 As shown, the measurement of the pseudo force 400 shows that the force-angle curve is strongly thermally perturbed. Figure 4 The graph 400 in FIG. 4 shows the course of the maximum values ​​of the force-angle curve. Contrary to what would be expected, the values ​​do not increase over time. In order to make the wear visible, Figure 4 It can be concluded that compensation for disturbance variables is necessary. This will be explained below.

[0111] As mentioned earlier, Figure 4 An exemplary diagram of the influence of temperature when determining the force-angle curve is shown, as shown above. Figure 2 and Figure 3 Graph 400 shows the variation of the measurement signal of a pseudo force or piezoelectric sensor over time. The graph shows two measurement curves 401 and 402, the first measurement curve 401 being recorded on the first day and the second measurement curve 402 being recorded on the second day.

[0112] It is noteworthy that the curves 401, 402 decrease over time. If there is a break between two measurement points, the maximum value will increase compared to the previous measurement point. It is assumed that the curve is mainly affected by the temperature of the punching tool.

[0113] from Figure 4 It can be concluded from the diagram that in order to make fatigue of the punching tool visible, it is imperative to compensate for temperature-dependent influences.

[0114] For this purpose, a function can be used which takes into account the heating of the punching tool on the basis of the presence of a large number of measurements during the heating process.

[0115] Figure 1 The device 100 shown takes into account the temperature behavior of the stamping process. The device 100 uses a function that compensates for the temperature influence during the stamping process.

[0116] To this end, two generalized equations can be considered to determine the compensation function for thermal effects. The first is a second-order polynomial function, and the second is an e function.

[0117]

[0118]

[0119] Both functions are suitable for approximating the compensation function. Once the compensation function is determined, the compensation for thermal disturbances can be determined as follows:

[0120]

[0121]

[0122] The thermal disturbance compensation determined in this way can be used to determine the state of the stamping process 118, as described above. Figure 1 Determination of the wear characteristics or fatigue of punching tools, for example.

[0123] Figure 5 a / b show exemplary representations of pseudo forces measured over time (500a without smoothing filter, 500b with smoothing filter) with compensation for thermal disturbances over several months.

[0124] In Figures 500a and 500b, thermal disturbances have been compensated, for example, by Figure 4 Describe the mechanism.

[0125] exist Figure 5 In a, the maximum values ​​of the first prominent peak are shown in chronological order. The considered data points are located between angles of 166 and 174 degrees. A decreasing trend in the values ​​is evident with increasing wear. After maintenance, the maximum values ​​rise moderately again.

[0126] exist Figure 5 b, yes Figure 5 The data in a have smoothing filtering applied. Figure 5 The maximum values ​​of a are smoothed using a moving average, with 50 values ​​being summed up in each case.

[0127] Figure 6 a / b show two exemplary diagrams of force-angle curves 600a, 600b after preprocessing of the detected state variables according to the present invention. Figures 1 to 5 The description is pre-processed by removing startup data 110, interpolating 112, removing anomalies 114 and thermal compensation 116.

[0128] Figure 600a shows the angle data corresponding to the detected angle data (corresponding to Figure 1 The detected force data (corresponding to the second data 102) pre-processed in Figure 1 600b shows the state of the punching process which can be determined thereby.

[0129] The two graphs 600a and 600b show a family of curves for two days, respectively, where the first day (November 2, 2020) is the beginning of the measurement period and the last day (February 15, 2020) is the end of the measurement period. Therefore, the measurement period lasted for several months. Figure 6 In Figure 600a of a, the force-angle curves of the first day (November 2, 2020) are shown as a first curve family 611, and the force-angle curves of the last day (February 15, 2020) are shown as a second curve family 612. Due to obvious wear, the measurement was stopped on the last day (February 15, 2020). Comparing the two curve families 611 and 612, it can be seen that the first prominent peak 614 in the second curve family 612 shows the change in shape and performance.

[0130] The lower graph 600b shows the same first family of curves 611 from the first day (November 2, 2020) and the third family of curves 613 from the last day of measurement (February 15, 2020), but this was recorded only after the successful maintenance. It can be seen here that the first prominent peak 614 is visible again after the maintenance. Therefore, the reduction of the peak 614 is a phenomenon caused by wear.

[0131] In addition, the false force 601 and the eccentric angles 602, 701 can also be detected by the corresponding sensor system, such as Figure 7 As the angle 701 is related to the distance travelled by the upper part of the tool, the area of ​​the force-angle curve contains information about the mechanical work required during punching. Assuming that the mechanical work required W is due to the blunting of the cutting element mech As wear continues, it increases. This relationship can be determined by measuring the forces acting on the cutting element. The tool is maintained and the cutting element is ground in this way, so that the mechanical work required to cut out the punched part is reduced.

[0132] Until now, tools have been maintained at static intervals. Unpredictable events such as burrs (i.e. parts detaching from the sheet), operating errors or breaks in the lubricant film can have a negative impact on the tool's service life, so that the tool's physical limits can be reached within the prescribed maintenance intervals. If this is the case, the worst-case scenario is a breakage of the cutting element.

[0133] With the help of machine learning methods, force-angle curves such as those shown in Figures 600a, 600b can be used for state determination. Machine learning is a general term for the "artificial" generation of knowledge from experience: an artificial system learns from examples and can generalize about these examples after the learning phase is completed. To do this, machine learning algorithms build statistical models based on training data. This means that instead of simply learning these examples repeatedly, patterns and regularities are identified in the learning data so that the system can also evaluate unknown data.

[0134] In addition, additional AI (artificial intelligence) models can be used to identify and classify error states. In the simplest case, the future evolution can be approximated by a linear function. However, if the data situation allows, regression models can be used to improve the accuracy of the prediction.

[0135] Figure 6 The data shown can be used to obtain information about fatigue or wear of the punching tool. The recorded angle 602 refers to the distance the top of the punching tool must move. Therefore, the area under the force-angle curve contains information about the mechanical work required during the punching process. mech Assume that W mech Increases with increasing fatigue level. After maintenance, W mech Should be restored to the initial value.

[0136] Currently, static maintenance intervals are used, which do not work well in the event of an unexpected event such as an accident or contamination. In this case, the stamping tool may already be in a dangerous state. If the worker then continues production, the tool will be damaged or possibly destroyed. With a machine learning model, regression can be performed to estimate W mech The most likely evolution. With this information combined with the knowledge of the physical limits, it is possible to predict the remaining life of the punching tool. If the historical data contains curves showing fractures, information about the physical limits can be obtained.

[0137] Figure 7 An exemplary illustration of states 700a, 700b, 700c of a punching process is shown using an eccentric press as an example.

[0138] Figure 7A punching tool 730, such as an eccentric press, is shown. The punching tool 730 periodically moves from a starting position 710 to an end position 710 for punching the material, and due to the periodicity of the punching process, the end position coincides with the starting position 710. In embodiment 700a, the punching tool 730 is in a first position, in which it presents an angle 701 of α=90°, corresponding to a path x 702. In embodiment 700b, the punching tool 730 is in a second position, in which it presents an angle 701 of α=120°, corresponding to a path x 702. In illustration 700c, the punching tool 730 is in a third position, in which it presents an angle 701 of α=160°, corresponding to a path x 702.

[0139] The punching tool 730 is periodically moved between a starting position 710 (only schematically shown here) and an end position 710 (only schematically shown here) to enable punching, through an intermediate position 720 (only schematically shown here), also referred to as the bottom dead center, which also corresponds to the starting position 710. The starting position 710 corresponds to an angle 701 of α=0°, the intermediate position 720 corresponds to an angle 701 of α=180°, and the end position 710 again corresponds to an angle 701 of α=0°.

[0140] At the beginning it was mentioned that as fatigue increases, mechanical work should increase. To determine an approximate value for mechanical work, the following formula is given, which is based on the systematics of eccentric machines, such as Figure 7 shown.

[0141] The mechanical work W as a function of the angle α can be approximately calculated as follows mech :

[0142] W mech (α)=∫(F N (α)·x(α))dα

[0143] W mech (α)=∫(F N (α)·(s(α)+w(α))dα

[0144]

[0145]

[0146] Only geometrical aspects are considered here, i.e. the properties of the rotating masses are not taken into account in this approximation.

[0147] By means of the mechanical work determined in this way, the wear characteristics or fatigue of the punching tool can be determined. The mechanical work determined in this way can also be a state variable 104 of the punching process, as described above for Figure 1This state variable can also be used to determine the wear state and / or maintenance time of the punching tool, as described above. Figure 1 As shown in the “status determination” module 118.

[0148] Figure 8 An exemplary illustration of a punching tool 800 according to the present invention is shown.

[0149] The punching tool is multi-layered and includes a functional plate 810 and a frame plate 820. The frame plate 820 includes a lower clamping strip 826, on which a bottom plate 825 is arranged. On the bottom plate 825, a punching strip 824 of the product is arranged on the processing plane. Above this is an intermediate plate 823, on which a top plate 822 is arranged, and on top of the top plate 822, an upper clamping plate 821 is arranged.

[0150] The functional plate 810 comprises a hardened cutting plate 814, with which cutting is performed. Above this is a hardened guide plate / peeling plate 813. Both plates 814, 813 are located between an intermediate plate 823 and a bottom plate 825. A hardened or soft punch fixing plate 812 is arranged above the intermediate plate 823, and a hardened pressing plate 811 is arranged above the punch fixing plate 812.

[0151] Fig. 9 A schematic diagram of a method 900 according to the present invention for determining a state of a punching process is shown.

[0152] As mentioned above about Figure 7 and Figure 8 As described above, during the punching process, the punching tools 730 , 800 periodically move from the starting position 710 to the ending position 710 to punch the material.

[0153] The method comprises the following steps:

[0154] Obtain 901 first data 101, the first data indicating the force applied by the punching tool to the material when the punching tool moves from a starting position to an end position in a plurality of cycles of the punching process, such as the force applied by the punching tool to the material as described above with respect to Figure 1 Description of

[0155] Obtain 902 second data 102, the second data indicating the angle or stroke traveled by the punching tool when moving from the starting position to the ending position during multiple cycles of the punching process, as described above for Figure 1 as stated;

[0156] Obtain 903 third data 103, the third data indicating the temperature of the stamping tool during multiple cycles of the stamping process, as described above Figure 1 Description of

[0157] The start-up data obtained during the start-up cycle of the punching tool are removed 904 from the first data 101 and the second data 102, as described above for Figure 1 as stated;

[0158] For multiple cycles of the stamping process, the first data 101 and the second data 102 are interpolated 905 to obtain a uniform length and step size of the second data 102 relative to the first data 101, as described above with respect to Figure 1 Description of

[0159] Identify and remove 906 anomalies in the first data 101 and the second data 102, as described above with respect to Figure 1 Description of

[0160] Determine and compensate 907 the thermal disturbance of the third data 103 to the first data 101 and the second data 102, as described above for Figure 1 as stated; and

[0161] The state of the stamping process is determined 908 based on the first data 101 and the second data 102 that are compensated for thermal disturbances, as described above. Figure 1 described.

[0162] The various steps of the method may correspond to the above Figures 1 to 8 function.

[0163] Reference numerals list

[0164] 100 Device for determining the state of a stamping process

[0165] 101 First Data

[0166] 102 Second Data

[0167] 103 Third Data

[0168] 104 Stamping process status

[0169] 110 Remove startup data

[0170] 112 Interpolation

[0171] 114 Remove exceptions

[0172] 116 Thermal compensation

[0173] 118 Status confirmed

[0174] 200a / b / c Force-angle curve in original state

[0175] 300a / b / c Force-angle curves for start-up data removed

[0176] 400 Measurement of pseudo force over time

[0177] 401 First day of measurement

[0178] 402 Measurements on the second day

[0179] 500a Pseudo force measurement over time without smoothing

[0180] 500b Pseudo force measurement over time with smoothing filter

[0181] 600a Force-angle curve

[0182] 600b Force-angle curve

[0183] 601 Force or pseudo-force

[0184] 602 Angle

[0185] 611 The first curve family

[0186] 612 The second curve family

[0187] 613 The third family of curves

[0188] 614 Peak or spike value

[0189] 700a The first state of the stamping process using an eccentric press as an example

[0190] 700b Second state of stamping process

[0191] 700c The third state of the stamping process

[0192] 701 Angle information

[0193] 702 Trip Information

[0194] 710 Starting and ending positions of punching tools

[0195] 720 Punch tool intermediate position

[0196] 730 Punching Tools

[0197] 800 Punching Tools

[0198] 810 Function Board

[0199] 811 Pressure Plate

[0200] 812 Punch fixing plate

[0201] 813 Guide Plate / Peel Plate

[0202] 814 Cutting Board

[0203] 820 frame board

[0204] 821 Upper splint

[0205] 822 Top Plate

[0206] 823 Middle Plate

[0207] 824 Stamping Strip (Product)

[0208] 825 Base Plate

[0209] 826 Lower clamp

[0210] 900 Method for determining the state of a stamping process

[0211] 901 First Method Step

[0212] 902 Second method step

[0213] 903 Third Method Step

[0214] 904 Fourth process step

[0215] 905 Fifth method step

[0216] 906 Sixth method step

[0217] 907 Seventh Method Step

[0218] 908 Eighth Method Step

Claims

1. A device (100) for determining a state of a stamping process, wherein: A punching tool (730, 800) for punching a material is periodically moved from a starting position (710) to an end position, the end position corresponding to the starting position due to the periodicity of the punching process, and the device (100) is designed as Obtaining first data (101), the first data representing the force applied by the punching tool to the material when the punching tool moves from a starting position to an ending position during a plurality of cycles of a punching process; Obtaining second data (102), the second data representing the angle or stroke traveled by the punching tool when moving from a starting position to an ending position during a plurality of cycles of the punching process; Obtaining third data (103), the third data representing the temperature of the stamping tool during a plurality of cycles of the stamping process; removing the start-up data detected during the start-up cycle of the punching tool from the first data (101) and the second data (102); interpolating the first data (101) and the second data (102) so that the second data (102) has a uniform length and step size relative to the first data (101) for a plurality of cycles of the stamping process; Identifying and removing anomalies in the first data (101) and the second data (102); Determining and compensating for thermal disturbance of the third data (103) to the first data (101) and the second data (102); as well as The state of the stamping process is determined based on the first data (101) and the second data (102) compensated for thermal disturbances.

2. The device (100) according to claim 1, in, The first data (101) represent the force acting on the punching tool (730, 800) due to the punching process by means of a bending or deformation of the punching tool (730, 800).

3. The device (100) according to claim 1 or 2, in, The first data (101) is sensor data of a piezoelectric sensor placed in a punching tool (730, 800).

4. The device (100) according to claim 3, in, The second data (102) includes angle information (602, 701) of a rotary encoder of a punching tool; or The second data (102) includes stroke information (702) of a stroke sensor of a punching tool.

5. The device (100) according to claim 4, in, The angle information of the rotary encoder of the punching tool is the eccentric angle information of the eccentric press (700a, 700b, 700c).

6. The device (100) according to claim 4, in, The stroke information of the stroke sensor of the punching tool is the stroke information of the piston of the hydraulic press.

7. The device (100) according to claim 4, in, The first data (101) and the second data (102) include force-angle curves (200a, 200b, 200c), and The device (100) is designed to determine the start-up data based on the deviation between the shape of the force-angle curve (200a, 200b, 200c) during the start-up cycle of the punching tool and the shape of the force-angle curve (200a, 200b, 200c) after the start-up cycle of the punching tool.

8. The device (100) according to claim 4, in, The device (100) is designed to determine the activation data based on the shift of the force-angle curve (200a, 200b, 200c) in the direction of the force during the activation cycle of the punching tool.

9. The device (100) according to claim 7, in, The device (100) is designed to remove force-angle curves (200a, 200b, 200c) generated when the number of strokes in a punching process is lower than a fixed percentage of the rated number of strokes in the punching process from the first data (101) and the second data (102).

10. The device (100) according to claim 1 or 2, in, The device is designed to extend the first data (101) and the second data (102) by one or more zero terms for each detection cycle of the stamping process until the second data (102) has a uniform length and step size relative to the first data (101).

11. The device (100) according to claim 4, in, The first data (101) and the second data (102) include force-angle curves (300a, 300b, 300c), and The device (100) is designed to generate a reference curve from the force-angle curves (300a, 300b, 300c) by averaging, and to determine anomalies based on deviations of the force-angle curves (300a, 300b, 300c) relative to the reference curves.

12. The device (100) according to claim 11, in, The deviation is based on the product of the standard deviation of the force-angle curve (300a, 300b, 300c) and an adjustable factor.

13. The device (100) according to claim 11, in, The first data (101) and the second data (102) are divided into different channels, each channel including the first data (101) and the second data (102) recorded by a corresponding piezoelectric sensor on the punching tool.

14. The device (100) according to claim 13, in, The device (100) is designed to, when an abnormality is detected in the force-angle curve (300a, 300b, 300c) of one of the different channels, also mark the corresponding force-angle curve (300a, 300b, 300c) of the other channels as abnormal.

15. The device (100) according to claim 1 or 2, in, The device (100) is designed to determine a compensation function of third data (103) for thermal disturbance (400) of first data (101) and second data (102).

16. The device (100) according to claim 15, in, The compensation function is based on a model which specifies deviations of the first data (101) as a function of the third data (103), wherein the model is based on a mathematical approximation function and / or on a prediction using artificial intelligence.

17. The device (100) according to claim 15, in, The device (100) is designed to determine a compensation function for each individual angle, and to compensate the third data (103) for thermal disturbances of the first data (101) and the second data (102) for each individual angle.

18. A method (900) for determining a stamping process state, wherein: A punching tool (730, 800) for punching a material is periodically moved from a starting position (710) to an end position, the end position corresponding to the starting position due to the periodicity of the punching process, and the method comprises the following steps: Obtaining first data (101), the first data indicating the force applied by the punching tool to the material when the punching tool moves from a starting position to an ending position during a plurality of cycles of a punching process; Obtaining second data (102), the second data indicating the angle or stroke traveled by the punching tool when moving from a starting position to an ending position during a plurality of cycles of the punching process; Obtaining third data (103), the third data indicating the temperature of the stamping tool during a plurality of cycles of the stamping process; removing start-up data obtained during a start-up cycle of the punching tool from the first data (101) and the second data (102); interpolating the first data (101) and the second data (102) so as to obtain a uniform length and step size of the second data (102) relative to the first data (101) for a plurality of cycles of the stamping process; Identifying and removing anomalies in the first data (101) and the second data (102); determining and compensating for thermal disturbance of the third data (103) to the first data (101) and the second data (102); and The state of the stamping process is determined (908) based on the first data (101) and the second data (102) compensated for thermal disturbances.

Images