WO2023134286A1

WO2023134286A1 - Online automatic quality testing and classification method for cathode copper

Info

Publication number: WO2023134286A1
Application number: PCT/CN2022/130854
Authority: WO
Inventors: 顾献代; 吴俊义; 张弛; 任禹桥; 方明; 吴家乐
Original assignee: 三门三友科技股份有限公司
Priority date: 2022-01-11
Filing date: 2022-11-09
Publication date: 2023-07-20
Also published as: CN114529510B; CN114529510A

Abstract

Disclosed in the present invention is an online automatic quality testing and classification method for cathode copper. The method comprises the following steps: performing primary image collection during a running process of a robot sheet-stripping unit; performing secondary image collection during a running process of a transverse conveying chain; performing online data analysis on collected images; and performing determination output according to an analysis result. In the technical solution, accurate image collection for a cathode copper plate is realized by means of two instances of image collection during a running process of a robot sheet-stripping unit and a running process of a transverse conveying chain, online data analysis is then performed on collected images according to a set standard, according to a data analysis result, different signals are given to a PLC by an industrial personal computer, and classification output of the cathode copper plate is realized by means of various forms, i.e. determination output of a robot unit, determination output of a chain unit, and manual intervention output, thereby greatly saving on the amount of labor, and improving the testing efficiency and accuracy.

Description

A method for automatic detection and classification of cathode copper online quality

technical field

The invention relates to the technical field of metal detection, in particular to an automatic detection and classification method for cathode copper online quality.

Background technique

The final link of the copper smelting industry is the refining workshop. The blister copper anode grows cathode copper on the surface of the cathode through electrolytic refining. After growing to a certain thickness, it is stripped by the cathode stripping unit. The stripped copper is electrolytic copper. At present, smelters process electrolytic copper according to different quality grades into the following categories: delivery grade A copper (including export copper), non-delivery grade A copper, No. 1 standard copper, and No. 2 standard copper, which cannot reach No. 2 Standard copper is required as scrap.

According to statistics, the amount of copper below grade A in each smelter is less than 10%, but at present, the judgment is mainly done manually, and there is only one output line plus an unqualified cathode copper station, and the operator sees unqualified products , When the product arrives at the designated station, press the reject button and the equipment will automatically place the unqualified product to the unqualified cathode copper position. The post-quality inspection personnel also need to re-inspect the products judged to be qualified by the operator. During the copper delivery period, personnel strengthened the inspection standards for unqualified products, and all grade A copper and below were judged unqualified, and the quality inspectors removed grade A copper in the follow-up. The technical inspection is mainly manual. The operator initially screens, and then the quality inspector re-inspects. If it is found to be unqualified, it will be re-classified and placed by a manual forklift. Manual inspection is costly, inefficient, and greatly affected by human influence. The whole process There is a lot of tedious and repetitive labor.

Chinese patent document CN211783203U discloses a "high-efficiency cathode copper surface quality detection system". A bottom plate is used, a base is installed on the upper side of the bottom plate, a row of parallel support rollers arranged in the transverse direction is installed on the base plate, and a limit frame is installed on the upper side of the bottom plate at the middle position in the transverse direction of the base , the limiting frame is equipped with a limiting bracket that can move up and down relative to it, a pair of parallel limiting rollers are installed on the limiting bracket, and the upper side of the bottom plate is installed at both ends of the base to push the cathode The propulsion mechanism for the movement of copper on support rolls. The above-mentioned technical solution still needs to manually realize the quality detection and judgment of cathode copper, and does not solve the technical problems of cumbersome process and heavy repetitive labor.

Contents of the invention

The present invention mainly solves the technical problems of cumbersome process and large amount of repetitive labor in the original technical solution, and provides an automatic detection and classification method for the online quality of cathode copper, which is realized by two image acquisitions during the operation process of the robot stripping unit and the operation process of the horizontal feeding chain. Accurate image collection of the cathode copper plate, and then online data analysis of the collected images according to the set standards, according to the data analysis results, the industrial computer will give different signals to the PLC, through the robot unit judgment output, chain unit judgment output and manual intervention output. This method realizes the classified output of cathode copper plates, which greatly saves labor and increases detection efficiency and accuracy.

Above-mentioned technical problem of the present invention is mainly solved by following technical scheme: the present invention comprises the following steps:

An image acquisition is performed during the operation of the S1 robot stripping unit;

S2 carries out secondary image acquisition during the operation of the cross-feeding chain;

S3 conducts online data analysis on the collected images;

S4 performs judgment output according to the analysis result.

As preferably, one image acquisition in step S1 includes the following steps:

S1.1 Wash the cathode copper plate under test and obtain the front and back surface images of the cathode copper plate through the line array camera image acquisition mechanism, and send them to the industrial computer;

S1.2 Carry out image acquisition through the line array camera image acquisition mechanism;

S1.3 Use the 3D laser scanning 3D modeling mechanism to reproduce the Y and Z axis dimensions of each frame of laser irradiation lines and merge them into a 3D model.

As preferably, the secondary image acquisition in step S2 includes the following steps:

S2.1 Take pictures with the line-scan camera during the operation of the horizontal feed chain;

S2.2 Carry out 3D laser scanning and three-dimensional image acquisition during the operation of the cross-feeding chain;

S2.3 Take pictures with the area array camera during the operation of the transverse feed chain.

As a preference, said step S1.1 area array camera image taking mechanism includes a front camera arranged between the robot stripping unit and the cathode copper plate to be tested, the front camera is provided with a front light source, and the cathode copper plate to be tested is far away from the robot One side of the peeling unit is provided with a reverse camera, and a reverse light source is provided next to the reverse camera, and the front camera and the reverse camera are connected through the industrial computer and the PCL in turn; the step S2. The reverse light source on one side is opposite to the reverse camera set on the side of the cathode copper plate to be tested, and the reverse camera is connected through the industrial computer and PCL in turn. A high-brightness projection light source is used to illuminate the cathode copper plate to be detected to ensure that the light source irradiation range includes the entire copper plate. The color camera captures the positive and negative surface images of the cathode copper plate through a high-quality lens, and sends them to the industrial computer.

As a preference, said step S1.2 line array camera image taking mechanism includes a front camera facing the transmission mechanism, a front light source and a reverse camera inserted towards the connecting board, and a reverse light source, and the front camera and the reverse camera are connected sequentially through the industrial computer and the PCL ; The step S2.1 line array camera image acquisition mechanism includes a front camera, a front light source, a back camera, and a back light source respectively arranged on both sides of the transmission mechanism, and the front camera and the back camera are sequentially connected through an industrial computer and a PCL. The line-array CCD camera and line light source are used to ensure that each frame of the picture is cleaned, and the code reader is used to ensure that the collected pictures are clear and usable. During the cathode copper transfer process, any position that conforms to the principle of line-scan camera image acquisition is acceptable, and the detection station can also be added separately or the existing mechanism can be rectified to achieve image acquisition.

Preferably, the step S1.3 3D laser scanning three-dimensional modeling mechanism includes a front 3D laser scanner facing the transmission mechanism and a reverse 3D laser scanner inserted toward the board, a front 3D laser scanner and a reverse 3D laser scanner Connected through the industrial computer and PCL in turn; the step S2.2 3D laser scanning three-dimensional modeling mechanism includes a front 3D laser scanner and a back 3D laser scanner respectively arranged on both sides of the transmission mechanism, a front 3D laser scanner and a back 3D laser scanner. The laser scanner is connected through the industrial computer and PCL in turn. Using industrial cameras and line lasers with code readers, the Y and Z axis dimensions of each frame of line laser irradiation lines are reproduced, and finally merged into a 3D model.

Preferably, said step S3 includes two-dimensional photo processing and three-dimensional model processing, specifically including:

S3.1 Preprocess the collected images, mainly divide the original image into multiple images for processing; preprocess the cathode copper surface images obtained by the industrial color camera.

S3.2 Annotate copper particles on the preprocessed cathode copper surface image to generate an annotation file, and the preprocessed cathode copper surface image and the annotated image form a copper particle defect data set;

S3.3 trains the copper particle defect data set so that the deep learning network can analyze the copper particle pixels;

S3.4 Set different testing standards according to different quality requirements, and output according to grades; set different testing standards according to different quality requirements, and output according to grades. The output parameters can be freely set according to a shape, b size, and c quantity. The degree of freedom is high and the customization is strong. The defect can be qualitatively determined by setting a convex coefficient, b ambiguity, and c aspect ratio.

S3.5 will detect the cathode copper output signal of NG to the PLC, and the PLC will record it and remove it after stripping;

S3.6 Calculate the three-dimensional coordinate data of the surface of the measured object, and analyze its surface condition according to the obtained information.

Preferably, the step S3.3 specifically includes applying a pyramid scene analysis network, obtaining the feature Map of the preprocessed image through convolution operations, and then through the multi-resolution convolution and fusion of the pyramid pooling module, the network is segmented into Background pixels and copper particle pixel categories, through copper particle image training, after the network recognition copper particle pixel accuracy reaches the specified value, a network parameter file is generated for the actual online copper particle detection system.

Preferably, the step S3.6 specifically includes: based on the triangulation method, the laser line is used as structured light, the camera captures the image to calculate the three-dimensional coordinate data of the surface of the measured object, and directly analyzes the surface condition according to the obtained information. The surface of the object is irregular, so it analyzes and compares the output of surface defects according to the relative data within a specific size and length range.

As a preference, said step S4 determines the output mode including the robot unit determination output, the chain unit determination output and the manual intervention output, and the industrial computer gives different signals to the PLC,

The robot unit judges and outputs different results. The PLC controls the robot to classify and place the copper plates according to the different signals received. The output path of cathode copper plate quality 1, the output path of cathode copper plate quality 2, the output path of cathode copper plate quality 3, and unqualified Cathode copper plate output path for output;

The method of judging and outputting different results by the chain unit is that the PLC does not peel off the abnormal copper cathode plate at the detection station, directly rejects it, and peels it separately at the rejecting station later;

Manual intervention, output judgment, output different results, the PLC gives different prompts according to different signals, and manually outputs according to the prompts.

The beneficial effects of the present invention are: the accurate image acquisition of the cathode copper plate is realized through the two image acquisitions of the operation process of the robot stripping unit and the operation process of the cross-feeding chain, and then online data analysis is performed on the acquired image according to the set standard, and the industrial control system is controlled according to the data analysis result. The machine gives different signals to the PLC, and realizes the classification output of cathode copper plates through various forms of robot unit judgment output, chain unit judgment output and manual intervention output, which greatly saves labor and increases detection efficiency and accuracy.

Description of drawings

Fig. 1 is a kind of flowchart of the present invention.

Fig. 2 is a diagram of an area array camera image acquisition mechanism during the operation of a robot peeling unit of the present invention.

Fig. 3 is a diagram of a line-scan camera image-taking mechanism in the operation process of a robot peeling unit of the present invention.

Fig. 4 is a 3D laser scanning three-dimensional modeling mechanism diagram during the operation of a robot peeling unit of the present invention.

Fig. 5 is a diagram of a line-scan camera image-taking mechanism in the running process of a transverse feed chain of the present invention.

Fig. 6 is a 3D laser scanning three-dimensional modeling mechanism diagram of the running process of a transverse feed chain of the present invention.

Fig. 7 is a diagram of an area-scan camera image-taking mechanism in the running process of a cross-feeding chain of the present invention.

Fig. 8 is a diagram of a robot unit judgment output system of the present invention.

Fig. 9 is a diagram of a judgment output system of a chain unit according to the present invention.

In the figure, 1 front camera, 2 front light source, 3 cathode copper plate to be tested, 4 back light source, 5 back camera, 6 industrial computer, 7PCL, 8 front 3D laser scanner, 9 back 3D laser scanner, 10 detection station, 11 Rejection station, 12 output paths of cathode

copper plate quality

1, 13 output paths of cathode copper plate quality 2, 14 output paths of cathode

copper plate quality

3, 15 output paths of unqualified cathode copper plates, 16 board sockets.

Detailed ways

The technical solutions of the present invention will be further specifically described below through the embodiments and in conjunction with the accompanying drawings.

Embodiment: A kind of cathode copper online quality automatic detection classification method of this embodiment, as shown in Figure 1, comprises the following steps:

An image acquisition is performed during the operation of the S1 robot peeling unit, including the following steps:

S1.1 Wash the cathode copper plate to be tested and obtain the front and back surface images of the cathode copper plate through the line array camera image acquisition mechanism, and send them to the industrial computer. As shown in Figure 2, the area array camera image taking mechanism includes a front camera 1 arranged between the robot stripping unit and the cathode copper plate 3 to be tested, the front camera 1 is provided with a front light source 2, and the cathode copper plate 3 to be tested is far away from One side of the robot peeling unit is provided with a reverse camera 4, and a reverse light source 5 is provided beside the reverse camera 4, and the front camera 1 and the reverse camera 4 are connected successively through the industrial computer and the PCL7. A high-brightness projection light source is used to illuminate the cathode copper plate to be detected to ensure that the light source irradiation range includes the entire copper plate. The color camera captures the positive and negative surface images of the cathode copper plate through a high-quality lens, and sends them to the industrial computer.

S1.2 Image acquisition is carried out through the line-scan camera image acquisition mechanism. As shown in Figure 3, the line scan camera image acquisition mechanism includes a front camera 1 facing the transmission mechanism, a front light source 2, and a back camera 4 and a back light source 5 plugged in toward the board. The front camera 1 and the back camera 4 pass through the industrial computer, PCL7 is connected. The line-array CCD camera and line light source are used to ensure that each frame of the picture is cleaned, and the code reader is used to ensure that the collected pictures are clear and usable. During the cathode copper transfer process, any position that conforms to the principle of line-scan camera image acquisition is acceptable, and the detection station can also be added separately or the existing mechanism can be rectified to achieve image acquisition.

S1.3 Use the 3D laser scanning 3D modeling mechanism to reproduce the Y and Z axis dimensions of each frame of laser irradiation lines and merge them into a 3D model. As shown in Figure 4, the 3D laser scanning three-dimensional modeling mechanism includes a front 3D laser scanner 8 facing the transmission mechanism and a reverse 3D laser scanner 9 inserted toward the board, the front 3D laser scanner 8 and the reverse 3D laser scanner 9 In turn through the industrial computer, PCL7 connected. Using industrial cameras and line lasers with code readers, the Y and Z axis dimensions of each frame of line laser irradiation lines are reproduced, and finally merged into a 3D model.

Secondary image acquisition is performed during the operation of the S2 cross-feeding chain, including the following steps:

S2.1 Take pictures with the line scan camera during the operation of the transverse feed chain. As shown in Figure 5, the line array camera image acquisition mechanism includes a front camera 1, a front light source 2, a back camera 4, and a back light source 5 respectively arranged on both sides of the transmission mechanism. The front camera 1 and the back camera 4 pass through the industrial computer, PCL7 connected.

S2.2 Carry out 3D laser scanning and three-dimensional drawing during the operation of the cross-feeding chain. As shown in Figure 6, the step S2.2 3D laser scanning three-dimensional modeling mechanism includes the front 3D laser scanner 8 and the back 3D laser scanner 9 respectively arranged on both sides of the transmission mechanism, the front 3D laser scanner 8 and the back The 3D laser scanner 9 is connected sequentially through the industrial computer and the PCL7.

S2.3 Take pictures with the area array camera during the operation of the transverse feed chain. As shown in Figure 7, the area array camera image-taking mechanism includes a reverse light source 5 arranged on one side of the cathode copper plate 3 to be tested and a reverse camera 4 oppositely arranged on the side of the cathode copper plate 3 to be tested, and the reverse camera 4 sequentially passes through the industrial computer, PCL7 is connected.

S3 conducts online data analysis on the collected images, including two-dimensional photo processing and three-dimensional model processing, including:

S3.1 preprocessing the collected image, mainly dividing the original image into multiple images for processing;

S3.3 trains the copper particle defect dataset, enabling the deep learning network to analyze copper particle pixels. Specifically, it includes applying the pyramid scene analysis network, obtaining the feature map of the preprocessed image through convolution operation, and then through the multi-resolution convolution and fusion of the pyramid pooling module, the network is segmented into background pixels and copper particle pixel categories. Particle image training, after the pixel accuracy of network recognition of copper particles reaches the specified value, a network parameter file is generated for the actual online copper particle detection system.

S3.5 will detect the cathode copper output signal of NG to PLC7, and PLC7 will record and remove it after stripping;

S3.6 Calculate the three-dimensional coordinate data of the surface of the measured object, and analyze its surface condition according to the acquired information. Specifically include: based on the triangulation method, the laser line is used as structured light, the camera takes pictures to calculate the three-dimensional coordinate data of the surface of the measured object, and directly analyzes the surface condition according to the obtained information. Since the surface of the measured object is irregular, according to the specific size length The relative data within the range is analyzed and compared to output surface defects. S4 performs judgment output according to the analysis result. The judgment output mode includes the robot unit judgment output, the chain unit judgment output and the manual intervention output. The industrial computer 6 gives different signals to the PLC7.

As shown in Figure 8, the way the robot unit judges and outputs different results is that the PLC7 controls the robot to classify and place the copper plates according to the different signals received, and the cathode copper plate quality 1 output path 12, the cathode copper plate quality 2 output path Copper plate quality 3 output path 14, unqualified cathode copper plate output path 15 for output;

As shown in Figure 9, the way the chain unit judges and outputs different results is that the PLC7 does not peel off the abnormal copper cathode plate at the detection station 10, and directly rejects it, and then peels it separately at the rejecting station 11 in the later stage;

Manual intervention, output judgment, and different output methods are that PLC7 gives different prompts according to different signals, and manually outputs according to the prompts.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs can make various modifications or supplements to the described specific embodiments or adopt similar methods to replace them, but they will not deviate from the spirit of the present invention or go beyond the definition of the appended claims range.

Although terms such as image acquisition, data analysis, and judgment output are frequently used in this paper, the possibility of using other terms is not excluded. These terms are used only for the purpose of describing and explaining the essence of the present invention more conveniently; interpreting them as any kind of additional limitation is against the spirit of the present invention.

Claims

A cathode copper online quality automatic detection and classification method is characterized in that it comprises the following steps:

An image acquisition is performed during the operation of the S1 robot stripping unit;

S2 carries out secondary image acquisition during the operation of the cross-feeding chain;

S3 conducts online data analysis on the collected images;

S4 performs judgment output according to the analysis result.
A kind of cathode copper online quality automatic detection classification method according to claim 1, is characterized in that, described step S1 once image collection comprises the following steps:

S1.1 Wash the cathode copper plate under test and obtain the front and back surface images of the cathode copper plate through the line array camera image acquisition mechanism, and send them to the industrial computer;

S1.2 Carry out image acquisition through the line array camera image acquisition mechanism;

S1.3 Use the 3D laser scanning 3D modeling mechanism to reproduce the Y and Z axis dimensions of each frame of laser irradiation lines and merge them into a 3D model.
A kind of cathode copper online quality automatic detection classification method according to claim 1, is characterized in that, described step S2 secondary image acquisition comprises the following steps:

S2.1 Take pictures with the line-scan camera during the operation of the horizontal feed chain;

S2.2 Carry out 3D laser scanning and three-dimensional image acquisition during the operation of the cross-feeding chain;

S2.3 Take pictures with the area array camera during the operation of the transverse feed chain.
According to claim 2 or 3, a cathode copper online quality automatic detection and classification method is characterized in that, said step S1.1 area array camera image acquisition mechanism includes a robot stripping unit and a cathode copper plate to be tested (3) Between the front camera (1), the front camera (1) is provided with a front light source (2), and the side of the cathode copper plate to be tested (3) away from the robot stripping unit is provided with a back camera (4), and the back camera (4) is provided with a reverse light source (5) next to it, and the front camera (1) and the reverse camera (4) are connected successively through the industrial computer and the PCL (7); The reverse light source (5) on one side of the cathode copper plate (3) to be tested is opposite to the reverse camera (4) arranged on the side of the cathode copper plate (3) to be tested, and the reverse camera (4) is connected through the industrial computer and the PCL (7) in sequence.
A method for automatic detection and classification of cathode copper online quality according to claim 2 or 3, characterized in that the step S1.2 line array camera image taking mechanism includes a front camera (1) facing the transmission mechanism, a front light source ( 2) The reverse camera (4), the reverse light source (5), the front camera (1) and the reverse camera (4) which are plugged towards the connecting board are connected successively through the industrial computer and the PCL (7); the step S2.1 line array The camera image acquisition mechanism includes front camera (1), front light source (2), back camera (4), back light source (5) respectively arranged on both sides of the transmission mechanism, front camera (1) and back camera (4) pass through in sequence The industrial computer and PCL (7) are connected.
A method for automatic detection and classification of cathode copper online quality according to claim 2 or 3, characterized in that, said step S1.3 3D laser scanning three-dimensional modeling mechanism includes a front 3D laser scanner (8) facing the transmission mechanism Connect to the reverse 3D laser scanner (9) inserted towards the board, the front 3D laser scanner (8) and the reverse 3D laser scanner (9) successively through the industrial computer and the PCL (7); the step S2.2 3D The laser scanning three-dimensional modeling mechanism includes a front 3D laser scanner (8) and a back 3D laser scanner (9) respectively arranged on both sides of the transmission mechanism, a front 3D laser scanner (8) and a back 3D laser scanner (9) Connect through industrial computer and PCL (7) successively.
A method for automatic detection and classification of cathode copper online quality according to claim 3, wherein said step S3 includes two-dimensional photo processing and three-dimensional model processing, specifically including:

S3.1 preprocessing the collected image, mainly dividing the original image into multiple images for processing;

S3.2 Annotate copper particles on the preprocessed cathode copper surface image to generate an annotation file, and the preprocessed cathode copper surface image and the annotated image form a copper particle defect data set;

S3.3 trains the copper particle defect data set so that the deep learning network can analyze the copper particle pixels;

S3.4 Set different testing standards according to different quality requirements, and output according to grades;

S3.5 will detect the cathode copper output signal of NG to PLC (7), and be removed after being stripped by PLC (7) record;

S3.6 Calculate the three-dimensional coordinate data of the surface of the measured object, and analyze its surface condition according to the obtained information.
A kind of cathode copper online quality automatic detection and classification method according to claim 7, it is characterized in that, described step S3.3 specifically comprises applying pyramid scene analysis network, obtains the feature Map of preprocessing image by convolution operation, and then Through the multi-resolution convolution and fusion of the pyramid pooling module, the network is segmented into background pixels and copper particle pixel categories. Through copper particle image training, after the network recognizes copper particle pixel accuracy reaches the specified value, a network parameter file is generated. In the actual online copper particle detection system.
The method for automatic detection and classification of cathode copper online quality according to claim 7, wherein said step S3.6 specifically includes: based on triangulation method, the laser line is used as structured light, and the camera takes a picture to calculate the surface of the measured object According to the obtained information, the surface condition is directly analyzed. Since the surface of the measured object is irregular, the relative data within a specific size and length range is analyzed and compared to output the surface defect condition.
A method for automatic detection and classification of cathode copper online quality according to claim 4, characterized in that, said step S4 determines the output mode including robot unit determination output, chain unit determination output and manual intervention output, and is controlled by the industrial computer (6) Give different signals to PLC(7),

The way that the robot unit judges and outputs different results is that the PLC (7) controls the robot to classify and place the copper plates according to the received different signals, respectively by the cathode copper plate quality 1 output path (12), the cathode copper plate quality 2 output path (13), Cathode copper plate quality 3 output path (14), unqualified cathode copper plate output path (15) for output;

The way the chain unit judges and outputs different results is that the PLC (7) does not peel off the abnormal copper cathode plate at the detection station (10), directly rejects it, and peels it separately at the rejecting station (11) in the later stage;

Manual intervention, output judgment, and different result output methods are that the PLC (7) gives different prompts according to different signals, and manually outputs according to the prompts.