Micro-plasma arc additive metal transition mode recognition method based on continuous multi-image fusion processing

Abstract

The invention relates to the field of micro-plasma arc additive materials, in particular to a micro-plasma arc additive material metal transition mode recognition method based on continuous multi-image fusion processing. In order to solve the problems that in the micro-plasma arc material increasing process, dynamic metal transition modes are varied, visual image processing is greatly influenced by the environment, technological parameters are difficult to regulate and control in a targeted mode, and consequently deposition quality is inconsistent, the micro-plasma arc material increasing metal transition mode recognition method based on continuous multi-image fusion processing is provided. According to the technical scheme, the method is characterized in that images obtained when a magnetic control micro plasma arc scans a left extreme value position, a middle extreme value position and a right extreme value position in a swing period serve as an image group, and difference characteristics of arc length ratios among the three extreme value positions in different metal transition modes are extracted from the continuous multi-image group; and metal transition mode identification is realized according to the difference characteristics. According to the method, the influence of inconsistent shooting exposure and low processing precision of a single image can be reduced, the limitation of identification of a dynamically changing arc process is broken through, and the state of metal transition in the deposition manufacturing process of the micro-plasma arc fuse is rapidly and accurately identified.

Description

TECHNICAL FIELD

[0001] The application particularly relates to a micro-plasma arc additive metal transition mode recognition method based on continuous multi-image fusion processing, and belongs to the field of micro-plasma arc additive manufacturing. BACKGROUND

[0002] As a kind of electric arc heat source with small heat input, micro-plasma arc has broad application prospects in metal additive manufacturing due to its excellent arc stability, concentrated heat source distribution and narrow heat affected zone. At the same time, micro-plasma arc has controllability of external field, which makes the additive manufacturing process more flexible, and the forming quality of micro-plasma arc wire deposition can be effectively improved by applying local alternating magnetic field. In the process of micro-plasma arc additive manufacturing, there is a correlation between metal transition mode and arc state, and both of them affect the quality of the manufactured workpiece.

[0003] At present, there is a lack of characteristic analysis of the arc in the process of micro-plasma arc additive manufacturing under external field control, and the recognition of the dynamic changing arc process has limitations. In view of the above problems, the application discloses a micro-plasma arc additive metal transition mode recognition method based on continuous multi-image fusion processing, and establishes the mapping relationship between the continuous multi-image fusion features of dynamic micro-plasma arc and the metal transition mode. SUMMARY

[0004] The purpose of the application is to provide a micro-plasma arc additive metal transition mode recognition method based on continuous multi-image fusion processing, and to design a metal transition mode recognition system with simple structure and clear principle. The metal transition mode recognition is more accurate when the arc swings under the action of magnetic field, and is used for real-time monitoring of the arc state in the process of magnetic control swing micro-plasma arc additive manufacturing, so as to ensure that the micro-plasma arc is in continuous deposition mode or continuous and intermittent deposition mode, and to carry out continuous and stable deposition to obtain relatively good deposition quality.

[0005] The purpose of the application is achieved by the following technical scheme: Figure 1As shown, the vision-monitored magnetically controlled oscillating micro-plasma arc additive metal transition pattern recognition system consists of a magnetically controlled arc system, a metal transition imaging system, an arc image processing system, and an image feature analysis system. Under the control of the magnetically controlled arc system, the micro-plasma arc oscillates left and right to reach three extreme positions: the leftmost, middle, and rightmost. In the metal transition imaging system, a high-speed camera continuously acquires images of the arc's extreme positions within one oscillation cycle and transmits the acquired high-precision images to the arc image processing system. The arc image processing system denoises the received images, extracts the precise arc region, and transmits the processed images to the image feature analysis system. The image feature analysis system uses a multi-image feature fusion algorithm to extract the ratio R1 (the arc length at the left extreme position to the arc length at the middle position) and R2 (the arc length at the middle position to the arc length at the right extreme position) as high-quality arc features within the image group. Based on the R1 and R2 values ​​of multiple image groups and the arc length, the micro-plasma arc additive metal transition pattern is comprehensively identified.

[0006] The magnetically controlled arc system includes a magnetic field generator, a frequency regulator, a welding torch, and a wire feed tube. The magnetic field direction is perpendicular to the welding torch's travel direction. The magnetic field generator produces a locally alternating controllable magnetic field based on the magnetic effect of current. It is fixed to the welding torch by a clamping device and moves with the torch, thereby controlling the arc behavior. The alternating magnetic field pulls the micro-plasma arc to oscillate regularly between the wire, the wire end, and the molten pool, reaching three extreme positions: the leftmost, the middle, and the rightmost. The frequency regulator can adjust the magnetic field frequency, changing the magnetic field alternation period and the arc oscillation period, thereby controlling the arc oscillation speed and the time spent at the three extreme positions.

[0007] The metal transition imaging system consists of a high-speed camera and an infrared auxiliary device. The high-speed camera continuously acquires images of the arc-metal transition process at a high frame rate, ensuring high-precision capture of the arc behavior during additive manufacturing. The infrared auxiliary device provides good illumination for the high-speed camera and enables automatic adjustment of exposure time and aperture. Both the high-speed camera and the infrared auxiliary device move with the welding torch via a clamping device, ensuring high-precision capture of the arc image during torch movement. Thus, the metal transition imaging system continuously captures images of the micro-plasma arc at three extreme positions during the additive manufacturing process. Based on the adapted shooting environment, the high-frame-rate images of the arc extreme positions are transmitted to the arc image processing system via a communication line.

[0008] The arc image processing system first sorts the acquired continuous arc extreme value position images according to the left, middle, and right extreme values ​​reached by the micro-plasma arc within one oscillation cycle. To accurately extract the arc contour, three consecutive arc extreme value position images within one oscillation cycle are grouped into one image set. Significant noise around the arc and reflected light from the welding wire and molten pool are processed. The arc image processing system uses a color space conversion method to convert the original arc image from the BGR color space to the HSV color space, thereby reflecting the hue characteristics of the arc and distinguishing the arc from other background colors. Then, a threshold segmentation method is used to process the color space converted image. By setting a grayscale threshold, the arc region is separated from the background. Essentially, this process converts the input image into a binary output image based on the segmentation threshold.

[0009]

[0010] In the formula, T is the segmentation threshold, g(x,y) is the output image, and f(x,y) is the input image.

[0011] After thresholding, the "salt-and-pepper noise" in the image is processed by median filtering algorithm. The gray values ​​of non-noise points are replaced with the gray values ​​of noise points, effectively removing salt-and-pepper noise and capturing arc features more accurately. Then, by calculating the number of pixels in the arc region and the interference region, the interference light in the background is effectively eliminated, capturing arc image features more accurately.

[0012] The image feature analysis system extracts the changes in arc length L at the left extreme position (A), middle position (B), and right extreme position (C). To avoid the problem of consistent arc lengths under different metal transition modes and improve pattern recognition accuracy, the vertical pixel unit Q of each arc image in the image group is used to replace the arc length L. To avoid changes in resolution caused by changes in camera focal length or exposure conditions and to overcome the limitation of judging metal transition modes solely by arc length, the ratio R between the arc lengths at the three extreme positions within each image group is used as a reference to select the arc length L at the left extreme position within each image group. A Arc length L at the middle position B Ratio R1, Arc length at midpoint L B Arc length L at the right extreme position C The ratio R2 is used as a feature parameter. By comparing the differences in arc feature parameters R1 and R2 under different metal transition modes, the metal transition mode can be identified using R1 and R2. The extraction of basic parameters is based on the starting point P (the uppermost point of the arc) and the ending point T (the lowermost point of the arc) of the arc. The calculation formula is as follows:

[0013] Q (PT) =T Y -P Y

[0014]

[0015] In the formula, Q (PT) T is the vertical pixel unit from point P to point T. Y P is the ordinate of the bottom edge of the arc image. Y Let L be the top vertical coordinate of the arc image, L be the arc length, R1 be the ratio of the length of arc A at the left extreme position to the length of arc B at the middle position, R2 be the ratio of the length of arc B at the middle position to the length of arc at the right extreme position, and Q be the arc length. A Q B Q C The pixel unit represents the location of each arc extreme value.

[0016] The main feature of this invention is that the arc trajectory is segmented and focused on the arc images at three extreme positions of the micro-plasma arc oscillation: the left extreme, the middle extreme, and the right extreme. A pixel unit Q is used to replace the arc length L at the three extreme positions. The arc length ratios R1 and R2 at adjacent extreme positions are selected as feature parameters, namely, the arc length ratio R1 between the left extreme position and the middle position, and the arc length R2 between the middle position and the right extreme position. Based on the differences in feature parameters under different metal transition modes, the micro-plasma arc additive metal transition mode recognition is completed.

[0017] The beneficial effects of this invention are that it provides a method for recognizing metal transition patterns in micro-plasma arc additive manufacturing based on the fusion processing of three consecutive images, enriching the feature analysis of the arc during micro-plasma arc additive manufacturing under external field control; the continuous multi-image feature fusion algorithm uses three consecutive images of the arc extreme position within one swing cycle as an image group, avoiding misjudgment of metal transition patterns from a single image and improving the sensitivity to metal transition patterns; by using the arc length ratio R at adjacent extreme positions within the image group as a feature parameter, it solves the problem of recognizing fluctuating metal transition patterns in dynamic swing arc additive manufacturing, breaking through the limitations of traditional static image length feature methods, and significantly improving the accuracy of metal transition pattern recognition. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the device.

[0019] Figure 2 This is a schematic diagram of the three extreme positions of the magneto-controlled micro-plasma arc.

[0020] Figure 3 This is a flowchart of an electric arc image processing system.

[0021] Figure 4 This is a flowchart of an image feature analysis system. Detailed Implementation

[0022] To better illustrate the technical solution and beneficial effects of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments:

[0023] See Figure 1 , Figure 2 This invention employs a multi-image feature fusion algorithm to identify metal transition patterns. The identification device includes a magnetically controlled arc system, a metal transition imaging system, an arc image processing system, and an image feature analysis system. The magnetically controlled arc system includes a magnetic field generator and a frequency regulator. The magnetic field direction is perpendicular to the welding torch's travel direction. The alternating magnetic field generated by the magnetic field generator pulls the micro-plasma arc to oscillate regularly between the wire, the wire end, and the molten pool, reaching three extreme positions: left extreme (A), middle (B), and right extreme (C). Infrared auxiliary equipment ensures good illumination conditions, and a high-speed camera continuously acquires arc images at the three extreme positions to ensure high-precision capture of the arc in different metal transition patterns.

[0024] See Figure 3 The electric arc image processing system arranges the collected images of electric arc extreme positions in the order of left extreme, middle, and right extreme. Three consecutive electric arc images at extreme positions within one swing cycle are grouped into one image set. After converting the original image to the HSV color space, the electric arc region is separated using a threshold segmentation method. The salt-and-pepper noise is suppressed by a median filtering algorithm to achieve image smoothing. By calculating the number of pixels in the electric arc region and the interference region, the influence of interference light on the image is removed. The electric arc image processing system extracts the precise electric arc region from the received image and transmits the processed image to the image feature analysis system.

[0025] See Figure 4 The image feature analysis system extracts the length changes and correlations of the micro-plasma arc at three extreme positions. The length L of the arc is replaced by the pixel unit Q in the vertical direction of the arc. The ratio R1 of the arc length at the left extreme position to the arc length at the middle position and the ratio R2 of the arc length at the middle position to the arc length at the right extreme position within the image group are used as feature parameters. After importing the feature fusion algorithm of multiple consecutive images, recognition is performed on an image group basis. The system uses a single image group to identify whether the metal transition pattern is abnormal. If there is no abnormality, the recognition result is exported. If an abnormality occurs, more image groups are selected for re-identification. If the pattern recognition result is stable, the result is exported. If the result is unstable after selecting multiple image groups for re-identification, the metal transition pattern is determined to be a non-contact pattern.

Claims

1. A method for recognizing metal transition patterns in micro-plasma arc additive manufacturing based on the fusion processing of consecutive multiple images, characterized in that: This method is implemented by an identification device, which consists of a magnetically controlled arc system, a metal transition imaging system, an arc image processing system, and an image feature analysis system. The arc is controlled by the magnetically controlled arc system to swing left and right, reaching three extreme positions: the left extreme, the middle extreme, and the right extreme. During the swinging of the micro-plasma arc, the metal transition imaging system continuously captures arc images during the metal transition and transmits the acquired high-precision images to the arc image processing system. The arc image processing system extracts the precise arc region from the received arc images, and the processed image is transmitted to the image feature analysis system via a communication line. The image feature analysis system uses a feature fusion method from multiple consecutive images to extract high-quality arc features, selecting the arc length L at the left extreme position within the image group. A Arc length L at the middle position B Ratio R1, Arc length at midpoint L B Arc length L at the right extreme position C The ratio R2 is used as a characteristic parameter to identify different metal transition modes by comparing R1, R2 and arc length.

2. The method for recognizing metal transition patterns in micro-plasma arc additive manufacturing based on the fusion processing of multiple consecutive images as described in claim 1, characterized in that: The magnetically controlled arc system includes a magnetic field generator, a frequency regulator, a welding torch, and a wire feed tube. It generates a localized alternating magnetic field covering the metal transition region. This localized alternating magnetic field controls the micro-plasma arc to oscillate left and right between the wire, the wire end, and the molten pool. The time taken for the micro-plasma arc to oscillate at the three extreme positions (leftmost, middle, and rightmost) is calculated based on the magnetic field frequency. The metal transition imaging system includes a high-speed camera and infrared auxiliary equipment, continuously capturing images of the micro-plasma arc at the three extreme positions during operation. The arc image processing system sorts the collected arc extreme position images in the order of left extreme, middle, and right extreme. Three consecutive arc images at extreme positions within one cycle are considered as an image group. The image group is processed using binarization thresholding to separate the arc region. A median filtering algorithm is used to suppress salt-and-pepper noise and smooth the image. The influence of interference light on the image is removed by calculating the number of pixels in the arc region and the interference region. The image feature analysis system obtains the arc features of the image group based on a feature fusion algorithm for multiple consecutive images, thereby identifying the metal transition pattern.

3. The method for recognizing metal transition patterns in micro-plasma arc additive manufacturing based on the fusion processing of consecutive multiple images, as described in claims 1 and 2, is characterized in that: The algorithm for fusing features from multiple consecutive images extracts the length changes and their correlations at three extreme positions (left extreme position (A), middle position (B), and right extreme position (C)) of an image group. The vertical pixel unit Q is used to replace the arc length L. Images captured under different camera parameters will change proportionally. To eliminate the influence of the pixel unit Q on the recognition accuracy due to the change of resolution and exposure, the ratio R between arc lengths is used as a feature parameter to analyze the arc evolution law of different metal transition modes.

4. The micro-plasma arc additive metal transition pattern recognition method based on continuous multi-image fusion processing according to claims 1 and 3, characterized in that: The algorithm for fusing features from multiple consecutive images calculates the arc length L based on the uppermost point P and the lowermost point T of the arc. The arc length L at the three extreme points is calculated using the vertical pixel unit Q from the uppermost point P to the lowermost point T. The arc length L at the left extreme point within each image group is used as the reference value. A Arc length L at the middle position B Ratio R1, Arc length at midpoint L B Arc length L at the right extreme position C The ratio R2 is used as a feature parameter. The arc length ratio features R1 and R2 of different metal transition modes have significant differences. Therefore, the metal transition mode can be identified by using the arc image length ratio features R1 and R2 during the manufacturing process.

Images