Intelligent welding spot leak detection system and leak detection method for electrical appliance refrigerant pipeline

Abstract

The present invention provides an intelligent weld spot leak detection system and leak detection method for electrical refrigerant pipelines. The leak detection system includes: a robot, a camera module, a leak detection gun, a computer system and a production line; the robot and the camera module are both fixed on one side of the production line; the leak detection gun is connected to the end of the robot; the computer system is electrically connected to the camera module, the robot and the leak detection gun. The leak detection method includes: installing a left camera and a right camera and calibrating them to obtain internal and external parameters of the camera; taking an image of the refrigeration appliance and performing image recognition, and performing image correction according to the internal and external parameters of the camera; obtaining the camera world coordinates of the refrigerant pipeline weld spot through the corrected image measurement, calibrating the robot, obtaining the robot world coordinates, and controlling the robot to carry the leak detection gun to perform weld spot leak detection. The present invention reduces the phenomenon of missed detection and wrong detection caused by worker fatigue, and has good industrial application prospects.

Description

Technical Field

[0001] The present invention belongs to the field of industrial detection and industrial intelligent robot arm, and in particular relates to an intelligent welding spot leak detection system and a leak detection method for electrical refrigerant pipelines. Background Art

[0002] Refrigeration appliances such as refrigerators and air conditioners are widely used in industrial and civilian fields. The principle is to achieve the refrigeration function through the refrigerant in the gas-liquid transmission pipeline inside the appliance. The outer wall of the refrigerant pipeline usually has several manually welded welds. These welds are prone to quality problems due to manual welding, which in turn causes refrigerant leakage, ultimately affecting the performance and even safety of refrigeration appliances. Therefore, the quality inspection of refrigerant pipeline welds is an important process in the production of refrigeration appliances. At present, the quality inspection of welds in the refrigeration appliance industry is mostly manual inspection, that is, workers hold refrigerant leak detection guns to inspect each weld, which has low operating efficiency, and the phenomenon of missed inspections and wrong inspections caused by worker fatigue occurs frequently, resulting in a low pass rate for refrigeration appliances.

[0003] Therefore, there is an urgent need for an intelligent weld spot leak detection system for electrical refrigerant pipelines to improve the weld spot detection efficiency and the detection pass rate. Summary of the invention

[0004] In view of the problem that the quality inspection of solder joints in the refrigeration and electrical industry is mostly manual inspection, resulting in low inspection efficiency and low inspection pass rate, the present invention proposes an intelligent solder joint leak detection system and leak detection method for electrical refrigerant pipelines, which reduces the phenomenon of missed inspection and wrong inspection caused by worker fatigue, is of great significance to improving the pass rate of refrigeration appliances, and has good industrial application prospects.

[0005] To achieve the above object, the present invention provides the following technical solutions: The present invention provides an intelligent welding spot leak detection system for electrical refrigerant pipelines, characterized in that it includes: a robot arm, a camera module, a leak detection gun, a computer system and a production line;

[0006] The robot arm and the camera module are both fixed on one side of the production line;

[0007] The leak detection gun is connected to the end of the manipulator;

[0008] The computer system is electrically connected to the camera module, the robot arm and the leak detection gun.

[0009] Preferably, the production line comprises a conveyor belt and a refrigeration appliance placed on the conveyor belt;

[0010] The refrigeration appliance includes a refrigerant pipeline area and a refrigeration appliance identification code.

[0011] Preferably, the robot comprises a robot arm and a robot end;

[0012] An extended thread is arranged at the end of the manipulator.

[0013] Preferably, a clamp is provided on the leak detection gun, and the clamp is fixed to the end of the robot arm through the extended thread.

[0014] Preferably, the camera module comprises a packaging shell; the packaging shell is connected with a flange;

[0015] The camera module further includes a left camera and a right camera, and the left camera and the right camera are fixed on the flange through screw holes;

[0016] The packaging shell is provided with an opening in the direction of the lenses of the left camera and the right camera, and a light source is provided on one side of the opening;

[0017] The packaging shell is also provided with a USB integrator, and the USB integrator is electrically connected to the computer system.

[0018] A leak detection method for an intelligent weld spot leak detection system for an electrical appliance refrigerant pipeline is also provided, which is characterized by comprising the following steps:

[0019] Install and fix the left camera and the right camera, and calibrate the left camera and the right camera to obtain camera internal and external parameters;

[0020] Perform system power-on and system pre-operation;

[0021] Controlling the left camera and the right camera to capture an image of the refrigeration appliance and perform image recognition, and performing image correction according to internal and external parameters of the camera to obtain a corrected image;

[0022] Obtaining a refrigerant pipe welding point image through the corrected image, measuring the camera world coordinates of the refrigerant pipe welding point, and calibrating the robot arm to obtain the robot arm world coordinates of the refrigerant pipe welding point;

[0023] According to the robot world coordinates of the refrigerant pipeline welding point, the robot arm is controlled to carry the leak detection gun to perform welding point leak detection, and the welding point leak detection results are recorded and sorted.

[0024] Preferably, the specific process of the robot arm calibration is: at the welding spot leak detection working distance, the checkerboard calibration plate is placed within the field of view of the left camera and the right camera, the checkerboard image is captured, and the camera world coordinates of each corner point in the checkerboard calibration plate are measured; the leak detection gun is carried by the robot arm to move, and when the leak detection gun touches each corner point in the checkerboard calibration plate, the robot arm world coordinates of each corner point are obtained; the camera world coordinates of each corner point are subtracted from the robot arm world coordinates to obtain the calibration coordinate difference, thereby completing the calibration of the robot arm.

[0025] Preferably, the measurement process of the camera world coordinates of the refrigerant pipe weld is as follows: based on the obtained refrigeration appliance image, a main image is obtained and Gaussian filtering is performed, and the filtered image is divided into different image areas according to the number of welds; each image area is matched separately according to the RGB channel information of the refrigerant pipe weld image to obtain welds with obvious features; the welds are identified and the positions of each weld are marked in the main image; the depth information of the refrigerant pipe area is obtained according to the internal and external parameters of the left camera and the right camera and the refrigeration appliance image; the camera world coordinates of the weld are obtained according to the depth information of the refrigerant pipe area and the positions of each weld in the main image.

[0026] Preferably, the method for acquiring the robot world coordinates of the refrigerant pipe weld is: adding the camera world coordinates of the weld and the calibration coordinate difference of the nearest corner point of the weld obtained by calibrating the robot to obtain the compensated coordinates; according to the welding pipe area and the actual working distance and field of view of the robot, the abnormal coordinates are eliminated to obtain the robot world coordinates of the refrigerant pipe weld.

[0027] Preferably, the weld leak detection process is specifically as follows: planning the path of the robot arm according to the robot arm world coordinates of the refrigerant pipe weld point, using the robot arm world coordinates of the weld point as the path node, and moving the leak detection gun by the robot arm according to the path node, when the end of the leak detection gun lightly touches the weld point or reaches a certain range from the weld point, stopping the movement and performing weld detection; after the detection is completed, controlling the robot arm carrying the leak detection gun to move to the next weld point, and continuing the detection.

[0028] The present invention discloses the following technical effects: the intelligent weld spot leak detection system and leak detection method for electrical appliance refrigerant pipelines disclosed by the present invention can identify the weld spots of the refrigerant pipelines of refrigeration appliances by visual detection, calculate their three-dimensional coordinate information, and finally transmit it to a robot arm, so that the robot arm replaces manual labor, which can not only improve the leak detection efficiency, but also reduce the phenomenon of missed detection and wrong detection caused by worker fatigue, which is of great significance to improving the qualified rate of refrigeration appliances and has good industrial application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0030] Figure 1 It is a schematic diagram of the overall structure of an intelligent welding spot leak detection system for electrical appliance refrigerant pipelines of the present invention;

[0031] Figure 2 A schematic diagram of a camera module in the intelligent leak detection system of the present invention;

[0032] Figure 3 It is a schematic diagram of the structure of the robot arm end, the leak detection gun and its fixture in the intelligent leak detection system of the present invention;

[0033] Figure 4 The present invention is a flow chart of an intelligent welding spot leak detection method for electrical appliance refrigerant pipelines. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0036] like Figure 1 As shown, this embodiment provides an intelligent welding spot leak detection system for electrical refrigerant pipelines, including: a robot arm 1, a camera module 2, a leak detection gun 3, a computer system 4 and a production line 5.

[0037] The production line 5 includes a conveyor belt 51 and a refrigeration appliance 50 placed on the conveyor belt 51 . The refrigeration appliance 50 includes a refrigerant pipeline welding point 500 and a refrigerator identification code 501 .

[0038] The robot arm 1 and the camera module 2 are fixed on one side of the production line 5. In order to avoid interference, the fixed position of the camera module 2 is limited outside the welding spot leak detection working range under the normal working condition of the robot arm 1; after the coordinate system calibration is completed, the relative position between the robot arm 1 and the camera module 2 remains unchanged.

[0039] The robot 1 of this embodiment is an AR5 six-axis robot with a repeatability of ±0.3mm, ensuring precise operation in complex applications. The robot 1 includes a robot arm and a robot end 10. The robot end 10 has a payload of 5kg and can carry a leak detection gun 3 to meet the requirements of refrigerant pipeline weld spot leak detection. The robot 1 of this embodiment can be set to one or more units, depending on the speed of the conveyor belt 51 and the number of weld spots in the refrigeration appliance 50.

[0040] This embodiment can transform the manual welding spot leak detection station on one side of the refrigeration appliance production line into a robot leak detection station, and can also cancel the manual welding spot leak detection station on one side of the refrigeration appliance production line to obtain a robot 1 leak detection station that can accommodate the working range of the robot 1.

[0041] Figure 2 The camera module includes a left camera 20 , a right camera 21 , a flange 22 , a packaging shell 23 , a light source 24 and a USB integrator 25 .

[0042] The left camera 20 and the right camera 21 are located in the outer shell 23, with the lenses facing the refrigerant pipe area 500 to be inspected, and the optical axes are perpendicular to the refrigerant pipe area 500; the field of view of the left camera 20 and the right camera 21 includes the entire refrigeration pipe area 500 and the refrigeration appliance identification code 501; after the left camera 20 and the right camera 21 are installed and fixed, the coordinate system is calibrated. The left camera 20 and the right camera 21 of this embodiment adopt Daheng MER-132-43U3C series CMOS industrial cameras, and the lens focal length is 8.5mm. The field of view range is 57.3°×43.8°(1 / 2″), and the resolution after correction can reach 0.2096mm / pixel, which meets the image requirements of the refrigerant pipe area 500.

[0043] The left camera 20 and the right camera 21 are installed together on the flange 22 , and screw holes 220 at different positions are reserved on the flange 22 for adjusting the positions of the left camera 20 and the right camera 21 to change the field of view and measurement accuracy.

[0044] The packaging shell 23 is connected to the flange 22 and can be disassembled. The packaging shell 23 is opened in the direction of the left camera 20 and the right camera 21 to avoid blocking the camera field of view. It is also used to protect the left camera 20 and the right camera 21 and provide mounting screw holes for the light source 24. The packaging shell 23 in this embodiment is made of aluminum alloy.

[0045] The light source 24 is installed outside the package shell 23, facing the refrigerant pipe area 500 of the electrical appliance, and is used to provide lighting for the refrigerant pipe area 500, improve the illumination distribution of the refrigerant pipe area 500, and improve the image quality captured by the left camera 20 and the right camera 21. The light source 24 of this embodiment uses two strip diffuse reflection light sources, Kemai Vision HF-TX14020, which can provide sufficient illumination in the refrigerant pipe area 500 to meet the image requirements of the refrigerant pipe area 500. At the same time, the voltage-stabilized power adapter of the light source 24 is powered by 220 volts and is installed inside the package shell 23 to provide stabilized DC power for the light source.

[0046] The USB hub 25 is installed on the packaging shell 23, and the bus is connected to the computer system 4, and is used to expand the USB interface of the computer system 4, wherein the two expanded interfaces are used to connect the left camera 20 and the right camera 21 respectively.

[0047] Figure 3 The schematic diagram of the structure of the leak detection gun of this embodiment. The leak detection gun 3 is fixed to the end of the robot arm 1 by a truncated cone-shaped fixture 30 installed on the extended thread 100 of the end of the robot arm, and the direction is perpendicular to the refrigerant pipeline area 500. After the installation of the leak detection gun 3, the robot arm 1 is calibrated. After the calibration, the leak detection gun 3 is not dismantled at will. The leak detection gun 3 of this embodiment adopts Qingdao Yijiayuan H25, which has a working range of 5mm and stays for more than 3 seconds to detect the leakage of the welding point, meeting the requirements of the refrigerant pipeline welding point leakage detection work.

[0048] The computer system 4 is installed near the robot 1 and the camera module 2, and its installation position is limited to outside the normal welding spot leak detection working range of the robot 1. The computer system 4 includes a computer host, a display, a keyboard, a mouse and other input devices, as well as a robot controller, a leak detection gun controller and a communication line connected to the computer host.

[0049] like Figure 4 As shown, a smart welding spot leak detection method for an electrical appliance refrigerant pipeline is also provided, comprising the following steps:

[0050] S1. Install the left camera 20 and the right camera 21 on the flange 22 and keep the relative position with the base of the robot 1 constant; calibrate the left camera 20 and the right camera 21 to obtain the internal and external parameters of the cameras.

[0051] This embodiment uses the Zhang Zhengyou calibration method to calibrate the left camera 20 and the right camera 21, that is, after fixing the left camera 20 and the right camera 21, at the working distance of the weld leak detection, the checkerboard calibration plate is placed within the field of view of the left camera 20 and the right camera 21, and the images of the calibration plate in different postures are taken by using the left camera 20 and the right camera 21, and imported into the Zhang Zhengyou calibration program, so as to obtain the internal and external parameters of the left and right cameras. The specific process is as follows:

[0052] definition represents a two-dimensional point on the image plane, and its augmented form is expressed as definition Represents a three-dimensional point in space, and the augmented form is

[0053] The conversion relationship can be obtained as follows:

[0054]

[0055] Where: s is the scale factor from the world coordinate system to the image coordinate system. For homogeneous coordinates, the scale factor does not change the coordinate value; Represents the intrinsic parameters of the camera; α and β represent the ratio of the image distance to the physical length of a pixel on the camera plate in the X and Y directions, respectively. The angle between the horizontal and vertical edges of the camera plate needs to be considered when calculating β; (u0, v0) are the coordinates of the image principal point; R is the rotation matrix; t is the translation matrix; Since pixels are not absolutely standard squares, γ is represented as the deviation of the pixel scale in the x and y directions.

[0056] By calculating the homography of formula (1), we get the following transformation form:

[0057]

[0058] in: [r1 r2 t] is the homography matrix, represented by H, that is, r1 and r2 represent the first and second columns of the rotation matrix R respectively; H is a 3×3 matrix, and one element is used as a homogeneous coordinate, so H has 8 unknown quantities to be solved, that is, at least eight equations are required, so four corresponding points are required. Four points can derive the homography matrix H from the image plane to the world plane.

[0059] From the matrix, we can know that by applying four points, we can obtain the homography matrix H, and H contains both the internal parameter matrix and the external parameter matrix, so we can first calculate the internal parameters and then the external parameters can be obtained.

[0060] definition:

[0061] Where: μ is the scale factor. Since r1T r2=0, so we can get two basic constraints:

[0062]

[0063] Among them: h1 and h2 are the values ​​solved by homography. At this time, only three photos of the calibration object plane are needed to solve the matrix Position parameter.

[0064] In this embodiment, when building a binocular camera detection platform, three different photos are obtained by changing the relative position between the binocular camera and the calibration plate. This process is completed during the platform construction and debugging stage.

[0065] definition:

[0066]

[0067] Define all unknown elements as a six-dimensional vector

[0068] Then the i-th column of the matrix H is Then get in:

[0069] Therefore, formulas (3) and (4) can be expressed as:

[0070]

[0071] By using three images with at least one checkerboard, B can be estimated by applying equations (4)-(6). After obtaining B, the camera's internal parameter matrix can be calculated by Cholesky decomposition: Specifically:

[0072]

[0073] The external parameters are calculated based on the calculated internal parameter matrix, and formula (2) can be obtained:

[0074]

[0075] Since equations (1) to (8) are calibration models established based on an idealized pinhole camera and do not consider geometric distortion, this embodiment adopts the maximum likelihood estimation method, substitutes the geometric distortion into the calibration model, and then establishes an evaluation function, which is specifically:

[0076]

[0077] Where: n is the number of images; m is the number of feature points; R irepresents the rotation matrix R corresponding to the i-th image; t i represents the translation matrix corresponding to the i-th image; M j Represents the corner point on the i-th image.

[0078] The Levenberg-Marquardt method is used to solve the minimum value of equation (9), thus completing the camera calibration.

[0079] S2, power on the leak detection system, turn on the computer system 4, start the robot control program, and reset the posture of the robot 1 to zero; start the camera control program, adjust the white balance, exposure value and other parameters of the left camera 20 and the right camera 21; start the leak detection gun control program; debug the communication between the robot 1 and the solder joint measurement program, and debug the communication between the leak detection gun 2 and the computer system 4.

[0080] This embodiment uses TCP communication to achieve communication between the computer system 4 and the robot 1 , the camera system 2 and the leak detection gun 3 .

[0081] S3. When the refrigeration appliance is transported through the production line 5 to the field of view of the left camera 20 and the right camera 21, the main program of the computer system 4 is started to control the left camera 20 and the right camera 21 to shoot and obtain RGB images, and based on the calibrated internal and external parameters of the left camera 20 and the right camera 21, the images obtained by the left camera 20 and the right camera 21 are corrected.

[0082] S4. At the working distance of the weld spot leak detection, the checkerboard calibration plate is placed within the field of view of the left camera 20 and the right camera 21, and the left camera 20 and the right camera 21 are controlled by a program in the computer system 4 to capture the checkerboard image. Then, the world coordinates (X, Y) of the camera of each corner point in the checkerboard calibration plate are measured by a visual measurement program. C (i),Y C (i),Z C (i)).

[0083] Among them, X C Indicates the X coordinate of each corner point in the camera's world coordinate system; Y C Indicates the Y coordinate of each corner point in the camera's world coordinate system; Z C represents the Z coordinate of each corner point in the world coordinate system of the camera; i represents the serial number of each corner point in the chessboard calibration plate (i=1, 2, 3, ...).

[0084] Control the robot 1 to move with the leak detection gun 3 until the end of the leak detection gun 3 pokes each corner point of the checkerboard calibration plate, while ensuring that the end joint of the robot 1 is horizontal, and record the three-dimensional coordinates in the robot control program to obtain the world coordinates (X R (i),Y R(i),Z R (i)).

[0085] Among them, X R represents the X coordinate of each corner point in the world coordinate system of robot 1; Y R represents the Y coordinate of each corner point in the world coordinate system of robot 1; Z R Indicates the Z coordinate of each corner point in the world coordinate system of the robot 1.

[0086] The world coordinates (X C (i),Y C (i),Z C (i)) and the obtained world coordinates (X R (i),Y R (i),Z R (i)) to perform a difference operation and obtain a set of coordinate difference values ​​(X D (i),Y D (i),Z D (i)), and obtain the final calibration result of robot arm 1.

[0087] S5. The conveyor belt 51 runs faster than the normal production line speed to send the refrigeration appliance 50 to the leak detection station. When the refrigeration appliance 50 arrives at the leak detection station, the conveyor belt 51 pauses, and the left camera 20 and the right camera 21 shoot the refrigeration appliance 50 and obtain an image. According to the image obtained by the left camera 20 and the right camera 21, the identification code 501 of the refrigeration appliance 50 is located and recognized, and the model and serial number of the corresponding refrigeration appliance 50 are obtained. According to the model and serial number of the refrigeration appliance 50, a dedicated welding point recognition program is called.

[0088] The refrigeration appliance 50 in this embodiment is a commonly used industrial or civilian appliance such as a refrigerator or an air conditioner.

[0089] S7. Select the main image taken by the left camera 20 and the right camera 21 as the basis for welding point recognition, and perform Gaussian filtering on the image. The filtered image is divided into different image areas according to the number of welding points. The number of divided sub-images can be 2, 3, etc., but the number should not be too large to avoid increasing the time complexity of the program.

[0090] According to the RGB channel information of the solder joints, the information is matched in the segmented image to identify the solder joints with obvious features, thereby improving the accuracy of solder joint image template matching.

[0091] The region growing algorithm is used to identify the weld spot at the end of the welded pipe, and the weld spot recognition results of the main image are supplemented by performing weld spot recognition on other images.

[0092] After all solder joints have been identified, the solder joint recognition program displays the main image and marks the locations of each solder joint.

[0093] The identification method of the image solder joints in the refrigerant pipe area 500 can adopt multiple matching methods, or multiple matching methods can be adopted at the same time, and finally a weighted matching method is adopted. The weighted matching method needs to calculate the color similarity (Bhattacharyya distance between HSV color histograms) and shape similarity (aHash value Hamming distance) between the target area photographed by the camera and the real solder joint through the color similarity measurement model and the shape similarity measurement model. Using two sets of similarity information for target recognition can improve the accuracy and applicability of solder joint recognition. When evaluating and judging a factor through two sets of feature information, the evaluation methods commonly used are: principal component analysis, data envelopment analysis, fuzzy evaluation method, weighted evaluation method, comprehensive evaluation method, etc.

[0094] This embodiment adopts a weighted evaluation method to establish a similarity evaluation model based on the actual situation of the production line and the fact that the automated leak detection system has high real-time performance and adaptability. The weighting function needs to be adjusted according to the different collected images. For example, in some refrigerant pipeline areas 500, the color in the area is close to the welding point and the shape is also similar. In this case, the weighted value should be set to an extreme value. In some areas with large differences in color and shape, the weighted value is further away from the extreme value. At this time, a linear model is used, and its formula is as follows:

[0095] S=a(1-d)+bh (10)

[0096] Among them: S is weighted similarity; d is color similarity; a is the weight of color similarity; h is shape similarity; b is the weight of shape similarity.

[0097] When d = 1, h = 0 (color and shape are the same), the weighted similarity S has an extreme value S min = 0. According to the actual situation of the refrigerant pipeline area 500, although the shape and color of each welding point are generally similar, they are different, so it is necessary to establish an evaluation threshold.

[0098] Set the threshold of the weighted similarity as S'. When a(1 - d)+bh < S', the target in the pipeline image region can be regarded as a solder joint; on the contrary, when a(1 - d)+bh > S', it is basically determined that the target in the region is not a solder joint. For different models of electrical appliances, the refrigerant pipeline region 500 is different, and its threshold S' is also different. The color similarity weight coefficient a and the shape similarity weight coefficient b in Equation (10) can be calculated from a large amount of pre-collected image data. Substitute a set of collected solder joint images into the threshold inequality, and calculate the ranges of the weight coefficients a and b through the inequality. Then, perform linear or non-linear fitting on the ranges of multiple sets of weight coefficients a and b to obtain the fitting values of the weight coefficients a and b.

[0099] The similarity weighting model of this embodiment is solved based on prior knowledge, and the accuracy and adaptability of the model can be enhanced through continuous training.

[0100] S8. According to the internal and external parameters of the left camera 20 and the right camera 21 obtained by calibration, the depth information of the refrigerant pipeline region is calculated and obtained by the vision measurement program based on the images of the left camera 20 and the right camera 21;

[0101] Take the center of each identified solder joint image as the image coordinates of each solder joint, and substitute them into the depth information obtained by the vision measurement program to obtain the world coordinates of the corresponding solder joint in the camera.

[0102] In the camera module 2, the left camera 20 and the right camera 21 have a translation difference on the X-axis. The line connecting the two optical centers is called the baseline. The u-axes of the coordinate systems of the two cameras coincide, the v-axes are parallel to each other, and the image planes of the left camera 20 and the right camera 21 are parallel and aligned. The intersection of the image plane and the camera optical axis is (u0, v0). Coincide the camera coordinate system of the left camera 20 with the world coordinate system, and the world coordinates of the solder joint P in the camera are (X P,C (i), Y P,C (i), Z P,C (i)). The pixel coordinates of the projection point p l are (u l , c l ), and the physical coordinates are (x l , y l ). The pixel coordinates of the point p r are (u r , v r ), and the physical coordinates are (x r , y r ).

[0103] For any solder joint P i , its pixel coordinates are respectively (u i , v i ), and its projections on the two cameras CL With C R The image points on the left and right cameras C are detected from the two images respectively. L With C R The label result, M L and M R are their projection matrices respectively, then:

[0104]

[0105] By eliminating the Z parameter of equation (11), the world coordinate information of the camera about the welding point (X P,C (i),Y P,C (i),Z P,C (i)) The operation matrix of the three unknowns is obtained, and the matrix parameters to be solved of the operation matrix are linear relations, specifically:

[0106]

[0107] From analytic geometry, we know that the plane equation in three-dimensional space is still a linear equation. When two plane equations are combined, they form the equation of a space line. The geometric meaning of equation (11) is that L With C R Based on the calibration parameters of the left camera 20 and the right camera 21, and by solving equation (12) in parallel, the three-dimensional coordinate information (X P,C (i),Y P,C (i),Z P,C (i))

[0108] S9, after obtaining the world coordinates of the camera of each welding point through visual measurement, the world coordinates of the welding point and the calibrated coordinate difference of the nearest corner point of the welding point are added to obtain the compensated coordinates.

[0109] Since there is a certain probability of gross errors in visual measurement, at this time, according to preset conditions such as the actual working distance between the weld pipe area and the robot 1 and the field of view, the coordinates that do not meet the preset conditions are eliminated, and the processed coordinates are output to the robot 1 via the computer. The robot control program plans the path of the robot 1, and the robot world coordinates of each weld point obtained after processing are used as path nodes. The robot 1 moves with the leak detection gun 3 based on the robot world coordinates of the weld point obtained by the final processing.

[0110] Between each path node, the robot's motion path can be selected as a linear motion, arc motion or other motion paths, and the motion speed and acceleration can be adjusted according to the production rhythm.

[0111] S10. According to the use requirements of the leak detection gun 3, the robot 1 carries the leak detection gun 3 and moves until the end of the leak detection gun 3 lightly touches each welding point or reaches a certain range from the welding point. At this time, there must be a dwell time at each welding point. At this time, the robot control program is in the timing stage and does not perform any movement. After the dwell time is reached, the robot control program automatically runs the next path node.

[0112] At each welding point, after the leak detection gun 3 finishes the detection, the computer system 4 communicates with the leak detection gun 3 to record and organize the detection results of each welding point.

[0113] The test results are saved, the robot control program controls the robot 1 to return to zero position, the leak detection gun program controls the leak detection probe of the leak detection gun 3 to zero, the main program enters the next cycle, the transmission belt 51 ends the pause state, and continues to run at a speed faster than the normal production line speed.

[0114] In the description of the present invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0115] The embodiments described above are only descriptions of the preferred modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A leak detection method for an intelligent weld spot leak detection system for an electrical refrigerant pipeline, wherein: An intelligent weld spot leak detection system for electrical appliance refrigerant pipelines comprises: a robot arm (1), a camera module (2), a leak detection gun (3), a computer system (4) and a production line (5), wherein the camera module comprises a left camera (20) and a right camera (21), the leak detection gun (3) is connected to the end of the robot arm (1), and the production line (5) comprises a conveyor belt (51) and a refrigeration appliance (50) placed on the conveyor belt (51), characterized in that the method comprises the following steps: The left camera (20) and the right camera (21) are installed and fixed, and the left camera (20) and the right camera (21) are calibrated to obtain camera internal and external parameters; Perform system power-on and system pre-operation; Controlling the left camera (20) and the right camera (21) to capture an image of the refrigeration appliance (50) and perform image recognition, and performing image correction according to internal and external parameters of the camera to obtain a corrected image; Obtaining a refrigerant pipeline welding point image through the corrected image, measuring the camera world coordinates of the refrigerant pipeline welding point, and calibrating the robot arm (1) to obtain the robot arm world coordinates of the refrigerant pipeline welding point; According to the robot world coordinates of the refrigerant pipeline welding point, the robot arm (1) is controlled to carry the leak detection gun (3) to perform welding point leak detection, and the welding point leak detection results are recorded and sorted; The specific process of calibrating the robot arm (1) is as follows: at a welding spot leak detection working distance, a checkerboard calibration plate is placed within the field of view of the left camera (20) and the right camera (21), a checkerboard image is captured, and the camera world coordinates of each corner point on the checkerboard calibration plate are measured; the robot arm (1) is equipped with the leak detection gun (3) to move, and when the leak detection gun (3) touches each corner point on the checkerboard calibration plate, the robot arm world coordinates of each corner point are obtained; the camera world coordinates of each corner point are subtracted from the robot arm world coordinates to obtain the calibration coordinate difference, thereby completing the calibration of the robot arm (1); The process of measuring the camera world coordinates of the refrigerant pipeline welding points is as follows: based on the obtained refrigeration appliance image, a main image is obtained and Gaussian filtering is performed, and the filtered image is divided into different image areas according to the number of welding points; each image area is matched according to the RGB channel information of the refrigerant pipeline welding point image to obtain welding points with obvious characteristics; the welding points are identified and the positions of each welding point are marked in the main image; the depth information of the refrigerant pipeline area is obtained according to the internal and external parameters of the left camera (20) and the right camera (21) and the refrigeration appliance image; and the camera world coordinates of the welding points are obtained according to the depth information of the refrigerant pipeline area and the positions of each welding point in the main image.

2. The leak detection method of the intelligent weld spot leak detection system for electrical refrigerant pipelines according to claim 1, characterized in that: The method for obtaining the robot world coordinates of the refrigerant pipeline welding point is as follows: adding the camera world coordinates of the welding point and the calibration coordinate difference of the nearest position corner point of the welding point obtained by calibrating the robot (1) to obtain the compensated coordinates; according to the actual working distance and field of view range between the welding pipe area and the robot (1), the abnormal coordinates are eliminated to obtain the robot world coordinates of the refrigerant pipeline welding point.

3. The leak detection method of the intelligent weld spot leak detection system for electrical refrigerant pipelines according to claim 1, characterized in that: The welding spot leak detection process specifically comprises: planning the path of the robot arm (1) according to the robot arm world coordinates of the welding spot of the refrigerant pipeline, taking the robot arm world coordinates of the welding spot as the path node, and moving the leak detection gun (3) through the robot arm (1) according to the path node; when the end of the leak detection gun (3) lightly touches the welding spot or reaches a certain range from the welding spot, stopping the movement and performing welding spot detection; after the detection is completed, controlling the robot arm (1) to carry the leak detection gun (3) to move to the next welding spot, and continuing the detection.

Images