METHOD FOR IDENTIFYING ANOMALIES AND THUS ACTUAL PROCESSING POSITIONS OF A RAILWAY AND FOR CLASSIFYING THESE ANOMALIES, PLANT AND COMPUTER PROGRAM PRODUCT

Abstract

The invention relates to a method, wherein: a web (4) is provided in which a number of machining operations (8, 10) have been made by means of a cutting / grooving unit (6), in each case on the basis of a predefined machining position (24) and a predefined cut type or groove type; a height profile (14, 16) of the web (4) is measured using a sensor unit (12) on at least one side of the web (4) and transversely to a conveying direction (F) of the web (4); a number of anomalies (20) and thus also actual machining positions (18) are identified in the height profile (14, 16) by identifying each section of the height profile (14, 16) where the height profile (14, 16) reaches a predefined threshold value (26) as an anomaly (20); and each anomaly (20) is classified by assigning same to the cut type or groove type said anomaly (20) most closely resembles. The invention also relates to a system (2) and to a computer programme product.

Description

[0001] Description

[0002] Method for identifying anomalies and thus also actual processing positions of a web and for classifying these anomalies, system and computer program product

[0003] The invention relates to a method for identifying anomalies, in particular cuts and / or creasing, and thus also actual processing positions of a web, in particular a corrugated cardboard web, as well as for classifying these anomalies. Furthermore, the invention relates to a system and a computer program product.

[0004] The web is, for example, a corrugated cardboard web and then composed of several layers of paper. The system is, for example, a corrugated cardboard plant, with which the web is first produced and then further processed. Typically, the web is processed in such a way that a number of cuts and / or creasing lines are introduced into it using a cutting and creasing unit. On the one hand, this is to divide the web into several separate blanks and then sheets using suitable cuts, and on the other hand, this is to provide the sheets with a number of folding, folding, separating edges or the like using suitable creasing. This means that the respective sheet can then be converted from an originally flat form into a typically three-dimensional final shape (e.g. a box or case).

[0005] For the most economical and environmentally friendly production of corrugated board webs, as well as the panels and sheets made from them, it is desirable to produce as little waste as possible during production and processing. Waste regularly occurs when cuts and / or creasing are incorrect or not made at all. The earlier and more reliably any defects in the application of cuts and / or creasing are detected, the less waste will be generated and the more targeted appropriate action can be taken, for example, by replacing a tool or adjusting an operating parameter.

[0006] DE 10 2015 200 397 A1 describes a testing device for testing the quality of processing results of a processing device for substrate sheets, comprising: a light source for illuminating a processing section of a substrate sheet that is mechanically processed by the processing device by means of a processing tool in a structure-changing manner, an image sensor that is configured to capture an image of the illuminated processing section and, on this basis, to generate actual image data of the processing section.Furthermore, an evaluation device is provided which is configured to output an evaluation result for the quality of the machining results, wherein target data for the machined machining section are stored in the evaluation device and the evaluation device is configured to compare the actual image data of the machining section with the target data under predetermined evaluation criteria and to determine correction data for the machining parameters as part of the evaluation result on the basis of a resulting comparison result.

[0007] DE 10 2019 105 217 A1 describes a folding machine for folding box blanks, comprising at least one control unit, means for conveying the folding box blanks from an input to an output of the folding machine, a device for folding the folding box blanks, and an alignment unit for folded folding box blanks. The folding machine has a device for optically detecting creasing and / or gaps in the folding box blanks, and a device for optically detecting creasing in the transport direction is arranged upstream of the means for folding the folding box blanks, and / or a device for optically detecting gaps is arranged downstream of the means for folding the folding box blanks and upstream of the alignment unit.US 5,581,353 A describes a device for measuring at least two properties of corrugated board, the device comprising at least one pair of laser triangulation sensors, one of the sensors being arranged on one side of the corrugated board and the second sensor on an opposite side. The distance between the respective sensor and the surface of the corrugated board is measured and converted into at least two properties of the corrugated board.

[0008] Against this background, the object of the invention is to improve the monitoring of the processing of a web by a cutting-creasing unit. For this purpose, a corresponding method, a system, and a computer program product are to be provided.

[0009] The object is achieved according to the invention by a method having the features according to claim 1, by a system having the features according to claim 16, and by a computer program product having the features according to claim 17. Advantageous embodiments, further developments, and variants are the subject of the dependent claims. The statements in connection with the method also apply mutatis mutandis to the system and the computer program product, and vice versa. Advantageous embodiments for the system arise from the fact that it is designed to carry out at least parts of the method or one or more steps of the method. For this purpose, the system then has a suitably designed control unit. Analogously, advantageous embodiments for the computer program product arise from the fact that it has instructions which, when executed by a computer, cause the computer to carry out at least parts of the method or one or more steps of the method.

[0010] The method is in particular a method for identifying anomalies and thus also actual processing positions of a web and for classifying these anomalies. The method enables in particular improved checking of processing operations by a cutting / creasing unit and in particular also improved checking of processing positions of these processing operations, i.e. cutting and / or creasing positions. Firstly, a web is provided into which a number of processing operations have been introduced by means of a cutting / creasing unit, each based on a predetermined processing position and a predetermined cut type or creasing type (generally referred to as a “processing type”). “A number of” is understood here and also generally to mean “one or more” or “at least one”. The processing operations are mechanical processing of the web and in each case either a cut or a crease in the web.Each cut was created according to a predetermined cut type and each crease according to a predetermined crease type. The cutting-creasing unit is therefore used to introduce cuts and / or crease marks into the web. A cut represents a separation point at which the web is severed. In contrast, a crease initially only represents an embossed point, in particular a predetermined crease (or similar) in the web, which allows a finished blank or sheet to be folded along this predetermined crease. Accordingly, cuts and crease marks differ in particular in that with a cut the web is completely severed, which is not the case or only partially the case with creasing. The cuts and / or creasing marks are used in particular for converting the web into individual blanks or sheets and / or for forming break, fold or crease edges of the blanks or the subsequent sheets, e.g.to form each one into a box, crate or similar.

[0011] A cutting / creasing unit is understood in particular to mean a cutting and / or creasing unit, i.e. a unit which is designed either to cut the web or to creasing the web, or both. In the following, without loss of generality, a cutting and creasing unit is assumed with which the web is both cut and creasing. The cutting / creasing unit is in particular part of a system for processing the web. Preferably, the system is a corrugated board system for producing corrugated board, in particular sheets of corrugated board. The web is made in particular from paper and is preferably a corrugated board web composed of several paper layers. Preferably, the cutting / creasing unit is part of the corrugated board system, and the web is a corrugated board web which is produced and processed using the corrugated board system.In a suitable embodiment, the cutting / creasing unit is arranged downstream of a so-called double-facer of the system. In the double-facer, several intermediate products for a corrugated board web are combined to form the actual corrugated board web. The double-facer also marks the end of a so-called "wet end" of the system (and is still part of this), followed by a so-called "dry end" with which the corrugated board web is converted, in particular into individual blanks and subsequently into individual sheets. The cutting / creasing unit is, in particular, part of the dry end of the system. In the following, it is assumed, without loss of generality, that the web is a corrugated board web. While such an embodiment is preferred, the statements made here generally apply to any web, in particular paper web, possibly even a single-ply web, and also to any system that processes a web.

[0012] The web is processed by the cutting / creasing unit at a number of actual processing positions (i.e., cutting and / or creasing positions). Due to the nature of the processing, a particularly characteristic anomaly is formed at each actual processing position. By identifying the anomalies (as explained in detail below), the actual processing positions are automatically identified, i.e., recognized, which are then also referred to as recognized processing positions. The processing positions are specified, in particular, by order data for an order for the system for processing the web. Accordingly, the order data then contains a number of specified processing positions.The cut or crease type to be performed at a given processing position is also specified in the order data, so that each specified processing position is also assigned a specified cut or crease type in the order data. The cut-and-crease unit is controlled according to the order data. However, due to errors or inaccuracies, the cut-and-crease unit does not necessarily perform the processing at the specified processing positions, but rather at the actual processing positions. It is possible that the required cut or crease type may not be reproduced exactly, or even that an incorrect cut or crease type may be used.In other words, the cuts and creasing that are actually created are not necessarily located exactly at the specified processing positions, and the execution of a cut or creasing can also be faulty, resulting in the wrong cut or creasing type being created at a processing position. It is therefore advantageous to check the processing result. Furthermore, an anomaly is not necessarily due to processing with the cutting / creasing unit, but can also be caused by other processing or damage. These anomalies and the corresponding actual processing position are also detected; however, the latter is not assigned to a specified processing position, nor does the anomaly necessarily correspond to a known cut or creasing type.Therefore, the inspection is advantageously designed in such a way that such anomalies are not mistakenly identified as predefined machining operations. Advantageously, the anomalies are evaluated and characterized independently of one another, i.e., classified as a whole, so that erroneous conclusions (e.g., based on the assumption that all anomalies are cuts or creasing) are avoided as far as possible from the outset.

[0013] In the method described here, a height profile of the web is measured using a sensor unit on at least one side of the web, preferably on both sides of the web, and transversely to a conveying direction (i.e. also the longitudinal direction) of the web. The sensor unit is preferably part of the system. The height profile is preferably measured inline in the system and downstream of the cutting / creasing unit. The web has a width perpendicular to the conveying direction and the height profile preferably extends over the entire width. The height profile is also referred to as the transverse profile because it is generally measured transversely and preferably - but not necessarily - perpendicular to the conveying direction. The height profile is also referred to as the surface contour because it reproduces the contour of a surface of the web. As described, deviations from the specifications according to the order data can generally occur during processing.Processing with the cutting and grooving unit generates characteristic changes in the height profile of the web, allowing the actual processing positions and the processing performed there to be identified and verified by measuring and evaluating the height profile. The height profile is measured, for example, with a distance sensor and then indicates the distance between the web and the sensor as a function along a transverse direction perpendicular to the conveying direction. The height profile can be represented, in particular, as a vector with numerous measured values, each at a different position in the transverse direction. Each measured value and a respective position (i.e., position value) together form a data point; the length of the vector corresponds to the number of data points.

[0014] In the height profile, the machining operations by the cutting-grooving unit are recognizable as anomalies at actual machining positions, i.e., as deviations, for example, from a normal state of the path, i.e., a path without machining operations. Therefore, a number of anomalies and thus also actual machining positions are identified in the height profile by identifying those sections of the height profile as an anomaly in which the height profile reaches, in particular exceeds or falls below, a predetermined (first) threshold value. A respective anomaly is therefore a section of the height profile and thus two-dimensional, i.e., contains several data points (measured values ​​and position values). Each anomaly corresponds to an actual machining position, which contains the positions of the corresponding section, i.e.,only the position values ​​of a respective anomaly, and is therefore one-dimensional and, so to speak, a position interval or position range. By using a threshold value, a definition of the normal state is not important; rather, the threshold value indicates what is considered normal / non-normal. A respective section is, in particular, a connected set of data points. A respective section is expediently not just the area in which the elevation profile actually reaches the threshold value, but contains a certain addition, particularly on both sides. The addition is expediently dimensioned such that the anomaly, which along the elevation profile regularly begins before the threshold value is reached, is captured as completely as possible by the section.

[0015] Each anomaly is then classified by means of a classification in which the anomaly is assigned to the cutting and / or creasing type to which the anomaly is most similar. Therefore, it is not just a check whether or to what extent a particular processing operation corresponds to the cutting or creasing type according to the order data, but rather a check is carried out to determine which cutting or creasing type is actually present. The check is initially independent of the order data. The classification therefore differs from a simple target / actual comparison in particular in that the anomaly is not merely compared with a single cutting or creasing type, but with several different ones. Furthermore, in this case, the threshold value is used to actively search for anomalies in the entire height profile and not just to examine the height profile at the specified processing positions, so that non-specified processing operations (e.g.Damage) can be detected and advantageously classified. Overall, this improves the inspection of the web's processing and, in particular, enables more detailed quality control than with a simple target / actual comparison and a review of the processing positions of the cuts and creasing in only predefined sections. This advantageously detects any incorrect cuts and creasing that occur outside the expected sections.

[0016] The classification described here is also more reliable than a simple minimum or maximum function for identification, which by its nature always detects a cut or a crease even when nothing of the sort has occurred. In addition, validation is possible in this case, which advantageously detects whether the correct creasing profile (i.e. the correct creasing element) was used for a particular creasing operation in accordance with the order data. Finally, the method described here can also be used to check the web for damage away from the cuts and creasing. Overall, the method described here makes it possible to detect missing or incorrectly positioned cuts and creasing, incorrectly inserted cutting elements and creasing elements, inadequately executed cuts and creasing, and damage to the web.In response, a corresponding warning is appropriately issued and / or a system operator is informed accordingly. Optionally, the system (and thus further processing of the web) is stopped or at least slowed down, thus preventing the production of scrap.

[0017] The procedure performed and described here, especially the classification, also contrasts with the aforementioned DE 10 2015 200 397 A1, DE 10 2019 105 217 A1, and US Pat. No. 5,581,353 A. During classification, a specific cut or crease type is assigned to a signal section of the elevation profile. This not only identifies the actual cut or crease produced, but also its type. Furthermore, the entire web, or at least extensive sections thereof, is examined, not just those individual positions where cuts or crease marks are expected.

[0018] For cutting and / or creasing, the cutting / creasing unit in particular has one or more corresponding cutting bodies and / or creasing bodies, i.e. preferably a plurality of cutting bodies and / or a plurality of creasing bodies, and not just a combination of a single cutting body and a single creasing body. In both cases, the web is mechanically processed; in the case of cutting, this merely involves separating it, and in the case of creasing, this involves forming it. A cutting body typically has a first knife, which rotates about an axis transverse to the web during operation and is then moved into the web to create a cut. The cutting body also has a counter-roller, which is arranged on the opposite side of the web and interacts with the first knife (a second knife instead of the counter-roller is also conceivable).Similarly, a creasing element usually has a first roller with a first profiled running surface, which rotates during operation about an axis transverse to the web and is then pressed against the web to create a crease. The creasing element also has a counter-roller, usually with a second profiled running surface, which is arranged on the opposite side of the web and interacts with the first profiled running surface. The first and second running surfaces do not necessarily have to be identically profiled. Various pairings of running surfaces are suitable for creating different creasing patterns, for example so-called one-point creasing, three-point creasing, five-point creasing, point-point creasing or asymmetric creasing. A particularly common creasing pattern is three-point creasing using a three-point creasing element.By shifting the running surfaces of a creasing element relative to each other in the transverse direction, so-called offset creasing is also possible. This shift is also referred to as offset. Such offset creasing can also be detected and verified using the method described here. A further shift is usually also possible perpendicular to both the transverse direction and the conveying direction. This sets a distance between the two running surfaces of a creasing element, the so-called creasing element spacing. This is conveniently done depending on the thickness of the web. The creasing element spacing can also be detected and verified using the method described here.

[0019] Advantageous aspects and preferred embodiments of the method are described in detail below. An overview is provided first, followed by a more detailed discussion of the following aspects: measurement, preprocessing, position detection, validation of the processing positions, classification, and validation of the classification result.

[0020] Overview

[0021] In this method, as already described, at least one elevation profile of the track is measured during a single measurement and then evaluated by identifying anomalies (i.e., conspicuous sections of the elevation profile). For evaluation, the elevation profile is expediently first processed and, if necessary, filtered in a preprocessing process, which particularly uses digital signal processing methods.

[0022] Subsequently, the anomalies and thus the detected processing positions, i.e., the actual processing positions, are identified in the elevation profile using position detection, which particularly applies suitable statistical methods. Position detection is performed particularly during ongoing system operation to ensure smooth processing of the web. Subsequently, the detected processing positions are advantageously assigned to the specified processing positions from the order data.

[0023] For further analysis, the method described here uses the aforementioned classification, by means of which the height profile is classified section by section at the detected processing positions with regard to the processing performed (cuts and creasing), in particular with regard to any damage, deviations, production errors, etc., i.e., the anomalies are classified. For the actual classification, i.e., the assignment of an anomaly to a specific cut or creasing type (preferably also to a damage type, deviation type, production error type, etc.), various configurations are suitable, some of which are described below.

[0024] In addition, validation is advantageously also carried out. The validation is a validation of the processing positions or a validation of the classification result, preferably both, so that the validation is then two-part, which is assumed below without loss of generality. During the validation of the processing positions, in particular, a note is issued, the system is stopped, or another suitable measure is initiated as soon as it is determined (e.g. already during position recognition) that the recognized processing positions do not correspond within a given tolerance with the specified processing positions from the order data or are directly recognized as damage, deviations, production errors, or the like. The validation of the classification result based on the order data is carried out analogously to the validation of the processing positions, i.e.A check is carried out to determine whether the correct cut or creasing type was used and / or executed correctly, in particular whether the cuts and creasing were produced using the correct cutting or creasing tool. As soon as this validation determines that the detected processing steps do not match the specified cut and / or creasing types from the order data within a given tolerance, a corresponding message is issued, the system is stopped, or another appropriate measure is initiated. Overall, validation in any form checks and ensures a certain quality of the processing positions and the processing performed there and, if necessary, initiates an appropriate response.

[0025] The method is preferably implemented by multiple modules, which are in particular functionally separate from one another. A first module is preferably a data preprocessing module, which performs the preprocessing and accordingly provides the measured height profile as a uniform database for the subsequent modules. A second module is preferably a detection module, which performs position detection and accordingly identifies the anomalies in the height profile, then calculates them into recognized (also detected or actual) processing positions, and finally outputs them, preferably to a third module, which is an assignment module. The assignment module performs the assignment of the processing positions and assigns the recognized processing positions to the specified processing positions according to the order data.A fourth module is preferably a classification module, which performs the classification and assigns a cut or crease type to each anomaly, or optionally even a damage type, deviation type, production error type, or the like. A fifth module is preferably a validation module, which performs the validation and compares the detected processing positions with the specified processing positions and / or compares the detected cut and / or crease types with the specified cut and / or crease types. Optionally, one or more of the modules are present multiple times, in particular to evaluate cuts and creasing separately, in particular to classify and / or validate them.

[0026] The various modules are suitably combined with one another in a control unit (integrated design); alternatively, one or more modules are housed separately in one or more other control units (distributed design). One or more of the aforementioned control units, and thus also the modules contained therein, are advantageously part of the system described here.

[0027] measurement

[0028] The height profile is suitably measured using a sensor unit (also referred to as a sensor module) with a number of sensors. A suitable sensor is, in particular, a distance sensor as already mentioned above, e.g., a laser triangulation sensor, which can be moved transversely to the web (e.g., by means of a linear motor) and measures a distance in the direction of the web, i.e., the distance between the sensor and the web. A camera is also suitable as a sensor, with which, in particular, an image of the web is taken, from which the height profile is then obtained. The sensor outputs a measured value (distance) for each measuring position across the web, which varies depending on the nature of the surface of the web. As a result, the sensor outputs a height profile containing a series of measured values, particularly in the form of a vector. Each measured value is also assigned a position (measurement position) along the web, which, for example,is determined from the travel path of the sensor. A pair of a measured value and the associated position is a data point. The sensor unit is preferably arranged immediately downstream of the cutting / creasing unit, i.e. the web path from the cutting / creasing unit to the sensor unit is at most 10 m, preferably at most 1 m. The sensor unit expediently has a pressure roller to press the web against a support as it passes through the sensor unit and when measuring the height profile, and to suppress any cambering of the web as far as possible, at least during the measurement.

[0029] The sensor unit preferably has two sensors as described, one of which measures an upper height profile along an upper side of the web and another, analogously, a lower height profile along an underside of the web. This advantageously takes into account detailed information which is only found in one of the two height profiles. In particular, both height profiles are processed in the same way, so that for the sake of simplicity we will only refer to one height profile below. In principle it is sufficient to use just one height profile (i.e. on the upper side or on the underside), but the use of both height profiles is preferred, especially when classifying creasing, since the height profiles on the upper side and underside regularly differ and thus enable more precise detection of the creasing type and thus classification.In addition, cuts and grooves can be more clearly distinguished from each other. However, the processing steps for both height profiles are, for convenience, identical.

[0030] In principle, the data points obtained with the measurement can be used directly for the subsequent steps and modules, but advantageously the data points are preprocessed, ie the measured height profile is preprocessed and is then a preprocessed height profile.

[0031] Pre-processing with the data pre-processing module

[0032] During preprocessing, the measured height profile is suitably prepared, particularly for subsequent position detection. For example, it is possible that the measuring range is exceeded during measurement of the height profile and no distance can be measured. This is regularly the case with a cut because the path is interrupted at the processing position. Accordingly, if the measuring range is exceeded, a valid measured value may not be generated; instead, a NaN (not a number) is returned as the measured value, and a corresponding gap is found in the height profile at the associated measuring position. During preprocessing, such gaps (NaNs) in the height profile are expediently removed in order to obtain a gap-free height profile. However, the gaps are not simply omitted, but replaced with suitable values, e.g. using linear interpolation based on the measured values ​​of neighboring measuring positions.Alternatively, a gap is simply replaced by a previously defined value, e.g., using a sample-and-hold function with the last valid measured value. Alternatively, a value of "0" is used for each gap (i.e., a vanishing distance). In any case, the number of data points should be maintained as much as possible. However, the options mentioned may have significant effects in the frequency domain, so it is advisable to additionally filter the elevation profile prepared in this way with an anti-aliasing filter.

[0033] Furthermore, the data preprocessing module conveniently trims the measured height profile to the width of the web, i.e., the zero point of the height profile is shifted to a side edge of the web. Alternatively or additionally, the height profile is also limited to the width of the web, which facilitates subsequent comparison of the detected machining positions with the specified machining positions.

[0034] The elevation profile is expediently mapped onto a grid, in particular an equidistant one, using the data preprocessing module. In this case, new data points with new (measured) values ​​and positions are generated from the measured values ​​and measurement positions, which are distributed along an equidistant grid. In this way, the distance between the data points of the elevation profile is determined. The elevation profile is then, in particular, a vector in which each individual component corresponds to a value at a position along the elevation profile. If necessary, further data points are derived (e.g., interpolated) based on the measured values ​​in order to obtain a specific grid and, in particular, a specific data point density (e.g., number of data points per millimeter). The data point density is particularly relevant for the accuracy when viewing and processing the elevation profile in the frequency domain.Studies have shown that a data point density of no more than 10 / mm is sufficient; with an equidistant grid, this corresponds to a distance of 0.1 mm between the individual positions of the elevation profile. Preferably, the data point density is specified, and based on this, equidistantly spaced support points are defined as new positions. At these new positions, a new value is then determined for each data point based on the measured values ​​and measurement positions. For example, if a measurement position and a support point match, this measurement position with the associated measured value is used as the data point; otherwise, the data point is interpolated from the nearest measured values ​​and measurement positions. The resulting elevation profile is also referred to as the adjusted elevation profile.The height profile is also calibrated to compensate for deviations in the parallelism of the two sensors due to manufacturing tolerances or temperature fluctuations. For this purpose, a calibration vector is determined using a calibration measurement, for example, and then subtracted from the height profile. However, the specific design of such a calibration is of secondary importance here and will therefore not be described in detail.

[0035] The prepared, cropped, cleaned and / or calibrated elevation profile is then output as a preprocessed elevation profile for the subsequent steps and modules.

[0036] filter

[0037] The detection module is designed to identify any anomalies (connected subsets of data points) in the processed elevation profile and thereby detect the actual processing positions (connected subsets of positions). First, it is assumed that the elevation profile for a web without cuts or grooves, i.e., in the normal state of the web, can be described, at least to a good approximation, as a constant function. Based on this, an anomaly is then conveniently defined as any deviation of the elevation profile from this constant function. In this case, it was found that anomalies can also be detected in the frequency domain, specifically in the spectrogram of the elevation profile. The spectrogram is a two-dimensional spectral representation and contains the frequency-dependent amplitude of the elevation profile as a function of the position of the respective data point (in particular along the grid).The spectrogram is generated from the height profile, for example, using a short-time Fourier transform (STFT), where the time dimension is replaced by position (i.e., a spatial dimension), so that the frequencies are then inverses of position / . The achievable frequency range depends on the data point density described above (Nyquist-Shannon sampling theorem). With ten data points per millimeter, a frequency range of 0 to 5 mm is possible. -1can be displayed. Investigations have shown that continuous lines (parallel to the frequency axis) appear in the spectrogram at exactly the same positions as the actually executed cuts. This means that a cut can be identified across the entire frequency spectrum and that by filtering out a suitable sub-range from the frequency spectrum (i.e., by selecting a specific frequency range) which only contains the frequencies of a cut, the influence of this cut on the measured height profile can be displayed in isolation, thus enabling improved identification of the cut. Accordingly, in an advantageous embodiment, for better identification of cuts in particular, a corresponding frequency range is isolated using a bandpass filter. The bandpass filter then has a passband which corresponds to the frequency range. A suitable frequency range ranges, for example, from 1.5 mm-1 up to 5 mm in particular -1 , since in such a frequency range, only the effect of the cuts is visible in the spectrogram, and all other influences on the height profile lie entirely or at least predominantly below this frequency range. Accordingly, the upper limit is not necessarily important; rather, it is determined in particular by the data point density. Similarly, at the processing positions of grooves in the spectrogram, anomalies in the frequency range from 0 mm are also visible. -1 up to about 0.1 mm -1 recognizable, so that in an advantageous embodiment, specifically for better identification of grooves, a corresponding frequency range is exposed using a bandpass filter. A suitable frequency range (ie passband of the bandpass filter) extends, for example, from a frequency > 0 mm -1 (e.g. 0.001 mm -1 ), to suppress the DC component in the height profile at 0 mm-1 , up to 0.1 mm -1 A combination of both of the bandpass filters mentioned above is also useful. Depending on the application, frequency ranges other than those mentioned above may also be advantageous.

[0038] In general, therefore, to better identify one or more anomalies, a frequency range containing the anomalies is advantageously isolated using a bandpass filter. In a suitable embodiment, the height profile is filtered with a bandpass filter before the anomalies are identified. The bandpass filter isolates a frequency range selected to specifically preserve those portions of the height profile that are generated either by cuts or by grooves. The isolated frequency range (i.e., passband) is realized by an appropriately selected bandwidth of the bandpass filter with suitable cutoff frequencies.The free frequency range is selected appropriately depending on the application, possibly also differently than described above, and thus also represents an adjustable parameter of the detection module, by adjusting which the identification of anomalies can be optimized.

[0039] In an advantageous embodiment, the preprocessing is further enhanced by one or more additional sensors for measuring vibrations on the track. A corresponding sensor measures vibrations of the track as it moves through the system. Furthermore, a vibration frequency is determined, and this vibration frequency is then expediently filtered out of the height profile, e.g., using a corresponding frequency filter or by selecting the passband of the aforementioned bandpass filter such that the vibration frequency is suppressed.

[0040] The filtering described above with one or more bandpass filters is optionally carried out within the detection module and then as part of the position detection.

[0041] Position detection with the detection module

[0042] The (first) threshold is suitably determined using a statistical method and, in a preferred embodiment, is a standard deviation of the elevation profile, e.g., 2o. The threshold depends in particular on the position and forms an envelope around the elevation profile. If the elevation profile breaks through this envelope, i.e. reaches the threshold, then an anomaly exists at this position and is identified as such. By multiplying the standard deviation by a factor, the threshold can be scaled as required, and this factor therefore represents a parameter for optimization. The threshold, and in particular its determination, benefit particularly from the above-described filtering of the elevation profile using a bandpass filter, as this limits the elevation profile to the anomaly being sought (section or Ri Hung), making its identification easier using statistical methods.The positions of the identified anomalies of the elevation profile are considered as candidates for recognized processing positions, in particular to be subsequently assigned to a given processing position using the assignment module.

[0043] To detect cuts, anomalies are expediently identified separately in both elevation profiles (upper and lower elevation profiles). An anomaly is recognized as a cut at an (actual) processing position if an anomaly has been identified at this processing position in both the upper and lower elevation profiles. Alternatively or additionally, an anomaly at an (actual) processing position is expediently recognized as a cut if there is a gap (NaN) in the upper or lower elevation profile at this processing position. This also includes the case where a gap exists in both elevation profiles at the same time. These three criteria are suitably concatenated (calculated) so that the presence of one of the criteria is sufficient to detect a cut.

[0044] To detect anomalies, especially creasing, a differential height profile is expediently calculated from the lower and upper height profiles, e.g., by subtracting them from one another. In other words, the height profile in which a number of anomalies are identified by identifying those sections of the height profile as an anomaly in which the height profile reaches a predetermined threshold value. It is assumed that an anomaly, specifically creasing, reduces the thickness of the web at an (actual) processing position. The differential height profile is then expediently filtered with a bandpass filter, as described above in the preprocessing section, making it particularly robust against measurement disturbances.If the difference height profile reaches the (first) threshold value at a processing position, an anomaly, especially a creasing, is detected at this processing position.

[0045] Using the two aforementioned steps for detecting cuts on the one hand and for detecting creasing on the other, the identified anomalies are appropriately divided into two groups before classification: a first group containing all anomalies identified as cuts, and a second group containing all anomalies identified as creasing. The anomalies are thus pre-sorted, so to speak. These two groups of anomalies are then further processed separately, i.e., classified and / or validated. In principle, the same classification module can be used for both groups, or a separate, specialized classification module can be used for each group.

[0046] Assignment of processing positions with the assignment module

[0047] As already mentioned above, each anomaly is assigned an actual processing position, which is automatically recognized by identifying the anomaly and is then also referred to as the recognized processing position. As part of the processing position assignment, the actual processing positions are assigned to the specified processing positions and, for this purpose, these are expediently compared. In a suitable embodiment, the assignment module generates a correlation table for this purpose, which contains the similarity of each actual processing position to each specified processing position. For example, the correlation table is a simple distance table containing the distances of each actual processing position to each specified processing position.The respective actual processing position is then assigned the specified processing position that is most similar to the actual processing position, e.g., that has the greatest similarity to it, e.g., that has the shortest distance to it. This advantageously also avoids double assignment, e.g., the assignment of multiple actual processing positions to only one specified processing position, or vice versa. Instead, each actual processing position is assigned at most one specified processing position, and vice versa.

[0048] By means of the comparison described, any excess processing positions are also advantageously detected, i.e., those processing positions that were detected but cannot be assigned to a predefined processing position in the order data. This is particularly the case if the number of predefined processing positions is fewer than the number of detected processing positions. Conversely, missing predefined processing positions are expediently detected in a similar manner if the number of predefined processing positions is greater than the number of detected processing positions. In this way, since not all detected / predefined processing positions can be assigned, excess or missing processing positions are detected accordingly.

[0049] Another advantageous embodiment is one in which a threshold value for the similarity is defined, which must be reached at least to perform a match. If the distance is used as a measure of similarity, the threshold value is a maximum distance (e.g., 5 mm). This avoids forced matchmaking and improves the detection of excess and missing machining positions.

[0050] In a practical embodiment, during validation, an anomaly is marked as an expected anomaly if a detected processing position could be assigned to a predetermined processing position, and otherwise, the anomaly is marked as an unexpected anomaly. Especially in the latter case, a suitable indication is issued that unexpected machining, damage, cuts, and / or creasing were detected. Conversely, during validation, a predetermined processing position is expediently marked as a found processing position if a predetermined processing position could be assigned to a detected processing position, and otherwise, the processing position is marked as not found.During validation, this is conveniently applied analogously to the cut or crease type, so that a given cut or crease type is marked as not found if it was not found by means of the classification.

[0051] Classification (characterization) with the classification module

[0052] The detected processing positions, which have optionally been assigned to predefined processing positions as described and preferably also validated in the process, are now classified using the classification module and each assigned to one of several cut or crease types. During classification, roughly speaking, a respective anomaly is assessed, particularly with regard to its shape and / or characteristics, and assigned to a cut or crease type based on this. Classification is an important component for quality assurance of the finished web (e.g., panels or sheets of corrugated board) and, in particular, also enables advantageous order tracking. For example, in the event of a complaint, the predefined cut and / or crease types from the order data are compared with the classified anomalies, thus identifying a possible reason for the complaint.The same applies analogously to the detected and specified processing positions. The classification described here advantageously allows the actual production result to be compared with the order data and verified, i.e., validated.

[0053] Since several different cutting and / or creasing elements are typically used, different cuts and / or creasing marks and thus also different anomalies are created in the web. Each cutting and creasing element has its own characteristic, which leads to a corresponding characteristic (shape and / or severity) of the anomaly created. This characteristic in particular is the subject of the classification described here. A cut usually has much simpler characteristics than a creasing mark. Since a cut is essentially a separation of the web at a specific processing position, it is initially sufficient to simply look for such a separation (e.g. by means of gaps in the measured values). Validation with the order data is therefore basically possible for cuts even without classification.It is also advantageously possible to distinguish between cuts and creasing without classification. Accordingly, in an advantageous embodiment, cuts are first distinguished from creasing as described, e.g., based on gaps in the height profile, and then only those anomalies are classified where creasing was detected at the processing positions.

[0054] The following primarily focuses on the classification of creasing. However, the same applies to cuts, as it is generally beneficial to consider the overall characteristics of a particular anomaly in order to draw conclusions, for example, whether a specific cut was executed correctly or to check whether the cutting tool used is still functioning properly or needs to be replaced.

[0055] In addition to the cut and crease types, it is also advantageous to define other types that can be recognized during classification. It is particularly advantageous to define a zero type that describes a section of the web with neither a cut nor a crease. This zero type makes it particularly easy to detect when no processing has been carried out at a specified processing position, i.e. when any creasing or cutting is missing there. In particular, when validating the processing positions, the height profile is then determined where processing is specified according to the order data but no anomaly was detected to determine whether the zero type is present at this position. Alternatively or additionally, the anomalies at any excess processing positions are classified by comparison with other types and then recognized, for example, as damage.The definition of other types for classification, especially the null type, is carried out in particular analogous to the options for defining cut and crease types described below, i.e., for example, by means of modeling, measurement, or machine learning. In a suitable embodiment, a respective anomaly for classification is compared with a representative of a respective cut or crease type, and the similarity (i.e., agreement) of the anomaly and the cut or crease type is calculated using a distance norm. The respective representative is, analogous to the anomaly, a series of data points, in particular a vector with the same length as the anomaly, and represents a typical or ideal shape of the corresponding cut or crease type. In other words: analogous to the anomaly, the representative is a set of data points (values ​​as a function of position).In this process, a difference is calculated from the data points of the anomaly and the representative at a respective position (i.e., component-wise in the case of a vector), and these differences are then summed. For example, the difference is calculated according to d = a - v, and thus the similarity is then calculated according to.

[0056] ||5||2= VS Ä) 2 . where a and v are the anomaly and the representative, respectively, and each have 2-N+1 data points (more precisely: values).

[0057] In a similarly suitable embodiment, the representative is not merely a one-dimensional vector as described, but has one or more additional dimensions in addition to the position, in particular the groove body spacing. Such a representative is then essentially a set of several one-dimensional representatives, which are parameterized by one or more parameters, such as the groove body spacing. The statements regarding the one-dimensional representative apply analogously to the comparison with the anomaly.

[0058] The existence of representatives is primarily important for the classifications described below using modeling and based on measured values. For classification using a learning machine, relevant information is incorporated through training the learning machine, which can in principle also be done using representatives as described.

[0059] Classification with modeling In a first embodiment for classification, the representatives are determined using a physical model that simulates the generation of creasing (and / or cuts) based on the geometry of a respective creasing body (or cutting body) of the cutting-Z-creasing unit. First, a reference model is created for each creasing type based on the associated creasing body, which reference model contains all relevant properties of the corresponding creasing. The reference model then serves as a representative for the classification using the classification module. A respective anomaly is then compared with various representatives and classified as the creasing type whose representative is most similar to the anomaly. In a suitable embodiment, the representative is generated from design data of the creasing body, e.g. by directly using its contour in the transverse direction as a representative.However, since the creasing body typically does not transfer its contour identically but only approximately to the web, it is advisable to take into account a corresponding transfer function when defining the representative or to determine the representative based on the design data or the contour of the creasing body using a finite element method, which then takes into account additional parameters relating to the process of creating a crease with the creasing body.

[0060] Modeling, i.e., the generation of representatives, is advantageously carried out using a learning machine. This will be described in more detail below.

[0061] Classification based on measured values

[0062] In a second embodiment of the classification, the representative of a respective crease type (or cut type) is determined based on previous measured values, suitably by recording a large number of height profiles for a respective crease type (or cut type) and then determining the representative for this crease type (or cut type) from these, in particular using statistical methods. This has the advantage that the representative is based more on the actual processing result and less on the crease body, as described above, whereby the similarity is potentially greater and the result more trustworthy. In addition to the crease type and the contour of the crease body, the parameters crease body spacing and / or web thickness, which are generally known and, for example, are part of the order data, are then taken into account. As a result, these parameters are not subject to any uncertainty.In a suitable embodiment, the representative is simply determined as the mean of the elevation profiles, i.e., the mean as a function of position along the elevation profile. It is advisable to first eliminate any outliers, e.g., by ignoring those elevation profiles that deviate from the mean by a certain (second) threshold, e.g., 2o, and then calculating the mean of the remaining elevation profiles again and using this as the representative.

[0063] In a suitable further development, the measured values ​​of different positions are weighted differently when calculating the similarity, e.g., with a weighting vector g containing a weight for each position. The weighting vector is multiplied element by element with the difference vector before the distance norm is then formed. Positions with a lower dispersion are expediently given a higher weight. The dispersion as a function of position is determined based on the measured height profiles. The dispersion is, for example, simply the width of an envelope at a respective position of all height profiles or the variance (i.e., the statistical quantity "variance"). In a particularly advantageous embodiment, the weighting vector is calculated using an exponential function as g = e~ b where b is the width of the envelope at the different positions.

[0064] Classification based on measured values ​​also enables continuous development of the representatives in the sense of training. In a suitable embodiment, the anomalies are each saved as a template and labeled by storing the cutting and / or creasing type to which this template belongs. In this way, known assignments of anomalies to cutting and / or creasing types are created, which are particularly suitable as representatives, for creating representatives, or as training data for training a learning machine (see below). The labeling is done, for example, via user input. The labeled templates therefore contain correct assignments of anomalies to cutting and / or creasing types and are then advantageously used as described above to continuously update a respective representative.

[0065] Classification with a learning machine

[0066] In a third embodiment for classification, a learning machine is used, which receives an anomaly (i.e., at least one or exactly one anomaly) as input parameter and then outputs the corresponding cut or crease type as output parameter. The learning machine is then part of the classification module. The learning machine is preferably a neural network with a large number of neurons, which is assumed below without loss of generality. In the previously described classification with modeling or based on measured values, a comparison of the anomaly is essentially carried out using a difference calculation and several representatives. Using this comparison, the anomaly is assigned to a corresponding cut or crease type.When classifying with a learning machine, the anomaly is also assigned to a corresponding cut or crease type. However, there is no direct comparison with representatives; rather, the corresponding information is encoded in the learning machine. For this purpose, the learning machine is appropriately trained, specifically using supervised learning, i.e., the learning machine is trained with known assignments of anomalies to cut and / or crease types. The representatives already mentioned are also generally suitable for training.

[0067] A respective anomaly is used as the input parameter for the learning machine. The input parameters form an input vector; preferably, the anomaly is represented as a vector (which then contains only the values, but not the positions of the anomaly data points) and used directly as the input vector. When classifying using a learning machine, it is advantageous to use both height profiles simultaneously as input parameters, i.e., not only passing an anomaly in the height profile on one side of the web to the learning machine, but also the corresponding section of the height profile (i.e., at the same processing positions) of the other side of the web, i.e., the opposite anomaly. Two such opposite anomalies form an anomaly pair. This doubles the number of input parameters and the length of the input vector accordingly. This is particularly advantageous for classifying creasing.

[0068] The cut and / or crease types are also referred to as target classes in the context of the learning machine. In a suitable embodiment, the target classes are made accessible to the learning machine using one-hot coding (or equivalently one-cold). This means in particular that the cut and / or crease types typically named with a character string (e.g. "three-point crease", "one-point crease", "point-point crease", "no crease (i.e., zero type)") are now each named with a vector which, depending on the cut and / or crease type, has a high bit at a single position and otherwise only low bits. The number of bits, i.e., the length of the vector, corresponds to the number of cut and / or crease types. Optionally, the aforementioned zero type and / or other types also exist. The output parameters of the neural network form, in particular, an output vector with the same length.

[0069] The so-called ReLu function (ReLu = rectified linear unit function), which is defined as relu(y) := max(y, 0), is preferably used as the activation function for the individual neurons of the neural network. Since this ReLu function only returns a zero value over a certain range of values, individual neurons responsible for detecting a different cut or groove type in the input vector are effectively switched off. The classification result is thus sharper. Furthermore, the ReLu function is advantageously differentiable (with the exception of the coordinate origin).

[0070] The neural network generally has an input layer for receiving the input vector and an output layer for outputting the output vector. The input layer and the output layer are connected via a number of hidden layers. In a suitable embodiment, the neural network has several, preferably three, hidden layers, with the number of neurons decreasing towards the output layer, preferably 300 in the first, 200 in the second, and 100 neurons in the third hidden layer. This type of embodiment of the hidden layers has proven particularly advantageous. For training the neural network, a gradient descent method, for example, is selected; in particular, a statistical measure, preferably cross entropy, is used as the cost function.

[0071] In a first advantageous embodiment, the learning machine is trained with correctly identified anomalies (e.g., based on the templates and / or representatives mentioned above) and is then a learning machine trained with supervised training. The length of the output vector corresponds in particular to the number of cut and / or crease types plus, if applicable, additional other types, so that the output vector then indicates how similar the anomaly (or anomaly pair, if both sides are input simultaneously) passed as an input vector is to a respective cut and / or crease type or other type.

[0072] In a second advantageous embodiment, the learning machine calculates the aforementioned respective representative of a crease type (or cut type), which is then used for classification as described above by comparing a respective anomaly with the representative. The representative is also referred to as a comparison signal. The learning machine therefore generates a representative for each cut and / or crease type and is thus used for modeling. The learning machine is expediently trained accordingly in advance. In a suitable embodiment, the learning machine described above is essentially used inversely. For training, the learning machine receives as input vector, in particular a crease type or cut type, e.g. appropriately marked templates, in particular from previous measurements, as well as optionally further parameters such as corrugated board thickness or crease spacing.The representative then serves as the output vector of the learning machine. The actual classification of an anomaly is then not performed by this learning machine, but rather as described above, possibly with a second learning machine. In other words, the learning machine calculates a representative, which is compared separately from this calculation of the representative with a measured height profile to determine its similarity to the representative and thus also the similarity of the crease type (or cut type) to the anomaly. In an advantageous embodiment, the learning machine calculates the representative as a function of the position on the one hand and additionally as a function of the crease body spacing.

[0073] Regardless of the specific implementation of the classification, the described optional detection of the zero type already makes it clear that, with the classification, any characteristics of the track beyond cuts and grooves can be recognized and classified, especially including damage to the track. For this purpose, only appropriate representatives or marked measured values ​​(i.e., correctly identified sections of the elevation profile) are used, and the classification is carried out as described. The same applies to the validation of the classification result described below.

[0074] Validation with the validation module

[0075] The previously described assignment and / or classification are advantageously additionally validated with the validation module. This means that the respective assignment of the anomalies at the detected processing positions to a cut and / or creasing type is compared with the order data, specifically with the cuts and / or creasing specified at the specified processing position, and in particular is also evaluated. Validation is carried out in particular based on the assignment of detected to specified processing positions, which was performed with the assignment module mentioned above, on the one hand, and on the classification of the anomalies, i.e., their assignment to specific cut and / or creasing types, on the other.

[0076] The validation module is used to suitably calculate the distances between a respective detected processing position and the assigned predefined processing position. A measure is then determined from all distances, e.g. the maximum or the mean value or something similar, and this measure is compared with a further (third) threshold value. If the measure does not reach the threshold value, the assignment is deemed valid (i.e., validated); otherwise, it is deemed invalid, and analogously, the order with regard to the processing position is deemed to have been correctly executed or to have been faulty. The validation module then outputs a corresponding message. Preferably, for each individual predefined processing position, it is output whether this is valid or not, i.e., whether it was carried out correctly or not.

[0077] Alternatively or additionally, the validation module is used in the same way to calculate the similarity of a respectively detected cut and / or creasing type with the cut and / or creasing type required in the job data at this processing position (i.e. the specified one). The similarity is then compared, preferably separately for each specified processing position, with a further (fourth) threshold value. If the similarity reaches the threshold value, the classification is deemed valid; otherwise, it is deemed invalid, and analogously, the job is deemed to have been correctly executed or to have errors with regard to the cut and / or creasing type. The validation module then issues a corresponding message. In principle, it is also advantageous to derive a single measure from the multiple similarities, e.g. the mean value, and to simply compare this with the threshold value in order to easily validate the job as a whole.

[0078] In the event of discrepancies between the detected processing positions and / or the detected cutting and / or creasing types and the order data, a message is expediently issued to an operator via an output element, e.g. as part of a human-machine interface (HMI), e.g. forwarded to a control station of the system. Preferably, such information about discrepancies is stored in a database with suitable key figures (timestamp, characteristics, etc.) for further analyses or preprocessors (machine learning, training for a learning machine, predictive and prescriptive analytics). Preferably, the system has an input element, e.g. also as part of the HMI, via which the output messages, the measured height profiles, the results of the position detection, the assignment, the classification and / or the validation can be combined with additional information, e.g.Key figures or indicators, especially "just-in-time" indicators (also known as "labeling"), can be used to supplement relevant information on production, process, and product quality or to make it available in the first place, and specifically to generate the templates mentioned above. This information is preferably stored in a database and used for further evaluations and learning processes, e.g., the training of the learning machine described above.

[0079] The above information is also suitable for forwarding to one or more higher-level systems so that it can be used outside of the operation of the system and the processing of the track, e.g. in production planning or the generation or adaptation of order data.

[0080] The database mentioned is implemented, for example, as a cloud solution and then separately from the system, or as part of it and thus as an on-premise solution. The process itself and, in particular, the computer program product are expediently executed on the system itself. Alternatively or additionally, the process itself and, in particular, the computer program product are implemented as a cloud solution to which the system is connected.

[0081] The computer program product according to the invention comprises instructions which, when executed by a computer, cause the computer to carry out, in particular, the method described above, partially or completely. In a suitable embodiment, the instructions, when executed by a computer, cause the computer to identify a number of anomalies in a measured height profile of a web by identifying those sections of the height profile as an anomaly in which the height profile reaches a predetermined threshold value, and then to classify a respective anomaly by assigning it to a cutting and / or creasing type (i.e., one of several cutting and / or creasing types) to which the anomaly is most similar.

[0082] In the following, exemplary embodiments of the invention are explained in more detail with reference to a drawing. In each case, the following schematically show:

[0083] Fig. 1 a method,

[0084] Fig. 2 a process diagram,

[0085] Fig. 3 a system,

[0086] Fig. 4 shows two elevation profiles,

[0087] Fig. 5 a creasing body in a cross-sectional view,

[0088] Fig. 6 two views of a sensor unit of the system from Fig. 3,

[0089] Fig. 7 a height profile and its spectrogram,

[0090] Fig. 8 to 10 each show a comparison of two height profiles with one representative each,

[0091] Fig. 11 a variety of height profiles for the same groove type,

[0092] Fig. 12 a learning machine,

[0093] Fig. 13 a representative.

[0094] Fig. 1 shows an embodiment of a method according to the invention. Fig. 2 then shows a process diagram with further details of the method. Fig. 3 shows an exemplary system 2 in which the method is implemented. First, in a first step S1, a web 4 is provided, into which a number of processing steps 8, 10 have been introduced by means of a cutting / creasing unit 6 of the system 2, each based on a predetermined cut type or crease type. “A number of” is understood here and generally to mean “one or more” or “at least one”. The processing steps 8, 10 are mechanical processing of the web 4 and in the present case each either a cut 8 or a crease 10 in the web 4. Each cut 8 was produced according to a predetermined cut type and each crease 10 according to a predetermined crease type.The cuts 8 and creasings 10 are used to cut the web 4 into individual panels or sheets and to form break, fold or crease edges of the panels or the subsequent sheets, e.g. in order to form them into a box, case or the like.

[0095] In the illustrated embodiment, the web 4 is both cut and scored using the cutting and creasing unit 6. The system 2 in this case is a corrugated board system for producing corrugated board, specifically sheets of corrugated board. The web 4 is made of paper and is a corrugated board web composed of multiple paper layers. As can be seen in Fig. 3, downstream of the cutting and creasing unit 6, a height profile 14, 16 of the web 4 is measured inline by means of a sensor unit 12, more precisely, two height profiles 14, 16: an upper height profile 14 on an upper side of the web 4 and a lower height profile 16 on an underside of the web 4.

[0096] The web 4 is processed by the cutting / creasing unit 6 at a number of processing positions 18 (i.e., cutting and / or creasing positions). In principle, an anomaly 20 is formed at each actual processing position 18. These actual processing positions 18 are detected together with the anomalies 20 and then also referred to as detected processing positions 18. The processing positions are specified by order data 22 for an order for the system 2. Accordingly, the order data 22 then contain specified processing positions 24. The cutting or creasing type to be performed at a respective processing position 24 is also specified in the order data 22, so that a specified cutting or creasing type is also assigned to each specified processing position 24 in the order data 22. The cutting / creasing unit 6 is controlled according to the order data 22.However, due to errors or inaccuracies, the cutting-creasing unit 6 does not necessarily produce the machining operations 8, 10 in such a way that the specified machining positions 24 correspond to the actual machining positions 18, and the required cutting or creasing type may not be reproduced exactly, or even an incorrect cutting or creasing type may be used.

[0097] In the method described here, in a second step S2, the sensor unit 12 measures a height profile 14, 16 on both sides of the web 4 and transversely to a conveying direction F of the web 4. An example of such height profiles 14, 16 is shown in Fig. 4, in which two anomalies 20 can be identified at each of two actual processing positions 18, i.e., a total of two pairs of anomalies, one of which belongs to a cut 8 and the other to a creasing 10, more precisely, a three-point creasing. Further height profiles 14, 16 are also shown in Figs. 7 to 10.

[0098] As can be seen in Fig. 4, in the respective height profile 14, 16, the processing steps 8, 10 by the cutting-grooving unit 6 can be identified as anomalies 20, i.e. as deviations from a normal state of the web. In a third step S3, the anomalies 20 are now identified in the height profiles 14, 16 by identifying those sections of the height profiles 14, 16 as an anomaly 20 in which the respective height profile 14, 16 reaches a predetermined (first) threshold value 26, i.e. exceeds or falls below it depending on the observation. As indicated in Fig. 4 by the crossed frames, the section is not just the area in which the height profile 14, 16 actually reaches the threshold value 26, but contains a certain allowance on both sides.

[0099] In Fig. 4, the reaching of the threshold value 26 is particularly clear for the creasing 10. The cut 8, on the other hand, is not necessarily detected as shown, but can also be identified in a different way, as described further below. In a fourth step S4, a respective anomaly 20 is then classified by assigning it to the cut and / or creasing type to which the anomaly 20 is most similar. Thus, it is not just checked whether or to what extent a respective processing 8, 10 corresponds to the cut or creasing type according to the order data 22, but it is checked which cut or creasing type is actually present (if necessary with specification of a confidence value for this). This classification thus differs from a simple target / actual comparison in that the anomaly 20 is not just compared with a single cut or creasing type, but with several different ones.In addition, by using the threshold value 26, anomalies 20 are actively searched for in the entire height profile 14, 16 and not only the height profile 14, 16 at the specified processing positions 24 is examined.

[0100] For cutting and / or creasing, the cutting / creasing unit 6 has one or more corresponding cutting bodies and / or creasing bodies 28. In both cases, the web 4 is mechanically processed; in the case of cutting, only separation, and in the case of creasing, deformation. A cutting body typically has at least one knife, which is inserted into the web 4 to produce a cut 8. Similarly, a creasing body 28 regularly has two profiled running surfaces 30, 32, which are arranged on opposite sides of the web 4 and interact. An example of a creasing body 28 for three-point creasing is shown in Fig. 5. By shifting the running surfaces 30, 32 of a creasing body 28 in the transverse direction Q, so-called offset creasing is also possible. This shift is also referred to as offset.A further displacement is also regularly possible perpendicular to both the transverse direction Q and the conveying direction F in order to adjust the so-called creasing body spacing 34.

[0101] Various aspects and embodiments of the method are described in detail below, for which specific reference is made to Fig. 2. To evaluate a height profile 14, 16 and to detect anomalies 20 therein, the height profile 14, 16 is first prepared and filtered as part of a preprocessing 36, which in this case uses digital signal processing methods. Subsequently, recognized processing positions 18 in the height profile 14, 16 are identified by means of a position detection 38, which in this case applies statistical methods; accordingly, the anomalies 20 are also identified here. Subsequently, an assignment 40 of the recognized processing positions 18 to the predetermined processing positions 24 from the order data 22 takes place. The position detection 38 and the assignment 40 are shown in Fig.2 is carried out separately once for cuts 8 and once for creasing 10, a division of the anomalies 20 into two corresponding groups is already carried out by the preprocessing 36.

[0102] For further analysis, the method described here uses the aforementioned classification 42, by means of which the height profile 14, 16 is classified section by section at the detected processing positions 18 with regard to the processing operations 8, 10 carried out, i.e. the anomalies 20 are classified. In the exemplary embodiment in Fig. 2, only the creasing 10 is classified; a classification of the cuts 8 does not take place, but is also optionally implemented. The cuts 8, or more precisely their processing positions 18, are here only validated with a validation 44. Optionally, a validation of the result of the classification 42 is also carried out; this is not explicitly shown in Fig. 2, but is possible analogously to the validation 44 already mentioned.During the validation 44 of the processing positions 18, a message is issued, the system 2 is stopped, or another suitable measure is initiated as soon as it is determined that the recognized processing positions 18 do not match the specified processing positions 24 from the order data 22 within a given tolerance. The validation of the result of the classification 42 based on the order data 22 is carried out analogously to the validation 44 of the processing positions 18, i.e., it is checked whether the correct cut or creasing type was used and / or was correctly executed and whether the cuts 8 and creasing 10 were produced with the correct cut or creasing body 28.As soon as it is determined during this validation that the recognized processing operations 8, 10 do not correspond within a given tolerance with the specified cutting and / or creasing types from the order data 22, a corresponding message is issued, system 2 is stopped, or another suitable measure is initiated.

[0103] Measurement (first step S1)

[0104] The height profile 14, 16 is measured, for example, with a sensor unit 12 as shown in Fig. 6. In Fig. 6, the sensor unit 12 is shown on the left as viewed in the conveying direction F, and on the right in the transverse direction Q perpendicular to the conveying direction F. The sensor unit 12 has two sensors 46, here distance sensors, specifically laser triangulation sensors, which can be moved transversely to the web 4 and measure a distance in the direction of the web 4 (i.e. towards it), i.e. the distance between sensor 46 and web 4. The sensor 46 outputs a measured value (distance) for each measuring position transversely across the web 4, which distance varies depending on the nature of the surface of the web 4. As a result, each sensor 46 outputs a height profile 14, 16 which contains a series of measured values. Each measured value is also assigned a position P (measurement position) along the web 4. A measured value forms a data point with the associated position P.The sensor unit 12 is arranged directly downstream of the cutting-creasing unit 6. Furthermore, the sensor unit 12 shown here has a pressure roller 48 to press the web 4 against a support 50 as it passes through the sensor unit 12 and during measurement of the height profile 14, 16, and to suppress any curvature of the web 4 as much as possible, at least during the measurement.

[0105] Preprocessing (second step S2)

[0106] During preprocessing 36, the measured height profile 14, 16 is suitably prepared for the subsequent position detection 38. For example, it is possible that the measuring range is exceeded when measuring the height profile 14, 16 and no distance can be measured, which is regularly the case with a cut 8 because the path 4 is interrupted at its processing position 18. Accordingly, if the measuring range is exceeded, a valid measured value may not be generated; instead, a NaN (not a number) is returned as the measured value, and a corresponding gap is found in the height profile 14, 16 at the associated position P. During preprocessing 36, such gaps in the height profile 14, 16 are now specifically removed in a block 52 in order to obtain a gap-free height profile 14, 16. However, the gaps are not simply omitted, but replaced with suitable values.In addition, in block 52, the elevation profile 14, 16 is mapped onto an equidistant grid. New data points with new (measured) values ​​and new positions P are generated from the measured values ​​and measurement positions, which are distributed along an equidistant grid.

[0107] The preprocessing 36 in Fig. 2 further comprises a further block 54 in which the gaps and their positions P are identified.

[0108] The anomalies 20 are also detectable in the frequency domain, specifically in the spectrogram of the height profile 14, 16. An exemplary spectrogram is shown in the lower part of Fig. 7, and the corresponding height profile 14, 16 is shown in the upper part. In Fig. 7, the data point density of the height profile 14, 16 is, for example, ten data points per millimeter, whereby the spectrogram in a frequency range from 0 to 5 mm' 1(in the vertical direction). As is clear from Fig. 7, continuous lines appear in the spectrogram at exactly the same positions P as actually executed cuts 8 (parallel to the vertical frequency axis). This means that a cut 8 is identifiable across the entire frequency spectrum and that by filtering out a suitable sub-range from the frequency spectrum (i.e., by selecting a specific frequency range 56) in which only the frequencies of a cut 8 are contained, the influence of this cut 8 on the measured height profile 14, 16 can be displayed in isolation, thus enabling identification of the cut 8. In one possible embodiment, a corresponding frequency range 56 is therefore isolated using a bandpass filter B. In Fig. 7, a suitable frequency range 56 ranges from 1.5 mm -1 up to 5 mm in particular -1, since in this frequency range 56 only the effect of the cuts 8 is visible in the spectrogram, and all other influences on the height profile 14, 16 lie entirely or at least predominantly below this frequency range 56. Accordingly, the upper limit is not necessarily important; rather, it is determined by the data point density. Similarly, at the processing positions 18 of grooves 10 in the spectrogram, anomalies in the frequency range 58 of 0 mm are also observed. -1 up to about 0.1 mm -1 recognizable, so that in a possible embodiment a corresponding frequency range 58 is released using a bandpass filter B. In Fig. 7 a suitable frequency range 58 ranges from 0.001 mm -1 (to suppress the DC component in the height profile 14, 16 at 0 mm -1 ) up to 0.1 mm' 1 .

[0109] Position detection (third step S3)

[0110] The threshold value 26 in this case is a standard deviation of the respective height profile 1, 16. The standard deviation forms a threshold value 26, which depends on the position P and, as it were, forms an envelope (cf. Fig. 4) around the height profile 14, 16. If the height profile 14, 16 breaks through this envelope, i.e. reaches the threshold value 26, then an anomaly 20 is present at this position P, which is also identified as such. This identification of an anomaly 20 by reaching the threshold value 26 takes place in Fig. 2 in block 60. The threshold value 26 and specifically its determination benefit from the above-described filtering of the height profile 14, 16 using a bandpass filter, since this limits the height profile 14, 16 to the respective anomaly 20 sought (cuts 8 or grooves 16) and facilitates its identification using statistical methods.The positions P of the identified anomalies 20 of the elevation profile 14, 16 are also considered as candidates for detected processing positions 18 in order to be subsequently assigned to a predetermined processing position 24 by means of the assignment 40.

[0111] To detect cuts 8, anomalies 20 are identified separately in both height profiles 14, 16 in block 60, and an anomaly 20 is then detected as a cut 8 at a processing position 18 if an anomaly 20 was identified at this processing position 18 in both the upper and lower height profiles 14, 16. Additionally, in block 62 (detection based on a gap), an anomaly 20 is detected as a cut 8 at a processing position 18 if there is a gap in the upper or lower height profiles 14, 16 at this processing position 18. For this purpose, the positions P with gaps identified there are transferred from block 54 to block 62. The aforementioned criteria are chained together so that the presence of one of the criteria is sufficient to detect a cut 8.

[0112] To detect creasing 10, a differential height profile is calculated from the lower and upper height profiles 14, 16 and then filtered with a bandpass filter B as described above. If the differential height profile reaches the threshold value 26 at a processing position 18 during identification in block 60, a creasing 10 is detected at this processing position 18.

[0113] Using the two aforementioned steps for detecting cuts 8 on the one hand and creasing 10 on the other, the identified anomalies 20 are divided into two groups before classification 42: a first group containing all anomalies 20 that were generally identified as cuts 8 (left in Fig. 2), and a second group that analogously contains all anomalies 20 that were identified as creasing 10 (right in Fig. 2). The anomalies 20 are thus pre-sorted. In Fig. 2, only the group of anomalies 20 that were identified as creasing 10 is then classified in order to assign a specific creasing type to each of these anomalies 20.

[0114] Assignment of processing positions

[0115] Each anomaly 20 is assigned an actual processing position 18, which is recognized by identifying the anomaly 20 and is then also referred to as the recognized processing position 18. As part of the assignment 40, the actual processing positions 18 are assigned to the predetermined processing positions 24 and these are compared. For this purpose, for example, a correlation table is generated which contains the similarity of each actual processing position 18 to each predetermined processing position 24. For example, the correlation table is a simple distance table which contains the distances of each actual processing position 18 to each predetermined processing position 24. A respective actual processing position 18 is then assigned that predetermined processing position 24 which is most similar to the actual processing position 18, i.e., has the greatest similarity to it, e.g.has the smallest distance to it. This avoids double assignment, so that each actual processing position 18 is assigned at most one predefined processing position 24, and vice versa.

[0116] By means of the comparison, any excess processing positions are also detected, i.e. those processing positions 18 which were detected but cannot be assigned to a predetermined processing position 24 in the order data 22. These can also be, for example, anomalies 20 which were not generated by the cutting / grooving unit 6. Conversely, missing predetermined processing positions 24 are detected in a similar way. For this purpose, block 64 is specifically used in Fig. 2, which generates feedback for the assignment 40 in order to adapt one or more parameters (e.g. threshold value 26, limit value (see below), search area in the height profile 14, 16) to identify the anomalies 20 if necessary. In the present case, a limit value for the similarity is also defined during the assignment 40, which limit value must at least be reached in order to carry out an assignment.If the distance is used as a measure of similarity, the threshold is accordingly a maximum distance (e.g.

[0117] 5 mm). This avoids forced assignment.

[0118] Classification (fourth step S4)

[0119] The detected processing positions 18, which were optionally also assigned to predefined processing positions 24 as described, are now classified and thus each assigned to one of several cut types or creasing types. Since several different creasing elements 28 are typically used, correspondingly different creasing patterns 10 and thus, depending on the order, also different anomalies 20 are generated (analogous for cuts 8). Each creasing element 28 has its own characteristic, which leads to a corresponding characteristic of the anomaly 20 generated thereby. A cut 8 usually has a significantly simpler characteristic than a creasing pattern 10. Since a cut 8 is essentially a separation of the web 4 at a specific processing position 18, it is initially sufficient to simply search for such a separation (e.g., by means of gaps in the measured values). Validation with the order data 22 is - as shown in Fig.2 - thus, for cuts 8, classification is generally possible without classification. It is also generally possible to distinguish cuts 8 from creasing 10 without classification 42. Accordingly, in the embodiment shown here, cuts 8 are first distinguished from creasing 10, and subsequently, only those anomalies 20 are classified at whose processing positions 18 creasing 10 were detected. Therefore, the classification 42 of creasing 10 is primarily described below. However, the explanations apply analogously to cuts 10.

[0120] For classification 42, in one possible embodiment, a respective anomaly 20 is compared with a representative 66 of a respective crease type, and the similarity (i.e., agreement) of the anomaly 20 and the crease type is calculated using a distance norm. This is shown as an example in Figs. 8, 9, and 10, each of which shows in the upper part a comparison of two height profiles 14, 16, each with three different representatives 66 (i.e., a total of six representatives 66), with two representatives 66 always forming a pair and being assigned to a specific crease type. The height profiles 14, 16 are the same in all three Figs. 8, 9, and 10. The creasing types shown are: point-flat creasing in Fig. 8, point-point creasing in Fig. 9, three-point creasing in Fig. 10. In the lower part, the pairs of results of the comparisons are shown as two bars, namely three times a comparison of the upper and lower height profile 14, 16 with a corresponding representative 66.A difference was calculated from the data points of the respective anomaly 20 and the respective representative 66 at a respective position P (i.e., component by component in the case of a vector), and these differences are then summed for each of the two height profiles 14, 16. The two sums, once for the upper and once for the lower height profile 14, 16, are then displayed as bars. A comparison of the three pairs of bars shows that the three-point profile in Fig. 10 has the lowest total sum (sum of both sums of the differences), i.e., the greatest similarity exists. Both anomalies 20 (i.e., the anomaly pair) are therefore assigned to the three-point creasing type.

[0121] Classification with modeling In one possible embodiment for classification 42, the representatives 66 are determined by means of a physical model which simulates the generation of creasing 10 based on a geometry of a respective creasing body 28. First, a reference model is created for each creasing type on the basis of the associated creasing body 28, which reference model contains all relevant properties of the corresponding creasing 10. The reference model then serves as representative 66 for classification 42. A respective anomaly 20 is then compared with various representatives 66 (see Fig. 8, 9 and 10) and classified as the creasing type whose representative 66 is most similar to anomaly 20. In one possible embodiment, representative 66 is generated from design data of creasing body 28, e.g. by directly using its contour in the transverse direction Q as representative 66, as can be seen in Fig. 5.

[0122] Classification based on measured values

[0123] In another possible embodiment of the classification 42, the representative 66 of a respective crease type is determined on the basis of previous measured values, e.g. by recording a plurality of height profiles 14, 16 for a respective crease type, e.g. as shown in Fig. 11, and then determining the representative 66 for this crease type from these, e.g. simply as the mean value of the height profiles 14, 16. Optionally, any outliers are eliminated beforehand, e.g. by ignoring those height profiles 14, 16 which deviate from the mean value by a certain threshold value and then again calculating the mean value of the remaining height profiles 14, 16 and using this as the representative 66. Likewise optionally, the measured values ​​of different positions P are weighted differently when calculating the similarity before the distance norm is formed as described above.In particular, those positions P which have a lower scatter are given greater weight; in Fig. 11 these are mainly the positions P at which the upper height profile 14 assumes a minimum and the lower height profile 16 a maximum (in Fig. 11 two maxima).

[0124] Classification with a learning machine

[0125] In another possible embodiment for classification 42, a learning machine 68 is used, which receives an anomaly 20 as an input parameter and then outputs the associated groove type as an output parameter. An exemplary embodiment of the learning machine 68 is shown in Fig. 12, in which the learning machine 68 is a neural network with a plurality of neurons 70. In this case, the learning machine 68 is trained using supervised training.

[0126] A respective anomaly 20, i.e., a section of the elevation profile 14, 16, is used as an input parameter for the learning machine 68. The input parameters form an input vector. In this case, both elevation profiles 14, 16 are used simultaneously, i.e., not only an anomaly 20 in the elevation profile 14, 16 on one side of the web 4, but also the opposite anomaly 20. This doubles the number of input parameters and the length of the input vector. The cut and / or crease types are made accessible to the learning machine 68 via one-hot coding. The output parameters of the neural network then form an output vector with a length corresponding to the number of crease types. The ReLu function, for example, is used as the activation function for the individual neurons of the neural network.The neural network has an input layer 72 for receiving the input vector and an output layer 74 for outputting the output vector. The input layer 72 and the output layer 74 are connected via a number of hidden layers 76, here three, e.g., with a decreasing number of neurons 70 toward the output layer 74. The learning machine 68 is trained, for example, with correctly identified anomalies 20 and is therefore a supervised learning machine 68.

[0127] A configuration is also possible in which the aforementioned respective representative 66 of a crease type is calculated with the learning machine 68, which is then used for classification 40 as described above by comparing a respective anomaly 20 with the representative 66. The learning machine 68 therefore generates a representative 66 for each crease type and is thus used for modeling. Fig. 13 shows a representative 66 calculated in this way, which is a function of the position P on the one hand and additionally a function of the crease body spacing 34. Validation

[0128] The validation 44 is based on the assignment 40 of detected to predetermined processing positions 18, 24, which was carried out as further described, on the one hand (shown in Fig. 2 for the cut types) or on the other hand on the classification 42 of the anomalies 20, ie their assignment to specific cut and / or creasing types (not explicitly shown in Fig. 2).

[0129] During validation 44 shown in Fig. 2, the distances between a respective detected processing position 18 and the associated predefined processing position 24 are calculated. Then, a measure is determined from all the distances, e.g., the maximum or the mean or something similar, and this measure is compared with a threshold value. If the measure does not reach the threshold value, the assignment 40 is considered valid (i.e., validated); otherwise, it is considered invalid, and, analogously, the order with respect to the processing position 18, 24 is considered to have been executed correctly or incorrectly.

[0130] Alternatively or additionally, the similarity of a respectively detected cut and / or creasing type with the cut and / or creasing type required (i.e., the specified) in the order data 22 at this processing position 18 is calculated analogously, and then the similarity is compared separately for each specified processing position 24 with a threshold value. If the similarity reaches the threshold value, the classification 42 is considered valid; otherwise, it is considered invalid, and analogously, the order is considered to have been properly executed or faulty with regard to the cut and / or creasing type.

[0131] In the event of discrepancies between the detected processing positions 24 and / or the detected cutting and creasing types and the order data 22, a note is output to an operator via an output element, e.g. as part of a human-machine interface 78, e.g. forwarded to a control station of system 2. The system 2 shown here also has, as part of the human-machine interface 78, an input element (not explicitly shown), via which the output notes, the measured height profiles 14, 16, the results of the position detection 36, the assignment 40, the classification 42 and / or the validation 44 can be marked by an operator with additional information, e.g. key figures or identifiers, in particular "just-in-time", in order to supplement relevant information on production, process and product quality or to make it available in the first place and specifically to generate the templates mentioned above.This information is stored in a database 80 and used for further evaluations and learning processes.

[0132] The system 2 further comprises a control unit 82 which is designed to carry out the method as described.

[0133] List of reference symbols

[0134] 2 Appendix

[0135] 4 lane

[0136] 6 Cutting-grooving unit

[0137] 8 Editing, cutting

[0138] 10 Processing, creasing

[0139] 12 Sensor unit

[0140] 14 upper elevation profile

[0141] 16 lower elevation profile

[0142] 18 actual / recognized machining position

[0143] 20 Anomaly

[0144] 22 Order data

[0145] 24 predefined processing positions

[0146] 26 (first) threshold

[0147] 28 creasing bodies

[0148] 30 profiled tread

[0149] 32 profiled tread

[0150] 34 Groove spacing

[0151] 36 Preprocessing

[0152] 38 Position detection

[0153] 40 Allocation (of processing positions)

[0154] 42 Classification

[0155] 44 Validation

[0156] 46 Sensor

[0157] 48 pressure roller

[0158] 50th edition

[0159] 52 Block (preprocessing)

[0160] 54 further blocks (of preprocessing)

[0161] 56 frequency range (for cuts)

[0162] 58 frequency range (for grooving)

[0163] 60 Block (identification of an anomaly by reaching the threshold)

[0164] 62 Block (detection based on a gap) 64 Block

[0165] 66 Representative

[0166] 68 Learning Machine

[0167] 70 Neuron 72 input layer

[0168] 74 output layers

[0169] 76 hidden layers

[0170] 78 Human-Machine Interface

[0171] 80 Database 82 Control unit

[0172] B Bandpass filter

[0173] F Conveying direction

[0174] P Position

[0175] Q Transverse direction S1 first step (measurement)

[0176] 52 second step (preprocessing)

[0177] 53 third step

[0178] 54 fourth step

Claims

Claims proceedings, - wherein a web (4) is provided into which a number of processing steps (8, 10) have been carried out by means of a cutting-grooving unit (6), each based on a predetermined processing position (24) and a predetermined cutting or grooving type, - wherein a height profile (14, 16) of the web (4) is measured by a sensor unit (12) on at least one side of the web (4) and transversely to a conveying direction (F) of the web (4), - wherein a number of anomalies (20) and thus also actual processing positions (18) are identified in the height profile (14, 16) by identifying those sections of the height profile (14, 16) as an anomaly (20) in which the height profile (14, 16) reaches a predetermined threshold value (26), - wherein a respective anomaly (20) is classified by means of a classification (42) in that the anomaly (20) is assigned to the cut type or crease type to which the anomaly (20) is most similar. The method according to claim 1, wherein a learning machine (68) is used for the classification (42), which receives an anomaly (20) as an input parameter and then outputs the associated cut or crease type as an output parameter. The method according to claim 1, wherein a respective representative (66) of a crease type or cut type is calculated with a learning machine (68), which is then used for classification by comparing a respective anomaly (20) with the representative (66). The method according to claim 2 or 3, wherein the learning machine (68) is a neural network with a plurality of neurons (70), the ReLu function being used as the activation function for the individual neurons (70), the neural network having an input layer (72) and an output layer (74), the neural network having a plurality of hidden layers (76), with the number of neurons (70) decreasing towards the output layer (74).Method according to claim 1, wherein for the classification (42) a respective anomaly (20) is compared with a representative (66) of a respective cut or crease type and the similarity of the anomaly (20) and the cut or crease type is calculated by means of a distance standard, wherein the representative (66) of a respective cut or crease type is determined on the basis of previous measured values ​​by recording a plurality of height profiles (14, 16) for a respective cut or crease type and then determining the representative (66) for this cut or crease type from these.Method according to claim 1, wherein for the classification (42) a respective anomaly (20) is compared with a representative (66) of a respective cut or creasing type, and the similarity of the anomaly (20) and the cut or creasing type is calculated using a distance standard, wherein the representatives (66) are determined using a physical model which simulates the generation of creasing (10) and / or cuts (8) based on a geometry of a respective creasing body (28) or cutting body of the cutting-creasing unit (6). Method according to one of claims 1 to 6, wherein the similarity of a respectively detected cut and / or creasing type with the respectively predetermined cut and / or creasing type is calculated and. then, the similarity for each predetermined processing position (24) is compared separately with a further threshold value, wherein the classification (42) is considered valid if the similarity reaches the threshold value, and otherwise invalid. Method according to one of claims 1 to 7, wherein a respective actual processing position (18) is assigned that predetermined processing position (24) which is most similar to the actual processing position (18). Method according to claim 8, wherein a limit value for the similarity is specified, which limit value must at least be reached in order to perform an assignment (40) in order to avoid a forced assignment.Method according to claim 8 or 9, wherein distances between a respective actual processing position (18) and the assigned predetermined processing position (24) are calculated, wherein a measure is determined from all distances and this measure is compared with a further threshold value, wherein the assignment (40) is considered valid if the measure does not reach this further threshold value, and otherwise invalid. Method according to one of claims 1 to 10, wherein the height profile (14, 16), before the anomalies (20) are identified therein, is filtered with a bandpass filter (B), wherein the bandpass filter (B) exposes a frequency range (56, 58) which is selected such that those portions of the height profile (14, 16) are specifically retained which are generated either by cuts (8) or by creasing (10). Method according to one of claims 1 to 11,. wherein the threshold value (26) is a standard deviation of the height profile (14, 16). Method according to one of claims 1 to 12, wherein anomalies (20) are identified separately in an upper height profile (14) on an upper side of the web (4) and in a lower height profile (16) on a lower side of the web (4), wherein an anomaly (20) at a processing position (18) is recognized as a cut (8) if an anomaly (20) has been identified in both the upper and the lower height profiles (14, 16) at this processing position (18), and / or wherein an anomaly (20) at a processing position (18) is recognized as a cut (8) if a gap is present in the upper or lower height profile (14, 16) at this processing position (18).Method according to one of claims 1 to 13, wherein a differential height profile is calculated from an upper height profile (14) on an upper side of the web (4) and a lower height profile (16) on an underside of the web (4), wherein, if the differential height profile reaches the threshold value (26) at a processing position (18), an anomaly (20) is detected at this processing position (18). Method according to one of claims 1 to 14, wherein the cutting and creasing unit (6) and the sensor unit (12) are part of a corrugated cardboard plant, and wherein the web (4) is a corrugated cardboard web that is produced and processed by the corrugated cardboard plant, wherein the height profile (14, 16) is measured inline in the corrugated cardboard plant and downstream of the cutting and creasing unit (6). Plant (2) comprising a control unit (82) designed to carry out a method according to one of claims 1 to 15. A computer program product that contains instructions that, when executed by a computer, cause it to - to identify a number of anomalies (20) and thus also actual processing positions (18) in a measured height profile (14, 16) of a path (4) by identifying those sections of the height profile (14, 16) as an anomaly (20) in which the height profile (14, 16) reaches a predetermined threshold value (26), - to classify a respective anomaly (20) by means of a classification (42) by assigning the anomaly (20) to a cutting and / or creasing type to which the anomaly (20) is most similar.