METHOD FOR DETECTING A FAULT IN A MACHINING PROCESS AND MACHINING MACHINE

Description

[0001] The present invention relates to a method for detecting at least one fault in a machining process, in particular in a cutting process, comprising: in particular cutting machining of a workpiece by moving a machining tool, in particular a laser machining head, and the workpiece relative to one another, recording an image of an area to be monitored on the workpiece, which area includes an interaction area of ​​the machining tool with the workpiece, and evaluating the image of the area to be monitored to detect the at least one fault in the machining process.The invention also relates to a processing machine, comprising: a processing tool, in particular a laser processing head, for the in particular cutting processing of a workpiece, a movement device for moving the processing tool and the workpiece relative to one another, an image capture device for recording an image of an area to be monitored on the workpiece, which area comprises an interaction area of ​​the processing tool, in particular the laser processing head, with the workpiece, and an evaluation device which is configured to detect at least one disturbance of the processing process based on the evaluation of the image of the area to be monitored.

[0002] A disruption in a machining process involving a cutting beam, such as a plasma or laser beam, can occur, for example, in a surface-processing machine such as a 2D laser cutting machine, in the form of a cut break. In a cut break, the machining beam no longer completely cuts through the (metallic) workpiece because the energy per unit length is insufficient to melt the entire cutting gap volume.

[0003] DE 10 2013 209 526 B4 proposes, in order to detect a cut break during cutting with a high-energy beam, in particular with a laser beam, to acquire an image of a region of the workpiece to be monitored, which region includes an interaction region of the high-energy beam with the workpiece. The image is evaluated to detect slag droplets at an end of the interaction region opposite a cutting front, and the cut break is detected based on the appearance of the slag droplets. The slag droplets can be detected based on a change in the geometry of the interaction region at its end opposite the cutting front and / or based on the appearance of a local intensity minimum in the image of the end of the interaction region opposite the cutting front.

[0004] WO 2016 / 181359 A1 describes that a detector in the form of a photodiode can be used to detect a temporary cut break. This detector is designed to be trailing, i.e., looking backward into the cutting gap, and whose observation direction is oriented at a polar angle greater than 5° with respect to the optical axis of a working laser beam. It also describes that pseudo-errors can occur when detecting a cut break because other effects of a similar magnitude, such as increased cutting speed or a wide cutting gap, can overlap with an effect caused by a cut break.

[0005] DE 10 2018 217 526 A1 discloses a method in which at least one parameter for the process quality of the machining process is determined based on a monitored area on a workpiece, which may include an interaction area of ​​a machining area with the workpiece. In the method, at least one position-dependent parameter for the process quality is determined based on a plurality of measured values ​​of the at least one parameter at the same machining position and / or at least one direction-dependent parameter for the process quality is determined based on a plurality of measured values ​​of the at least one parameter in the same machining direction.

[0006] Using the method described there, interference position areas can be identified that depend on the machining position or the position in the working area, but essentially do not depend on the geometry of the contour to be cut, the type of machining process (e.g. flame cutting or fusion cutting) and the machining parameters. Examples of position-dependent interference include support bars arranged in the working area, which, if contaminated with slag, can negatively influence the machining process and thus the cutting result. Interference in the form of incorrect cuts, cut breaks, slag adhesion, burr or spatter formation and burn-out of the cut edges are just a few examples of how contaminated support bars can negatively influence the cutting process.

[0007] DE 10 2017 210 182 A1 discloses a method in which the actual state of a transverse extent of a support strip is detected by means of a detection device. The actual state of the support strip can be detected, for example, using an optical method, i.e. a method of optical metrology and / or an imaging method. In the first case, the support strip can be arranged between an optical emitter and an optical sensor of the detection device, and the transverse extent of the support strip can be imaged onto the optical sensor by means of an optical beam. In the second case, the support strip and a camera of the detection device can be opposite one another, and an image of the transverse extent of the support strip is recorded by means of the camera.

[0008] A method in which an actual value of a support bar contour is detected by means of an optical sensor in order to check a workpiece support for the presence of deposits to be removed is described in EP 2 082 813 A1.

[0009] US 2014 / 175 071 A1 describes a method in which brightness profiles of the workpiece surface are recorded using CMOS sensors in the range between 450 nm and 800 nm. These profiles are compared with models to determine the processing quality. Aufgabe der Erfindung

[0010] The invention is based on the object of providing a method and a processing machine for the reliable detection of faults in a processing process. Gegenstand der Erfindung

[0011] This object is achieved according to the invention by a method of the type mentioned at the outset, which is characterized in that, in order to detect the disturbance when evaluating the image, the presence or absence of a local intensity drop of an intensity profile within the interaction region along a feed direction of the machining process is detected and that the detection of the local intensity drop at a machining position is assigned to a position-dependent disturbance of the machining process at the machining position.

[0012] During the machining process, e.g. during laser cutting, an image of the interaction area (process emission zone or process lights) is captured in real time using an imaging sensor or image acquisition device (e.g. with a frequency of more than 400 Hz) and evaluated in real time using suitable image processing algorithms in order to extract or analyze features of the interaction area that enable the detection of disturbances in the machining process.

[0013] A plausibility check is essential for detecting faults based on features extracted during the evaluation of the image of the interaction area in order to reliably identify the presence of a fault or the type of fault. It has been shown that when detecting a fault in the machining process in the form of a cut break based solely on geometric features of the interaction area, for example the length of the interaction area, it is not possible to clearly distinguish between a real cut break and a pseudo-cut break. Such a pseudo-cut break can occur if other faults occur during the cutting process, for example, due to support webs of a workpiece storage or due to the type of workpiece (e.g.a double sheet), which change the geometric characteristics of the interaction area in the same way as would be the case with a cut tear.

[0014] In these cases, i.e., in a "disturbed" cutting process, reliable detection of a cut break cannot usually be achieved based on geometric features of the interaction area alone. This means that a cut break may be detected even though no cut break occurred during the cutting process. However, the distinction as to whether a cut break has occurred can be made based on the intensity profile of the interaction area along the feed or cutting direction, i.e., along the direction of the current relative movement between the workpiece and the machining tool. However, to detect a cut break, another geometric feature of the interaction area is usually required, as described in more detail below.

[0015] The local intensity drop in the intensity curve typically occurs near the expected (nominal) end of the cutting front or the interaction region in an undisturbed cutting process. However, in the event of a cut break, the length of the interaction region is extended in the feed direction, so that the local intensity drop does not occur at the end of the interaction region, but rather within the interaction region. This means that both before and after the intensity drop or a local intensity minimum, the intensity is increased along the feed direction compared to the local intensity minimum.

[0016] The presence of a local intensity drop can be detected, for example, if the intensity curve within the interaction region exhibits a local minimum whose intensity value is below a predefined percentage, e.g., below 80%, of a maximum intensity value within the interaction region. The intensity value of the intensity curve increases in the feed direction before and after the local intensity minimum until the intensity curve drops steeply at the front or rear end of the interaction region, respectively. The percentage of the maximum intensity value at which a local intensity drop is detected is not necessarily predetermined but can be determined depending on the current operating point.

[0017] To detect the presence or absence of a local intensity drop, other criteria can be used alternatively or additionally. For example, the presence or absence of a local intensity drop can be detected based on a gradient of the intensity curve, i.e. based on the derivative of the intensity curve along the feed direction. The gradient of the intensity curve has a zero crossing at a local minimum of the intensity curve within the interaction region, at which the gradient increases from negative values ​​to positive values. The zero crossing of the gradient lies between a local minimum and a local maximum of the gradient. The local intensity drop can also be detected based on the curve of the gradient, for example based on the absolute value of the local minimum or the local maximum, which represents a measure of the steepness of the local intensity drop.For example, the absolute value of the gradient at the location of the local minimum or the local maximum or the increase of the gradient from the local minimum to the local maximum can be used for this purpose and compared with a threshold value.

[0018] In one variant, a cut break is recognized as a disturbance during cutting of the workpiece only if the absence of a local intensity drop is detected within the interaction area. This exploits the fact that in a real cut break, the intensity curve along the cutting direction exhibits no local intensity drop ("discontinuity") and thus no local intensity minimum, or only a slightly deviating intensity value, from a maximum intensity value. This is typically the case with a disturbance in the form of a support web or the presence of a double sheet, since such a "disturbance" in the material flow generates lower process emission at a particular point in the interaction area. Based on the intensity curve within the interaction area, a distinction can therefore be made between a real cut break and a pseudo-cut break.

[0019] In one variant, at least one geometric feature of the interaction region, in particular a length of the interaction region along the feed direction, is detected or determined by image analysis to detect the disturbance during image evaluation. As described above, geometric features of the interaction region can be used to detect disturbances in the machining process, in particular a disturbance in the form of a cut break. The geometric features can be combined with other features of the interaction region, for example with the intensity or the intensity curve in the interaction region or in a process light region of interest (see below), in order to detect a respective disturbance.

[0020] In a further development of this variant, a cut break is detected during cutting processing if a parameter dependent on the length of the interaction area exceeds a threshold value and if the absence of the local intensity drop within the interaction area is detected.

[0021] The parameter dependent on the length of the interaction region is a function that depends on the length of the interaction region. In the simplest case, the parameter is the length of the interaction region itself. It is also possible for the parameter to be a value proportional to the length of the interaction region, i.e., the length of the interaction region is multiplied by a weighting factor. However, it is also possible for a more complex relationship between the parameter and the length of the interaction region. In addition to the length of the interaction region, other parameters of the recorded image can be included in the parameter, such as the (average) intensity of the recorded image or a sub-area of ​​the recorded image, the width of the interaction region, etc.

[0022] The threshold can be an absolute value, but it is also possible that the threshold is a percentage change from a current / defined operating point of the parameter. In this case, the parameter determined from the image is related to a currently specified parameter. For this purpose, for example, a quotient can be calculated from the parameter determined from the image and the parameter at the operating point. In this case, the quotient is compared with the threshold.

[0023] In this case, two criteria must be met to detect a cut break: Firstly, the parameter must exceed a specified threshold (cut break threshold), and secondly, the criterion that no local decrease in intensity occurs in the intensity curve within the interaction area must be met. If both criteria are met over a specified (short) distance, e.g., on the order of approximately 10 mm, in the images recorded of the interaction area, a cut break has occurred. In this case, intervention can be made in the machining process, for example, a feed stop can be triggered, or an information / warning / error message can be generated or output.

[0024] To calculate the parameter, the exact length of the interaction region, i.e. the process zone of the process light, is determined. The size of the process light region of interest is then determined using the data from the process zone or the area to be monitored. The size of the process light region of interest typically extends - when the image is taken through a processing nozzle - in the feed direction up to the nozzle edge and extends transversely to the feed direction across the entire width of the interaction region. The length of the interaction region is determined, for example, by comparing the intensity with an intensity threshold value or by evaluating the intensity gradient. The intensity threshold value for the length measurement can, for example, be calculated in the form of a quotient of the average intensity of the process light region of interest (see above) and a weighting factor.

[0025] The visible characteristics of the cutting front or the interaction zone (process emission) for the image capture device, e.g., in the form of a camera, are generally detectable up to the underside of the workpiece. A change (extension) of the interaction zone is caused by disturbances in the cutting process, such as feed rate, focus position, gas pressure, contamination of the optics, support bars, a workpiece in the form of a double sheet, etc. In real cutting operations, support bars, and less frequently double sheets, are predominantly responsible for a significant change in the cutting front or the interaction zone, particularly its length in the feed direction.

[0026] As described above, when these disturbances occur, the characteristic changes by a magnitude that corresponds to the change in a real cut break. It is therefore not possible to distinguish between a real cut break and a pseudo-cut break based solely on the characteristic. Only through plausibility checks, such as checking whether a local decrease in intensity occurs in the intensity curve, and through a temporally or distance-dependent analysis, can a cut break be reliably detected. A cut break is typically only detected when both criteria are met over a specified distance.

[0027] In another variant, the repeated detection of a local intensity drop at a processing position is attributed to the presence of a support ridge at the processing position. As described above, dirty support ridges, in particular, can represent a position-dependent disturbance that negatively impacts the processing process, especially the cutting process. If the processing positions of the support ridges or other position-dependent disturbances are known, they can be taken into account accordingly for the processing process, e.g., for the parts or contours to be cut.

[0028] To locate the machining positions where position-dependent disturbances occur, e.g., due to support bars, the occurrence of a disturbance in the form of a local decrease in intensity in the intensity curve of the interaction area is assigned to a current machining position in machine coordinates (X / Y / Z). Using these coordinates, position-dependent disturbances such as support bars or hotspots at individual positions of the support bars in the machining area can be precisely determined and, if necessary, classified, as described below.

[0029] In a further development, a degree of position-dependent disturbance, in particular a (local) degree of contamination of the support web, is determined based on the intensity profile, in particular based on a gradient of the intensity profile. The severity, i.e. the size, of the position-dependent disturbance can be determined based on the gradient of the intensity profile in the area of ​​the local intensity drop. For example, a degree of contamination of a support web can be determined based on the absolute value of a local minimum or a local maximum of the gradient of the intensity profile in the area of ​​the local intensity drop. In principle, a smaller local minimum or local maximum of the gradient can be assigned to a greater degree of contamination of the support web, and vice versa.Alternatively or additionally, the degree of position-dependent interference can also be determined based on the magnitude of the local intensity drop in the intensity curve of the recorded image. In general, a larger intensity drop can be attributed to a lower (local) degree of contamination of the support bar at the processing position, and vice versa.

[0030] The degree of contamination of the support bar influences the degree of disruption to the machining process: A "new", only slightly slagged or contaminated support bar at a certain machining position has only a minor influence on the cutting process, whereas an "old", heavily contaminated support bar at a certain machining position can have a critical influence on the machining process.

[0031] The magnitude or degree of the position-dependent disturbance can be used to adapt the machining process, for example, a cutting process, or more precisely, the cutting program of the cutting process, to take the position-dependent disturbance variable or the degree of disturbance into account accordingly, or to make changes to the sequence or control of the machining process. For example, depending on the degree of the position-dependent disturbance(s), the workpiece parts to be cut can be (re)nested on the workpiece, job planning can be adjusted, heavily contaminated support bars can be replaced if necessary, etc. It goes without saying that information about the degree of disturbance or the degree of contamination can also be formulated or displayed accordingly for the user.

[0032] The type of disturbance can also be classified based on further characteristics of the interaction zone, for example its width and / or length. The severity, i.e. the size or degree of the disturbance, can also be determined. For this purpose, the local severity and shape of the disturbance during the cutting process and / or its history, e.g. by repeatedly overlapping or passing over one and the same processing position, can be recorded. In addition to or alternatively to position-dependent disturbances, by evaluating the image of the interaction zone or a temporal sequence of images, disturbances dependent on the processing direction can also be detected and their type or severity can be determined, as is described, for example, in DE 10 2018 217 526 A1 cited at the beginning, which is incorporated into this application in its entirety by reference.

[0033] In another variant, the nature of the disruption to the machining process is determined based on the evaluation of a plurality of chronologically successive images of the interaction area. As described above, the trajectory during machining and thus the current machining position relative to the workspace are known. This allows the recorded images to be assigned to the machining positions or the workspace.

[0034] The support webs are typically arranged in the workspace at predetermined coordinates in a first direction (e.g., X-direction) and extend in a second direction (Y-direction). A recurring or—in the case of extremely slagged webs—possibly constant disturbance in the Y-direction is thus present at specific positions in the X-direction. In the X-direction, on the other hand, the disturbances are generally locally limited, namely to the extent of the support web in the X-direction. By evaluating a plurality of consecutive images, it is therefore possible to conclude that a position-dependent disturbance in the machining process in the form of a support web is occurring.

[0035] Location- and direction-dependent disturbances in the workspace that are characteristic of a double sheet can also be determined using a plurality of images of the interaction area, which are assigned to a respective machining position and machining direction during movement along the trajectory. In this way, for example, a disturbance in the shape of a double sheet can be deduced.

[0036] A further aspect of the invention relates to a processing machine of the type mentioned above, in which the evaluation device is configured to detect the presence or absence of a local intensity drop in an intensity profile within the interaction region along a feed direction of the processing process in order to recognize the disturbance during the evaluation of the image, and is further configured to associate the detection of the local intensity drop at a processing position with a position-dependent disturbance of the processing process at the processing position. If a support web or a double sheet is currently being traversed during the processing process, a local intensity drop occurs in the intensity profile, which is otherwise essentially constant within the interaction region.

[0037] A local intensity drop can be detected, for example, if the intensity curve within the interaction region exhibits a local minimum whose intensity value is below a specified percentage, e.g., below 80%, of a maximum intensity value within the interaction region. The intensity value of the intensity curve increases in the feed direction before and after the local intensity minimum until the intensity curve drops steeply at the front and rear ends of the interaction region, respectively.

[0038] Alternatively or additionally, the presence or absence of a local intensity decrease can also be detected based on the gradient of the intensity curve. As described above, for this purpose, for example, the absolute value of the gradient at the location of a local minimum or a local maximum within the interaction region, or the gradient increase from the local minimum to the local maximum, can be used and compared with a threshold value.

[0039] In one embodiment, the evaluation device is configured to detect a cut break as a disturbance during the cutting process of the workpiece only if the absence of a local intensity drop is detected within the interaction zone. If no local intensity drop is present, typically no support web is passed over or there is no double sheet, i.e., the cutting process is not influenced by these disturbances. In this case, the presence of a cut break can be determined solely based on a geometric criterion or on geometric features of the interaction zone.

[0040] In a further embodiment, the evaluation device is configured to detect at least one geometric feature of the interaction region, in particular a length of the interaction region along the feed direction, to detect the disturbance when evaluating the image. As described above, the type of disturbance can be detected, i.e., the disturbance can be classified, based on geometric features of the interaction region, among other things. For example, based on the length of the interaction region in the feed direction, it can be detected that there is no cut break if a predefined cut break criterion is not met.

[0041] In a further development, the evaluation device is configured to detect a cut break during cutting if a parameter dependent on the length of the interaction zone exceeds a threshold value and if the absence of a local intensity drop within the interaction zone is detected. For the reliable detection of a cut break, both criteria must be met: on the one hand, the parameter must exceed a specified threshold value, and on the other hand, no local intensity drop must be detected within the interaction zone.

[0042] In a further embodiment, the evaluation device is configured to associate the detection of a local drop in intensity at a processing position, in particular multiple detections, with the presence of a support web at the processing position. If a local drop in intensity occurs at least twice or more than twice at one and the same processing position - when processing different workpieces - it is very likely that a support web is present at this processing position, i.e., a defect in the form of a double sheet can be practically ruled out. In this way, a defect in the form of a heavily contaminated support web can be reliably distinguished from a defect in the form of a double sheet.

[0043] In a further embodiment, the evaluation device is configured to determine a degree of position-dependent disturbance, in particular a degree of contamination of the support web, based on the intensity profile, in particular based on a gradient of the intensity profile. As described above, a comparatively small absolute value of the local minimum or local maximum of the gradient in the area of ​​the position-dependent disturbance indicates a heavily contaminated support web or a double sheet. If the absolute value of the local minimum or local maximum of the gradient is comparatively large, this indicates the presence of a virtually uncontaminated support web that only slightly influences the cutting process. It is understood that the gradient of the intensity profile can also be evaluated in other ways to determine the degree of position-dependent disturbance.The intensity curve itself can also be evaluated for this purpose, for example by determining the absolute value of the local intensity drop, i.e. the difference between the maximum intensity value in the interaction region and the intensity value of the local intensity minimum.

[0044] In a further embodiment, the evaluation device is configured to infer the type of disruption in the machining process based on the evaluation of a plurality of temporally successive images of the interaction area. As described above, types of disruption can be detected or differentiated, for example, from support webs or the presence of a double sheet.

[0045] Further advantages of the invention will become apparent from the description and the drawings. Likewise, the features mentioned above and those listed below can be used individually or in combination in any desired manner. The embodiments shown and described are not intended to be exhaustive, but rather are exemplary in nature for describing the invention.

[0046] They show: Fig. 1 a schematic representation of a laser processing machine for carrying out a laser cutting process, Fig. 2 a schematic representation of a device for monitoring the laser cutting process by recording an image of a region of the workpiece to be monitored, which contains an interaction region, Fig. 3a-c schematic representations of images of the interaction region and an intensity profile along the interaction region during an undisturbed cutting process, during a cut break and when passing over a support web, and Fig. 4a,b schematic representations analogous to Fig. 3a-c when driving over a heavily soiled or almost new support bridge.

[0047] In the following description of the drawings, identical reference symbols are used for identical or functionally identical components.

[0048] Fig. 1 shows a processing machine in the form of a laser processing machine 1 with a laser source 2, a laser processing head 4 and a workpiece support 5. A laser beam 6 generated by the laser source 2 is guided to the laser processing head 4 by means of a beam guide 3 with the aid of deflecting mirrors (not shown), where it is focused and aligned perpendicular to the surface 8a of a workpiece 8 with the aid of mirrors (likewise not shown), i.e. the beam axis (optical axis) of the laser beam 6 runs perpendicular to the workpiece 8. In the example shown, the laser source 2 is a CO 2 laser source. Alternatively, the laser beam 6 can be generated, for example, by a solid-state laser.

[0049] To laser cut the workpiece 8, the laser beam 6 is first pierced, i.e., the workpiece 8 is melted or oxidized at a specific point, and the resulting melt is blown out. The laser beam 6 is then moved across the workpiece 8, creating a continuous cutting kerf 9 along which the laser beam 6 cuts through the workpiece 8.

[0050] Both piercing and laser cutting can be assisted by adding a gas. Oxygen, nitrogen, compressed air, and / or application-specific gases can be used as cutting gases 10. Developing particles and gases can be extracted using an extraction device 11 from a suction chamber (not shown), which is located below the workpiece support 5.

[0051] The laser processing machine 1 also includes a movement device 12 for moving the laser processing head 4 and the workpiece 8 relative to one another. In the example shown, the workpiece 8 rests on the workpiece support 5 during processing, and the laser processing head 4 is moved along two axes X, Y of an XYZ coordinate system. For this purpose, the movement device 12 has a gantry 13 that can be displaced in the X direction by means of a drive indicated by a double arrow. The laser processing head 4 can be displaced in the X direction by means of a further drive of the movement device 12, indicated by a double arrow, in order to be moved to any processing positions BX,Y in the X direction and in the Y direction within a working field predetermined by the displaceability of the laser processing head 4 or by the workpiece 8.At a respective processing position BX,Y, the laser beam 6 has an (instantaneous) feed direction V which corresponds to the (instantaneous) relative speed between the laser processing head 4 and the workpiece 8.

[0052] Fig. 2 shows an exemplary structure of a device 14 for process monitoring and control of a laser cutting process on the workpiece 8 by means of the laser processing machine 1 of Fig. 1 , of which the laser processing head 4 is shown only very schematically with a focusing lens 15 made of zinc selenide for focusing the laser beam 6 of the laser processing machine 1, a cutting gas nozzle 16, and a deflecting mirror 17. In the present case, the deflecting mirror 17 is partially transparent and forms an entry-side component for the device 14 for process monitoring.

[0053] The deflection mirror 17 reflects the incident laser beam 6 (with a wavelength of approximately 10 µm) and transmits radiation 19 relevant for process monitoring, reflected by the workpiece 2 and emitted by an interaction region 18 of the laser beam 5 with the workpiece 2, in a wavelength range that, in the present example, lies between approximately 550 nm and 2000 nm. As an alternative to the partially transmissive deflection mirror 17, a scraper mirror or a pinhole mirror can also be used to feed the process radiation 19 to the device 14.

[0054] In the device 14, behind the semi-transparent mirror 17, another deflecting mirror 20 is arranged, which deflects the process radiation 19 onto a geometrically high-resolution camera 21 as the image capture unit. The camera 21 can be a high-speed camera arranged coaxially to the laser beam axis 22 or as an extension of the laser beam axis 22a and thus independent of direction. In principle, the camera 21 can also record the image using the incident light method in the VIS wavelength range, possibly also in the NIR wavelength range, provided an additional illumination source radiating in this wavelength range is provided, and alternatively, the recording of the process's intrinsic luminescence in the UV and NIR / IR wavelength ranges.

[0055] For the illustration in the present example, a Fig. 2 An imaging, focusing optical system 23, shown as a lens, is provided, which focuses the radiation 19 relevant for process monitoring onto the camera 21. In the Fig. 2 In the example shown, a filter 24 in front of the camera 21 is advantageous if further radiation or wavelength components are to be excluded from detection by the camera 21. The filter 24 can be designed, for example, as a narrow-band bandpass filter.

[0056] In the present example, the camera 21 is operated using the incident light method, ie an additional illumination source 25 is provided above the workpiece 8, which couples illumination radiation 27 coaxially to the laser beam axis 22 into the beam path via a further partially transparent mirror 26. Laser diodes or diode lasers can be provided as additional illumination source 25, which, as in Fig. 2 shown coaxially, but also off-axis to the laser beam axis 22. The additional illumination source 25 can, for example, also be arranged outside (in particular next to) the laser processing head 4 and directed towards the workpiece 8; alternatively, the illumination source 25 can be arranged inside the laser processing head 4, but not directed coaxially to the laser beam 6 onto the workpiece 8. It is understood that the device 14 can also be operated without an additional illumination source 25.

[0057] In the Fig. 2 In the example shown, during a laser cutting process, the camera 21 records an image B of an area 28 of the workpiece 8 to be monitored, which area contains the interaction area 18. During the cutting process, a relative movement occurs between the workpiece 8 and the laser processing head 4 due to the movement of the laser processing head 4 along the positive Y direction (see arrow) at the relative speed referred to as the feed speed V. During the cutting process, a cutting front 29 forms in the lead to the interaction area 18, to which the cutting kerf 9 adjoins in the trailing direction (in the negative Y direction).

[0058] The image capture device 21 in the form of a camera is connected to an evaluation device 30 via signaling. The evaluation device 30 is configured or programmed to detect at least one disturbance in the machining process, for example, a cut break, based on the evaluation of the recorded image B or a temporal sequence of images B of the area 28 to be monitored.

[0059] For this purpose, the evaluation device 30 evaluates the image B or a series of consecutively recorded images B of the interaction area 18 in order to extract or identify features of the interaction area 18 that indicate a disturbance in the cutting process.

[0060] Fig. 3a shows an example of image B of an interaction area 18 in a good cut, ie in an undisturbed cutting process. As in Fig. 3a As can be seen, the interaction region 18 has a comparatively short length L along the feed direction V, which in the example shown corresponds approximately to twice the width of the interaction region 18 transverse to the feed direction V. The length L of the interaction region 18 in the feed direction V is determined by the evaluation device 30 by comparing the location-dependent intensity I within the image B with an intensity threshold value. The intensity threshold value for the length measurement can, for example, be calculated in the form of a quotient of an average intensity of a process light region of interest (ROI) B ROI and a weighting factor. In the example shown, the process light region of interest B ROI is a rectangular sub-region of the image B of the region 28 of the workpiece 8 to be monitored.The shape of the process light region of interest B ROI extends - when the image B is taken through a processing nozzle 16 - in the feed direction V from the cutting front 29 to the nozzle edge 16a of a nozzle opening 16b of the processing nozzle 16. Transversely to the feed direction V, the process light region of interest extends over the entire width of the interaction area 18.

[0061] From the length L of the interaction area 18 in the feed direction V, a characteristic can be formed, which is compared with a threshold value S that defines a cut break threshold. For the sake of simplicity, it is assumed below that the length L of the interaction area 18 itself forms the characteristic. If the characteristic, in the example shown the length L of the interaction area 18, is smaller than the threshold value S, as shown in Fig. 3a If this is the case, no cut break is detected by the evaluation device 30. The threshold value S, at which no cut break is usually present, can be determined experimentally before the cutting process. It is also possible for the threshold value S to be not an absolute value but a percentage change in a current / defined operating point of the parameter. In this case, the parameter determined from the image is related to a currently specified parameter. For this purpose, for example, a quotient can be formed from the parameter determined from the image (here: L) and the parameter at the operating point. The quotient is then compared with the threshold value.

[0062] For the evaluation device 30 to detect a cut break, it is not sufficient if the parameter or the length L of the interaction area 18 is greater than the threshold value S, since in this case a cut break may be present, but the exceeding of the threshold value S may also be caused by other disturbances in the cutting process, as will be explained below with reference to Fig. 3b und Fig. 3c is explained.

[0063] Fig. 3b shows the image of the interaction area 18 in the presence of a cut break as well as an intensity curve I(Y) along the feed direction V, which in the example shown corresponds to the Y direction, as well as the gradient dI / dY of the intensity curve I(Y). As in Fig.3b As can be seen, the interaction region 18 has a significantly greater length L than in the Fig. 3a shown good cut is the case, ie the threshold value S of the length L of the interaction area 18 is exceeded. The threshold value S of the length L of the interaction area 18 is also exceeded if a support web 7 is passed over during the cutting process, as shown in Fig. 3c is shown using a corresponding image B of the interaction area 18. Based on the parameter in the form of the length L of the interaction area 18 alone, it is therefore not possible to distinguish between a real cut break, as shown in Fig. 3b shown, and a pseudo-section as shown in Fig. 3c shown, can be distinguished.

[0064] However, such a distinction can be made based on the intensity profile I(Y) within the interaction region 18 along the feed direction V or the gradient dI / dY of the intensity profile I(Y) along the feed direction Y, as shown in Fig. 3b and in Fig. 3c shown below. In the Fig. 3b In the intensity curve I(Y) shown, no local intensity drop ΔI or discontinuity occurs within the interaction region 18, ie no local intensity minimum I MIN occurs, as in Fig. 3c is the case. As in Fig. 3c As can be seen, the local intensity minimum I MIN or the intensity drop ΔI occurs at a point along the feed direction V, at which Fig. 3a shown interaction region 18 in an undisturbed cutting process, approximately the end of the interaction region 18 would be expected. This fact makes it easier to detect the occurrence of the intensity drop ΔI or the discontinuity in the intensity curve I(Y) in real time using suitable image evaluation algorithms.

[0065] The presence of the local intensity drop ΔI is detected in the example shown by the evaluation device 30 when both a local minimum (dI / dY) MIN of the gradient dl / dY and a local maximum (dI / dY) MAX of the gradient dl / dY of the intensity curve I(Y) have a Fig. 3b,c exceed the (absolute) threshold value (dl / dY)s shown in dashed lines. The threshold value (dl / dY)s can be determined experimentally or set depending on the current operating point. Detection of the presence or absence of the local intensity drop ΔI by evaluating the gradient dl / dY has proven advantageous.

[0066] The occurrence of the intensity drop ΔI can alternatively be detected by the evaluation device 30 if the value of the intensity minimum I MIN falls below a predetermined percentage, for example less than 80%, of a maximum intensity I MAX of the intensity curve I(Y) in the interaction region 30, i.e. if the local intensity drop ΔI is at least 20% of the maximum intensity I MAX. It is also possible to combine the criterion for the presence of the intensity drop ΔI based on the gradient dl / dY described above with the criterion described here. For example, the criterion described here can be used to check the plausibility of the criterion described above.

[0067] If the evaluation device 30, which can be a computer or suitable hardware and / or software, e.g. in the form of an ASIC, FPGA, etc., detects the absence of the local intensity drop ΔI within the interaction region 18 over a predetermined (short) distance, e.g. in the order of magnitude of approximately 10 mm, and if the threshold value S of the parameter or the length L of the interaction region 18 is exceeded, the evaluation device 30 detects a cut break.

[0068] As in Fig. 2 As can be seen, the evaluation device 30 is in signal communication with a control or regulating device 31, which controls or regulates the laser cutting process. If a cut break is detected by the evaluation device 30, the control / regulating device 31 can suitably adapt the cutting parameters of the laser cutting process in order to counteract a continuation of the cut break during the further execution of the laser cutting process. Alternatively, however, it is also possible for the control or regulating device 31 to abort the cutting process upon detection of the cut break or, if necessary, to restart the cutting process in order to machine the affected area again and to completely cut through the workpiece 8 in the area affected by the cut break, or for the control / regulating device 31 to output information about the cut break to an operator.

[0069] Based on features of the interaction area 18, in addition to process-related disturbances such as a cut break, position-dependent disturbances 37 and / or angle-dependent disturbances 38 of the cutting process can be detected (cf. Fig. 1 ). For example, upon (particularly repeated) detection of a local intensity drop ΔI at one (or at one and the same) machining position BX,Y of the workpiece support 5, a position-dependent disturbance 37 of the machining process in the form of a support ridge 7 or, if applicable, local contamination of the support ridge 7 (hotspot) can be assigned to this machining position BX,Y. The respective machining position BX,Y can be assigned to a respective recorded image B based on the associated machine coordinates X, Y, Z of the movement device 12 when the image B was recorded.

[0070] As shown by Fig. 4a ,bcan be seen, the degree or strength of the position-dependent disturbance can be determined based on the amount or size of the local intensity drop ΔI along the intensity curve I(Y) and / or based on the gradient dl / dY of the intensity curve I(Y): In the case of Fig. 4a In the example shown, the local intensity drop ΔI is approximately 65% ​​of the maximum intensity value I MAX of the intensity curve I(Y), while in the Fig. 4b In the example shown, the local intensity drop ΔI is more than approximately 95% of the maximum intensity value I MAX. Fig. 4a In the example shown, the support bar 7 is heavily soiled (possibly only locally), while in the example shown Fig. 4b In the example shown, the support web 7 is practically uncontaminated. Based on the magnitude of the intensity drop ΔI, the evaluation device 30 can thus determine the degree of contamination of the support web 7.

[0071] To determine the strength of the position-dependent disturbance, the Fig. 4a,b The gradient dl / dY of the intensity curve I(Y) (not shown in the image) can be evaluated, as described in connection with Fig. 3b,c In this case, the degree or strength of the disturbance can be determined, for example, using the absolute value of a local minimum (dI / dY) MIN of the gradient dI / dY or the absolute value of a local maximum (dI / dY) MAX of the gradient dI / dY of the intensity curve I(Y).

[0072] The Fig. 4a The intensity curve I(Y) shown can, in principle, also be caused by another type of disturbance, for example, by a double sheet. However, if the intensity drop ΔI is detected several times at the same processing position BX,Y, it can be assumed that a support web 7 is located at this processing position BX,Y.

[0073] Based on the evaluation of a plurality of temporally successive images B of the interaction area 18, as shown for example in Fig. 4a,b are shown, it is therefore possible to draw conclusions about the type of disturbance in the machining process, for example the presence of a support web or a double sheet.

[0074] The method described above for detecting malfunctions in a machining process is not limited to a cutting process, but can also be used in other machining processes, such as welding processes. A different type of machining beam, such as a plasma beam, can also be used instead of a laser beam 6. In this case, a plasma head is used as the machining tool 4 instead of a laser machining head.

Claims

1. A method for recognizing at least one fault in a machining process, comprising: machining a workpiece (8) by moving a machining tool and the workpiece (8) relative to each other, capturing an image (B) of an area (28) to be monitored on the workpiece (8), which comprises an interaction area (18) of the machining tool with the workpiece (8), and evaluating the image (B) of the area (28) to be monitored in order to detect at least one fault in the machining process, wherein, for the purpose of recognizing the fault when evaluating the image (B), the presence or absence of a local intensity drop (ΔI) in an intensity profile (I) within the interaction area (18) along a feed direction (V) of the machining process is detected, characterized in that the detection of the local intensity drop (ΔI) at a machining position (BX,Y) is assigned to a position-dependent fault in the machining process at the machining position (BX,Y).

2. The method according to claim 1, in which a cut-off as a fault during the cutting of the workpiece (8) is only recognized in the event that the absence of the local intensity drop (ΔI) within the interaction area (18) is detected.

3. The method according to claim 1 or 2, in which at least one geometric feature of the interaction area (18) is detected in order to recognize the fault when evaluating the image (B).

4. The method according to claim 3, in which a cut-off during cutting is recognized when a characteristic dependent on the length (L) of the interaction area (18) along the feed direction (V) exceeds a threshold value (S), and when the absence of the local intensity drop (ΔI) within the interaction area (18) is detected.

5. The method according to any of the preceding claims, wherein detecting the local intensity drop (ΔI) at a machining position (BX,Y) is associated with a presence of a support web (7) at the machining position (BX,Y).

6. The method according to one of the preceding claims, in which a degree of position-dependent fault is determined on the basis of the intensity profile (I(Y)).

7. The method according to claim 6, in which a degree of position-dependent fault is determined on the basis of a gradient of the intensity profile (I(Y)).

8. The method according to one of the preceding claims, in which the type of fault in the machining process is inferred on the basis of the evaluation of a plurality of temporally successive images (B) of the interaction area (18).

9. A processing machine (1), comprising: a machining tool for machining a workpiece (8), a movement device (12) for moving the machining tool and the workpiece (8) relative to each other, an image detection device (21) for capturing an image (B) of an area (28) to be monitored on the workpiece (8), comprising an interaction area (18) of the machining tool with the workpiece (8), and an evaluation device (30) configured to recognize at least one fault in the machining process on the basis of the evaluation of the image (B) of the area (28) to be monitored, wherein the evaluation device (30) is configured to detect the presence or absence of a local intensity drop (ΔI) in an intensity profile (I) within the interaction area (18) along a feed direction (V) of the machining process in order to detect the fault when evaluating the image (B), characterized in that the evaluation device (30) is configured to assign the detection of the local intensity drop (ΔI) at a machining position (BX,Y) to a position-dependent fault in the machining process at the machining position (BX,Y).

10. The processing machine according to claim 9, in which the evaluation device (30) is configured to recognize a cut-off as a fault during the cutting processing of the workpiece (8) only in the event that the absence of the local intensity drop (ΔI) within the interaction area (18) is detected.

11. The processing machine according to claim 9 or 10, in which the evaluation device (30) is configured to detect at least one geometric feature of the interaction area (18) in order to recognize the fault when evaluating the image (B).

12. The processing machine according to claim 11, in which the evaluation device (30) is configured to recognize a cut-off during cutting when a characteristic dependent on the length (L) of the interaction area (18) exceeds a threshold value (S), and when the absence of the local intensity drop (ΔI) within the interaction area (18) is detected.

13. The processing machine according to one of claims 9 to 12, in which the evaluation device (30) is configured to associate the detection of the local intensity drop (ΔI) at a machining position (BX,Y) with the presence of a support web (7) at the machining position (BX,Y).

14. The processing machine according to one of claims 9 to 13, in which the evaluation device (30) is configured to determine a degree of the position-dependent fault on the basis of the intensity profile (I(Y)).

15. The processing machine according to claim 14, in which the evaluation device (30) is configured to determine a degree of the position-dependent fault on the basis of a gradient (dl / dY) of the intensity profile (I(Y)).

16. The processing machine according to one of claims 9 to 15, in which the evaluation device (30) is configured to infer the type of fault in the machining process on the basis of the evaluation of a plurality of temporally successive images (B) of the interaction area (18).