Method for determining a quality characteristic of a welded connection between two conductor ends, method for providing a training data record, training data record, method for welding conductor ends and device for determining a quality characteristic of a welded connection
Patent Information
- Application Number
- EP2023728736
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-05-31
- Filing Date
- 2023-05-24
- Publication Date
- 2025-04-09
AI Technical Summary
Existing methods for evaluating the quality of welded connections between conductor ends in electrical machines are time-consuming and costly, particularly when using X-ray analysis, which is not efficient for in-process monitoring.
A method that records intensity and depth information of image points in a detection area using an image capture device and a depth detection device, combined with a processor unit to determine the quality feature of the welded connection, allowing for in-process quality monitoring with reduced time expenditure, and utilizes a machine learning system, such as an artificial neural network, for accurate determination.
Enables efficient and accurate assessment of welded connection quality, allowing for real-time monitoring and improved control of subsequent welding processes, reducing time and equipment costs.
Smart Images
Figure 1.1
Abstract
Description
[0001] Method for determining a quality characteristic of a welded joint between two conductor ends, method for providing a training data set, training data set, method for welding conductor ends and device for determining a quality characteristic of a welded joint
[0002] The invention relates to a method for determining a quality characteristic of a welded joint between two conductor ends. Furthermore, the invention relates to a method for welding conductor ends. Furthermore, the invention relates to a method for providing a training data set for a machine learning system for determining a quality characteristic of a welded joint between two conductor ends. A further subject of the invention is a training data set for a machine learning system for determining a quality characteristic of a welded joint between two conductor ends. A further subject is a method for welding conductor ends. Finally, the invention relates to a device for determining a quality characteristic of a welded joint.
[0003] Such welded joints are typically formed between two conductor ends of hairpin elements of an electrical machine to join several hairpin elements—hairpin-shaped copper wires with a mostly rectangular cross-section—to form a winding of the electrical machine. An electrical machine typically comprises a large number, often hundreds, of such welded joints.
[0004] The welded connections between the conductor ends are typically made using a beam welding process, e.g. laser welding, and must meet certain quality requirements. For example, the connection area, i.e. the cross-sectional area available for the current flow from one conductor to the other, must be sufficiently large. A connection area that is too small can result in an undesirably high ohmic resistance and negatively impact the efficiency of the electrical machine. It is therefore necessary to check the quality of the welded connections. A common method in the prior art involves determining the connection area using X-ray analysis. In this case, the joined connection area of the two conductors is placed in a suitable device and examined with X-rays, which involves increased equipment expenditure and time expenditure.
[0005] Against this background, the task arises of enabling sensitive testing of welded joints between conductor ends with minimal time expenditure.
[0006] To achieve the object, a method is proposed for determining a quality feature of a welded joint between two conductor ends, in particular between two conductor ends of hairpin elements of an electrical machine, wherein for a plurality of pixels of a detection area comprising the conductor ends, at least intensity information of the pixels and depth information of the pixels are detected, and wherein a quality feature, in particular a connection area or a pore volume, is determined depending on the detected intensity information and depth information.
[0007] A further subject of the invention is a device for determining a quality feature of a welded joint between two conductor ends, in particular between two conductor ends of hairpin elements of an electrical machine, comprising an image capture device for capturing intensity information for a plurality of pixels of a capture area comprising the conductor ends, a depth information capture device for capturing depth information of the pixels, and a processor unit for determining a quality feature, in particular a connection area or a pore volume, depending on the captured intensity information and depth information. In the method and device according to the invention, intensity information and additional depth information are captured for a plurality of pixels of a capture area.The combination of this information makes it possible to derive a quality characteristic of the welded joint between the conductor ends based on this information. The method can be performed during the actual welding process or shortly after completion, thus enabling in-process quality monitoring with minimal time expenditure.
[0008] The method step of determining the quality characteristic based on the acquired intensity information and depth information is preferably carried out by computer implementation, for example, using a processor device. The intensity information can, in particular, be image data from a 2D camera. The depth information is, in particular, information on the topological position of a surface of the weld joint in a direction perpendicular to the plane of the detection area.
[0009] The conductor ends are preferably made of copper. The hairpin elements are particularly preferably made of copper. In this respect, the connection is preferably a copper-to-copper welded joint.
[0010] According to an advantageous embodiment, a first acquisition of the intensity information and the depth information takes place before the formation of the weld joint, and a second acquisition of the intensity information and the depth information takes place after the formation of the weld joint, and the determination of the quality feature takes place as a function of the intensity information and depth information determined before and after the formation of the weld joint. By determining the quality feature as a function of the information acquired before and after welding, the quality feature can be determined with improved accuracy. In this case, it is advantageous if, in addition to the second acquisition of the intensity information and the depth information, the intensities of the process emissions are recorded after the formation of the weld joint.The process emissions can be recorded in particular with detectors based on photodiodes for process-typical wavelengths.
[0011] According to an advantageous embodiment, the depth information is determined using a triangulation device. The triangulation device can enable optical distance measurement and thereby provide depth information for the pixels in the detection area. The triangulation device is preferably designed as a laser triangulation device. The laser triangulation device can comprise one or more line lasers.
[0012] According to an alternative, advantageous embodiment, it is provided that the depth information is determined by means of an optical coherence tomography device or a grazing light device or a stereo camera.
[0013] According to an advantageous embodiment, the intensity information is captured using an intensity channel. In this respect, the intensity information can form a grayscale image of the detection area.
[0014] Alternatively, it is possible for the intensity information to be captured using multiple, in particular three, intensity channels for different colors. The intensity information can form multiple grayscale images for different colors, which, when combined, represent a color image of the detection area. For example, intensity channels can be provided for the colors red, yellow, and blue.
[0015] According to an advantageous embodiment, the quality characteristic is determined using a regression analysis. The regression analysis can be performed using statistical analysis methods.
[0016] According to an advantageous embodiment, the quality feature is determined using a machine learning system. The machine learning system can be embodied, for example, as an artificial neural network. The invention further relates to a method for providing a training data set for a machine learning system for determining a quality feature of a welded joint between two conductor ends, wherein intensity information of the pixels and depth information of the pixels are acquired for several pixels in a detection area encompassing the conductor ends.
[0017] Furthermore, the invention relates to a training data set for a machine learning model for determining a quality characteristic of a welded joint between two conductor ends, comprising intensity information from multiple pixels of a detection area encompassing the conductor ends and depth information of the pixels. Additionally, information from time series formed from the intensities of the process emissions can be provided.
[0018] A further subject of the invention is a method for welding conductor ends, in particular hairpin elements of an electrical machine, using a beam welding process, in particular a laser welding process, wherein - before welding a first pair of conductor ends - intensity information of the pixels and depth information of the pixels are recorded for several pixels of a detection area comprising the conductor ends, wherein the conductor ends of the first pair are welded, wherein - after welding the first pair of conductor ends - intensity information of the pixels and depth information of the pixels are recorded for several pixels of a detection area comprising the conductor ends, wherein depending on the recorded process emission intensity information, the intensity information of the pixels and the depth information, a quality feature, in particular a connection area or a pore volume,is determined, whereby the welding of a second pair of conductor ends following the first pair is controlled depending on the determined quality characteristic. In this case, the intensities of the process emissions can advantageously be recorded during welding.
[0019] The method makes it possible to test welded joints between conductor ends with little time expenditure and to adjust the creation of subsequent welded joints depending on the result of the test.
[0020] In the method for welding conductor ends, the advantageous embodiments and features explained in connection with the method for determining a quality characteristic of a welded joint between two conductor ends can also be used, alone or in combination.
[0021] Further details and advantages of the invention will be described below with reference to an embodiment illustrated in the drawings.
[0022] Fig. 1 shows a schematic sequence of an embodiment of a
[0023] Method according to the invention;
[0024] Fig. 2 exemplary method for determining a quality characteristic using machine learning,
[0025] Fig. 3 shows another exemplary method for determining a
[0026] Quality feature using machine learning, and
[0027] Fig. 4 shows another exemplary method for determining a
[0028] Quality feature using machine learning.
[0029] Fig. 1 schematically shows a sequence of an exemplary embodiment of a method according to the invention for determining a quality feature of a welded joint between two conductor ends 1. The conductor ends 1 shown here are conductor ends 1 of hairpin elements of an electrical machine. The left-hand illustrations show intensity information IR(XY), IG(X,Y), IB(X,Y) of pixels in a detection area that includes the conductor ends 1. The intensity information IR(X,Y), IG(X,Y), IB(X,Y) in the upper illustration was acquired before the welded joint was formed, and the intensity information IR(XY), IG(X,Y), IB(X,Y) in the lower illustration was acquired after the welded joint 2 was formed. The intensity information IR(XY), IG(X,Y), IB(X,Y) can be acquired using an image capture device, for example an optical camera.According to the exemplary embodiment, the intensity information IR(X,Y), IG(X,Y), IB(X,Y) was acquired using several, here three, intensity channels. These can, for example, correspond to the intensity of the colors red, yellow, and blue in the detection area.
[0030] In this detection area, in addition to the intensity information IR(X,Y), IG(X,Y), and IB(X,Y), depth information Z(x,y) of the pixels was also acquired. The depth information Z(x,y) can be acquired using a depth information acquisition device. This is preferably designed as a triangulation device, in particular as a laser triangulation device. Alternatively, a depth information acquisition device designed as an optical coherence tomography device, a grazing light device, or a stereo camera can be used.
[0031] According to the invention, a quality characteristic, in particular a bonding area or pore volume, of the weld joint 2 is determined based on the acquired intensity information IR(X,Y), IG(X,Y), IB(X,Y), and depth information Z(x,y). The quality characteristic is determined using a regression analysis. Preferably, the quality characteristic is determined using a machine learning system 3.
[0032] The illustration in Fig. 2 schematically shows a machine learning system 3 designed as an artificial neural network. To train the artificial neural network, it is fed a training data set 4 comprising intensity information from several pixels of a detection area encompassing the conductor ends and depth information of the pixels. The training data set 4 contains the information necessary for supervised training of a model about the fundamental truth, the so-called "ground truth", for example the connection cross-section and the number of pores and their volume. The training data set 4 contains both the intensity and depth information acquired before the formation of the weld joint and the intensity and depth information acquired after the formation of the weld joint. The training data set 4 according to Fig. 2 comprises intensity information with three channels.
[0033] The trained system 3 is used within the quality characteristic determination process to determine the quality characteristic of the weld joint 2 based on the recorded intensity and depth information. The process can be performed during the actual welding process or shortly after completion of the welding process, thus enabling process-accompanying quality monitoring with minimal time expenditure.
[0034] Figure 3 shows another machine learning system 3, which is implemented as an artificial neural network. Unlike the embodiment shown in Figure 2, the training data set 4 includes intensity information with exactly one intensity channel.
[0035] Fig. 4 shows another machine learning system 3, which is designed as an artificial neural network and is constructed from the network 6, consisting of LSTM and FCN layers, and from the network 7, consisting of CNN and FCN.
[0036] In contrast to the embodiment shown in Fig. 2, the training data set 4 comprises intensity information with, for example, an intensity channel, depth information and time series consisting of the intensities of the process emissions. As an example, three time series are shown, corresponding to the intensities at three selected wavelengths 5. The model to be trained is structured in such a way that it has a branch specifically designed for time series regression, based, for example, on an LSTM-FCN (neural network consisting of long short-term memory layers and fully connected layers), and a branch for image regression, based, for example, on a CNN-FCN (convolutional layers and fully connected layers). Both branches are based on the so-called
[0037] Feature layers merged.
Claims
Claims: Method for determining a quality characteristic of a welded joint (2) between two conductor ends (1), in particular between two conductor ends of hairpin elements of an electrical machine, wherein intensity information of the image points and depth information of the image points are recorded for several pixels of a detection area encompassing the conductor ends (1), and wherein a quality characteristic, in particular a bonding area or a pore volume, is determined depending on the recorded intensity information and depth information.A method according to claim 1, characterized in that a first acquisition of intensity and depth information is performed before the formation of the weld joint (2), and a second acquisition of intensity and depth information is performed after the formation of the weld joint (2), and the determination of the quality characteristic is carried out as a function of the intensity and depth information determined before and after the formation of the weld joint. A method according to claim 2, characterized in that, in addition to the second acquisition of intensity and depth information, the intensities of the process emissions are acquired after the formation of the weld joint (2). A method according to any one of the preceding claims, characterized in that the depth information is determined by means of a triangulation device, in particular a laser triangulation device.
5. Method according to one of claims 1-3, characterized in that the depth information is determined by means of an optical coherence tomography device or a grazing light device or a stereo camera.
6. Method according to one of the preceding claims, characterized in that the acquisition of the intensity information is carried out with one intensity channel or with several, in particular three, intensity channels for different colors.
7. Method according to one of the preceding claims, characterized in that the determination of the quality characteristic is carried out by means of a regression analysis.
8. Method according to one of the preceding claims, characterized in that the determination of the quality characteristic is carried out using a machine learning system.
9. Method for providing a training data set (4) for a machine learning system (3) for determining a quality feature of a weld joint between two conductor ends, wherein intensity information of the image points and depth information of the image points are recorded for several pixels of a detection area encompassing the conductor ends.
10. Training data set for a machine learning system (3) for determining a quality characteristic of a welded joint between two conductor ends, comprising: Intensity information from multiple pixels of a detection area encompassing the ends of the ladder and Depth information of the pixels.
1. Training data set according to claim 10, characterized in that additional information from time series formed from the intensities of the process emissions is provided.
2. Method for welding conductor ends (1), in particular hairpin elements of an electrical machine, using a beam welding process, in particular a laser welding process, wherein – before welding a first pair of conductor ends (1) – intensity information of the pixels and depth information of the pixels are acquired for several pixels of a detection area encompassing the conductor ends (1), wherein the conductor ends (1) of the first pair are welded, wherein – after welding the first pair of conductor ends (1) – intensity information of the pixels and depth information of the pixels are acquired for several pixels of a detection area encompassing the conductor ends.wherein, depending on the acquired intensity and depth information, a quality characteristic, in particular a bonding area or a pore volume, is determined, wherein, depending on the determined quality characteristic, the welding of a second pair of conductor ends (1) following the first pair is controlled.
3. Method for welding conductor ends (1) according to claim 12, characterized in that the intensities of the process emissions are recorded during welding.
14. Device for determining a quality characteristic of a welded joint (2) between two conductor ends (1), in particular between two conductor ends of hairpin elements of an electric machine, with an image acquisition device for acquiring intensity information for multiple pixels of a detection area encompassing the conductor ends, with a depth information acquisition device for acquiring depth information of the pixels, and with a processor unit for determining a quality characteristic, in particular a bonding area or a pore volume, depending on the acquired intensity and depth information.
15. Device for determining a quality characteristic of a weld joint (2) between two conductor ends (1) according to claim 14, characterized in that a process radiation intensity detection unit is additionally provided.