Method and device for monitoring and controlling laser machining processes

Abstract

The invention relates to a method and a device for monitoring and controlling laser machining processes on a workpiece (13). The method and device are characterized in that recordings (16) of the process zone of the workpiece (13) are pre-processed (17, 18) using an AI method, the interference factors are eliminated from the recordings (16), geometric workpiece properties are detected, and an input signal (19) for an image processing algorithm (20) is generated therefrom, said algorithm calculating correction values for at least one control parameter (21) of the laser path.

Description

[0001] Method and device for monitoring and controlling laser processing operations

[0002] When laser processing workpieces, the correct alignment of the laser beam relative to the workpiece is crucial for high-quality processing. Therefore, various monitoring and control systems with sensors for object detection and distance measurement between the focusing optics of the laser beam and the workpiece are used on laser processing systems.

[0003] With the help of a line laser positioned upstream of the processing laser beam and a camera that records both the reflections of the line laser beam and the process zone created by the processing laser beam in the workpiece, distance measurements can be performed and height differences in the workpiece can be detected using an image processing system. During laser welding, it is thus possible to detect the transition from a top sheet to a bottom sheet and guide the processing laser along this transition line, creating a precise weld seam. However, this solution is limited to monitoring in the feed direction of the processing laser beam.

[0004] Using an OCT scanner for monitoring, however, allows for the detection of height differences in the workpiece and distance measurements in all spatial directions, independent of the feed direction of the processing laser. However, this monitoring system is relatively complex and expensive due to the use of an OCT scanner.

[0005] EP 2 365 890 B1 discloses a monitoring system for a laser processing system that includes an OCT scanner, a camera with HDR functions, and an image processing device. The camera takes multiple images of the process zone with different exposure times and calculates images with a high contrast ratio from them, which are then fed to the image processing device as measured values. The image processing device compares the measured values ​​with key values ​​determined using a data reduction method.

[0006] Alternatively, the parameters can also be calculated using a neural network.

[0007] The determined parameters allow for a classification of the laser processing process. Furthermore, the determined parameters can be used to readjust the path and / or the distance of the processing laser from the workpiece.

[0008] The invention is based on the object of monitoring and controlling a laser processing process using structurally simple means.

[0009] The task is solved by a method for monitoring and controlling laser processing operations on a workpiece with the following steps:

[0010] - Recording the environment of a process zone in a workpiece created by a processing laser beam using an image recording device;

[0011] - Elimination of disturbing factors in the recording such as process lights, smoke or material splashes using an artificial intelligence method in an image processing device;

[0012] - Detection of geometric workpiece properties such as workpiece edges and / or height jumps in the recording using the artificial intelligence method;

[0013] - generating an input signal for a downstream image processing algorithm from the image cleaned of interference factors and containing information on the geometric workpiece properties by the artificial intelligence method;

[0014] - Calculating a correction value for at least one control parameter for the path of the processing laser beam over the workpiece by the image processing algorithm.

[0015] The method according to the invention allows using only a few

[0016] Devices that are usually already present on the processing heads of laser processing systems allow monitoring and controlling the path of a processing laser beam. No additional line lasers or OCT scanners are required. The process light from the laser processing process can be used as indirect illumination for the images.

[0017] The monitoring and control according to the invention is independent of the feed direction of the processing laser beam, so that, particularly in laser welding processes, seam position control can be carried out even with more complex weld seam geometries.

[0018] The artificial intelligence method also allows for continuous improvement of the preprocessing of the images of the process zone, so that the method according to the invention can make a robust statement about the position of the processing laser beam relative to workpiece edges, weld seams, and other elements of the workpiece. By preprocessing the images of the process zone and its surroundings, an input signal can be provided to the actual image processing algorithm that is free of all interference factors. From this input signal, the image processing algorithm can detect the geometric properties of interest to the workpiece, such as a height jump, and use this to calculate a correction value for at least one control parameter of the path of the processing laser beam. The parameterization of the image processing algorithm can be selected more easily than with prior art methods.

[0019] Further advantages arise when the artificial intelligence method not only eliminates interference factors in the images but also enhances the information about the geometric workpiece properties in the input signal for the image processing algorithm. The image processing algorithm can then identify these workpiece properties with greater reliability and, if necessary, generate appropriate correction values ​​for the laser path control parameters. Another advantage is that the parameters of the image processing algorithm do not have to be changed by the processed input signal, as is the case with state-of-the-art methods. Preprocessing the images makes it possible to keep the parameters of the image processing algorithm constant thanks to the uniqueness of the information in the input signal. The overall parameterization can also be significantly simpler than with conventional methods.

[0020] A further amplification of the input signal can be achieved by using the image recording device to create several images with different exposure times and to combine them using a high dynamic range (HDR) method to create an image with a high contrast value from which the interference factors are eliminated using the artificial intelligence method and the geometric workpiece properties are recorded.

[0021] The images are preferably taken with the image recording device coaxially to the processing laser beam.

[0022] The robustness of the method can be further increased by incorporating information about the workpiece shape and the laser beam path into the preprocessing of the process zone image. In this way, the AI ​​method can narrow down the search space for interference factors and geometric workpiece properties of interest in the images.

[0023] The artificial intelligence method can preferably be used to calculate a single-layer or multi-layer neural network for detecting interference factors and geometric workpiece properties. However, the AI ​​method can also be based on other methods such as linear regression, etc. The method according to the invention can be used in all laser processing processes. However, particular advantages arise when it is a method for monitoring and controlling a laser welding process. In these processes, precise detection of the joining line between two workpieces is essential. If a top sheet is welded to a bottom sheet, the joining line can be detected by a height difference in the workpiece.

[0024] The invention also relates to a device for carrying out a method according to the invention for monitoring and controlling laser processing operations on a workpiece, which device has an image recording device arranged coaxially to a processing laser beam of a laser processing system and an image processing device, characterized in that the image processing device is designed such that it carries out a pre-processing of the image of the image recording device with the aid of a method based on artificial intelligence, with which interference factors in the image are removed and information about geometric properties of the workpiece that are essential for the processing process is enhanced, before it subjects the pre-processed image to an image processing algorithm and generates control parameters for the path of the processing laser beam therefrom.

[0025] The process light from the laser processing can be used as indirect lighting for the images. However, providing additional lighting sources is also possible.

[0026] The image processing device can preferably calculate a neural network for the artificial intelligence method as well as the image processing algorithm to which the images are subjected. This allows adaptation to new workpieces.

[0027] The image capture device can advantageously be a camera with HDR capabilities. The device can also be relatively easily retrofitted to existing laser processing systems.

[0028] The invention also includes a processing head for a laser processing system, which has a device according to the invention.

[0029] In the following, a preferred embodiment of a device according to the invention is explained in more detail with reference to the drawing.

[0030] The sole FIGURE shows a schematic block diagram of a processing head 10 and an image processing device 11 of a laser processing system (not shown further). The processing head 10 directs a processing laser beam 12 onto a workpiece 13. The light 14 reflected by the workpiece 12 is captured by an image recording device 15, preferably an HDR camera, in images 16, which are then forwarded to the image processing device 11.

[0031] In the image processing device 11, a processing 17 of the images 16 from the image recording device 15 takes place, in which interference factors such as reflections from splashes, smoke, or process lights are eliminated from the images 16 using an artificial intelligence method. The processed images 16' are then subjected to further processing 18, in which workpiece edges, height changes, and other geometric workpiece properties are detected in the images using the artificial intelligence method. This generates an input signal 19 for an image processing algorithm 20, which generates control signals 21 for the processing head 10 from these signals 19, which contain the images 16' freed of interference factors and enhanced information about the geometric workpiece properties. These control signals 21 are then subjected to an error check 22 before being forwarded to the processing head 10.The control signals 21 can be used in particular to make corrections to the optical axis of the processing head 10.

[0032] List of reference symbols 10 Machining head

[0033] 11 Image processing device

[0034] 12 processing laser beam

[0035] 13 Workpiece

[0036] 14 reflected light 15 image recording device

[0037] 16, 16' recording

[0038] 17 Processing

[0039] 18 Processing

[0040] 19 Input signal 20 Image processing algorithm

[0041] 21 control parameters

[0042] 22 Error control

Claims

Patent claims 1 . Method for monitoring and controlling laser processing operations on a workpiece, comprising the steps: - Recording the surroundings of a process zone generated by a processing laser beam (12) in a workpiece (13) with an image recording device (15); - Elimination of disturbing factors in the recording (16) such as process lights, smoke or material splashes using an artificial intelligence method in an image processing device (11); - detecting geometric workpiece properties such as workpiece edges and / or height jumps in the holder (16, 16') by the artificial intelligence method; - generating an input signal (19) for a downstream image processing algorithm (20) from the image (16) cleaned of interference factors and containing information about the geometric workpiece properties by the artificial intelligence method; - Calculating a correction value for at least one control parameter (21) for the path of the processing laser beam (12) over the workpiece (13) by the image processing algorithm (20).

2. Method according to claim 1, characterized in that the process lighting of the laser processing operation is used as indirect lighting for the recordings (16).

3. Method according to claim 1 or 2, characterized in that the artificial intelligence method amplifies the information about the geometric workpiece properties in the input signal (19) for the image processing algorithm (20).

4. Method according to one of the preceding claims, characterized in that the parameters of the image processing algorithm (20) are not changed by the processed input signal (19).

5. Method according to one of the preceding claims, characterized in that a plurality of images (16) with different exposure times are generated with the image recording device (15) and are offset against one another using a high dynamic range (HDR) method, and an image (16) with a high contrast value is generated from which the interference factors are eliminated using the artificial intelligence method and the geometric workpiece properties are recorded.

6. Method according to one of the preceding claims, characterized in that the images (16) are recorded with the image recording device (15) coaxially to the processing laser beam (12).

7. Method according to one of the preceding claims, characterized in that the artificial intelligence method takes into account information about the shape of the workpiece (13) and the path of the laser beam (12) in the preprocessing of the recording of the process zone.

8. Method according to one of the preceding claims, characterized in that the artificial intelligence method calculates at least one neural network for detecting the disturbance factors and the geometric workpiece properties.

9. Method according to one of the preceding claims, characterized in that it is a method for monitoring and controlling a laser welding process.

10. Device for carrying out a method for monitoring and controlling laser processing operations on a workpiece (13) according to one of the preceding claims, which device has an image recording device (15) arranged coaxially to a processing laser beam (12) of a laser processing system and an image processing device (11), characterized in that the image processing device (11) is designed such that it carries out a pre-processing (17) of the image (16) of the image recording device (15) with the aid of a method based on artificial intelligence, with which interference factors in the image (16) are removed and information about geometric properties of the workpiece (13) that are essential for the processing process is enhanced, before it subjects the pre-processed image (19) to an image processing algorithm (20) and generates control parameters (21) for the path of the processing laser beam (12) therefrom.

11. Device according to claim 10, characterized in that the image processing device (11) calculates a neural network for the artificial intelligence method and the image processing algorithm (20).

12. Device according to claim 10 or 11, characterized in that the image recording device (15) is a camera with HDR functions.

13. Device according to one of claims 10 to 12, characterized in that it can be retrofitted to existing laser processing systems.

14. Processing head for a laser processing system, characterized in that it comprises a device according to one of claims 10 to 13.

Images