Machine vision-based automatic docking system and method for reduction tank

Abstract

The application provides a machine vision-based automatic butt joint system and method for reduction tanks, and belongs to the technical field of magnesium metal smelting; solves the problem that the standard and uniformity of the longitudinal and transverse positions of the reduction tank cannot be guaranteed during installation or production tank replacement, and the problem of low butt joint efficiency caused by tank body deformation in high-temperature production; comprises a coarse positioning subsystem, a fine positioning subsystem and a control subsystem; the coarse positioning subsystem is used for realizing coarse positioning of the position of the reduction tank to be loaded; the fine positioning subsystem realizes accurate butt joint of the guide cylinder and the reduction tank to be loaded through machine vision; the fine positioning subsystem is integrated with a fine adjustment algorithm, a fine positioning algorithm and a fine adjustment algorithm; the fine adjustment algorithm is used for realizing preliminary determination of the position of the guide cylinder; the fine positioning algorithm is used for simultaneously starting two guide cylinders to perform butt joint with respective reduction tanks; and the fine adjustment algorithm is used for realizing final butt joint of the guide cylinder and the reduction tank; the application is applied to automatic butt joint of the horizontal reduction tank and the guide cylinder for magnesium metal.

Description

TECHNICAL FIELD

[0001] The application provides a machine vision-based automatic butt joint system and method for reduction pots, and belongs to the technical field of magnesium smelting. BACKGROUND

[0002] The Pidgeon process is a common magnesium smelting process, and has a relatively mature process flow, small early-stage investment, high purity of finished magnesium, and rich dolomite resources as a magnesium smelting raw material. During smelting, a worker first puts the dolomite pellets into a high-temperature reduction pot, then seals the reduction pot for 12 hours, allows the dolomite pellets to fully react, and finally removes the slag from the reduction pot.

[0003] At present, most magnesium plants use manual pushing. The manual pushing method is that a worker pushes the pellets into the reduction pot one by one through a pushing rod until the loading is qualified. Manual pushing leads to long loading time, high labor intensity, low work efficiency, unstable production indexes, and a relatively poor working environment. Long-term dust and high-temperature environment seriously harm the health of workers, and fewer and fewer workers engage in pushing. Some magnesium plants use a forklift to push the material, which improves the efficiency and reduces the physical exertion of workers, but the forklift driver is still in a dust and high-temperature environment, which is not conducive to health.

[0004] At present, there are three main ways of butt joint for coal-based magnesium loading: manual butt joint, semi-automatic butt joint, and automatic butt joint.

[0005] (1) Manual butt joint mainly refers to that a worker stirs dolomite into the reduction pot one shovel at a time, and the butt joint is eye aiming and shovel butt joint. An improved version of manual butt joint is that a worker inserts a guide barrel into the reduction pot, then continuously pours dolomite into the guide barrel, and the worker pushes the dolomite in the guide barrel into the reduction pot one by one using a horseshoe iron. At this time, the butt joint is eye aiming, and the guide barrel is butt jointed with the reduction pot. An advanced version can use a forklift to replace the worker to push the reduction pot back and forth. At this time, the worker on the forklift needs to master the force back and forth, which is consistent with the above. The butt joint method is eye aiming, and the guide barrel is butt jointed with the reduction pot.

[0006] (2) Semi-automatic docking involves a mechanical loading trolley, which is usually installed on a workshop track and moves along the track. During loading, workers do not directly participate, but operators still need to control the loading trolley on-site using a remote control to dock the guide bucket of the loading trolley with the reduction tank. An improved method is to install a camera at a suitable position on the trolley, allowing operators to control the loading trolley by observing the camera images from inside the trolley, thus docking the guide bucket of the loading trolley with the reduction tank. In short, whether observing visually or through a camera, operators need to be proficient in docking techniques. Furthermore, similar to manual methods, the long-term exposure to dust and high temperatures can negatively impact the health of operators.

[0007] (3) Fully automatic docking refers to being completely controlled by a mechanical loading trolley without the need for human intervention. Currently, the fully automatic docking system for reduction tanks is not yet mature and is difficult to achieve in practical applications.

[0008] To address the aforementioned problems, this invention proposes an automatic docking system for restoration tanks based on machine vision, achieving precise and rapid docking between the guide cylinder and the restoration tank. This system effectively meets the requirements for creep adaptability of the deformable tank and uniform material loading. Summary of the Invention

[0009] To address the issues of inconsistent longitudinal and transverse positioning during the installation or production of reduction tanks, coupled with low docking efficiency due to tank deformation caused by high temperatures during production, this invention provides an automated docking system and method for reduction tanks based on machine vision.

[0010] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: an automatic docking system for reduction tanks based on machine vision, including a coarse positioning subsystem, a fine positioning subsystem, and a control subsystem. The coarse positioning subsystem is used to achieve coarse positioning of the reduction tank to be loaded. The fine positioning subsystem achieves precise docking of the guide cylinder and the reduction tank to be loaded through machine vision. The control subsystem is used to control the electrical components in the coarse positioning subsystem and the fine positioning subsystem.

[0011] The fine positioning subsystem integrates a fine adjustment algorithm, a fine positioning algorithm, and a fine adjustment algorithm. The fine adjustment algorithm is used to control the position of the moving trolley after the coarse positioning of the reduction tank is completed and before the left and right guide cylinders are simultaneously connected to the reduction tank, so that the distance between the guide cylinders and the corresponding calibration points is within the range.

[0012] The precise positioning algorithm is used to simultaneously activate two guide cylinders to dock with their respective reduction tanks;

[0013] The fine-tuning algorithm is used to achieve the final docking of the guide cylinder and the reduction tank.

[0014] The coarse positioning subsystem includes RFID electronic tags, card readers, a mobile trolley, and coarse positioning software. The mobile trolley is installed on a track in the workshop and moves along the track. Each reduction tank is equipped with a numbered, high-temperature resistant RFID electronic tag. The mobile trolley is equipped with a card reader. The coarse positioning software sends commands to the control subsystem to control the movement and stopping of the mobile trolley based on the information of the RFID electronic tags read by the card reader, so that the mobile trolley stops near the reduction tank to be loaded.

[0015] The precision positioning subsystem includes a camera, a guide cylinder, a moving platform, and precision positioning software. The camera is mounted on the guide cylinder to capture video images of the reduction tank. The guide cylinder is mounted on the moving platform for loading materials after the reduction tank is docked. The moving platform can move within a set range in the horizontal and vertical directions and drive the guide cylinder to move in the same direction.

[0016] The precision positioning software has built-in computer programs for fine-tuning, precision positioning, and fine-tuning algorithms. The fine-tuning algorithm keeps the position of the moving trolley within the calibrated range, the precision positioning algorithm accurately positions the guide cylinder with the corresponding reduction tank, and the fine-tuning algorithm enables continuous automatic docking between the guide cylinder and the reduction tank.

[0017] The camera is specifically installed at the rear and top of the guide tube.

[0018] The automatic docking method for restoration tanks based on machine vision employs an automatic docking system for restoration tanks, and includes the following steps:

[0019] S1: Coarse positioning of the guide tube: During the movement of the mobile trolley, the information of the RFID electronic tag installed on the reduction tank is read in real time using a card reader. If the reading result meets the requirements, the coarse positioning software sends a command to the control subsystem to require the mobile trolley to stop moving. Otherwise, the mobile trolley continues to move until it reads an RFID electronic tag that meets the conditions and finds the position of the reduction tank to be loaded.

[0020] S2: Center point calibration of the reduction tank to be filled and guide cylinder calibration: Select the reduction tank with the RFID electronic tag that meets the conditions installed in step S1, and then guide the guide cylinder into the reduction tank manually. The coordinates of the guide cylinder are calibrated at this time. Then, the reduction tank in the current image is detected and identified using deep learning-based target detection technology. If the identification is successful, the center point coordinates of the reduction tank are calculated through the detection box, and the final calibration center point coordinates of the reduction tank are obtained by calculating the median of the nth reduction tank detection and identification structure.

[0021] S3: Fine-tuning the moving trolley through a fine-tuning algorithm: Before the left and right guide cylinders are simultaneously connected to the reduction tank, the position of the moving trolley is controlled by the control subsystem so that the distance between the guide cylinders and the corresponding calibration points is within the range.

[0022] S4: Precise positioning of the guide cylinder is achieved through a precision positioning algorithm: the movement of the mobile platform is controlled based on the allowable error between the coordinates of the center point of the restoration tank detected in the video image and the coordinates of the calibration point of the guide cylinder, thereby completing the precise positioning of the guide cylinder and the restoration tank.

[0023] S5: The guide cylinder position is precisely adjusted through a fine-tuning algorithm to achieve docking with the restoration tank: The target detection algorithm is used to identify the restoration tank in the current image, calculate the coordinates of the center point of the current restoration tank after fine-tuning, and call the fine positioning algorithm based on the coordinates of the center point of the restoration tank after fine-tuning to finally complete the docking of the guide cylinder and the restoration tank.

[0024] The specific steps of the fine-tuning algorithm are as follows:

[0025] Let the ranges of the two moving platforms mounted on the two guide cylinders be C1 and C2, respectively, and assume that C1 ≤ C2. Let the x-coordinate of the calibration point on the left guide cylinder be X. A The horizontal coordinate of the current position of the left guide tube is X' A The x-coordinate of the calibration point of the right guide tube is X. B The x-coordinate of the current position of the right guide tube is X' B ;

[0026] The calculation process of the fine-tuning algorithm is as follows:

[0027] S31: Calculate the difference between the x-coordinate of the left guide tube calibration point and its current x-coordinate: a = X A -X' A ;

[0028] S32: Calculate the difference between the x-coordinate of the right guide tube calibration point and its current x-coordinate: b = X B -X' B ;

[0029] S33: Determine whether the left and right guide cylinders are within the range: a≤C1 and b≤C2;

[0030] S34: If within the range: no adjustment is needed, the algorithm ends;

[0031] S35: If not within the range, then calculate:

[0032] When |ab|>2C1, it means that the left and right guide tubes cannot enter at the same time and need to enter one tube at a time, and the algorithm ends.

[0033] When |ab| < 2C1, it means that the left and right guide cylinders can advance simultaneously by adjustment. The process of adjusting the moving trolley is as follows:

[0034] When a > 0 and b > 0, the moving cart moves to the left by |a + b| / 2.

[0035] When a < 0 and b < 0, the moving cart moves to the right by |a + b| / 2.

[0036] When a+b>0, the moving cart moves to the left by |a+b| / 2;

[0037] When a+b<0, the moving cart moves to the right by |a+b| / 2;

[0038] S36: Send a command to the control subsystem to control the mobile trolley to move to the adjusted position;

[0039] S37: After the trolley moves, check whether the positions of the left and right guide cylinders are within the range of C2;

[0040] S38: If present, adjustment stops, algorithm ends;

[0041] S39: Otherwise, proceed to step S4 to continue the adjustment.

[0042] The specific steps of the precise localization algorithm are as follows:

[0043] The precision positioning algorithm includes simultaneous precision positioning of the left guide cylinder and the right guide cylinder. The principle of precision positioning of the left and right guide cylinders is the same. The adjustment process for precision positioning of the left guide cylinder is as follows:

[0044] Let the coordinates of the calibration point of the left guide tube be (X... A Y A ), where ε represents the allowable error value between the coordinates of the center point of the restoration tank detected in the video image and the coordinates of the calibration point;

[0045] S41: Calculate the difference between the current horizontal position of the left guide tube to be docked with the reduction tank and the x-coordinate of its calibration point:

[0046] c = X' A -X A ;

[0047] Where X' A X is the x-coordinate of the current position of the left guide tube. A The x-coordinate of the calibration point of the left guide tube;

[0048] S42: When c > ε, the mobile platform moves c pixels horizontally to the right.

[0049] Convert pixel distance to physical distance;

[0050] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0051] S43: When c < -ε, the mobile platform moves c pixels horizontally to the left.

[0052] Convert pixel distance to physical distance;

[0053] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0054] S44: When |c|<ε, it means that the horizontal fine positioning is completed, and jump to S45 to complete the vertical fine positioning;

[0055] S45: Calculate the difference between the current vertical position of the left guide tube to be docked with the reduction tank and the longitudinal coordinate of its calibration point:

[0056] d=Y' A -Y A ;

[0057] Among them, Y' A Y is the ordinate of the current position of the left guide tube; A The ordinate of the calibration point of the left guide tube;

[0058] S46: When d > ε, the mobile platform moves d pixels downwards in the vertical direction.

[0059] Convert pixel distance to physical distance;

[0060] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0061] S47: When d < -ε, the mobile platform moves d pixels vertically upwards;

[0062] Convert pixel distance to physical distance;

[0063] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0064] S48: When |d|<ε, it means that the vertical precision positioning is complete and the algorithm ends.

[0065] The specific steps of the fine-tuning algorithm are as follows: The fine-tuning algorithm includes simultaneous fine-tuning of the left guide cylinder and the right guide cylinder. The principle of fine-tuning the left and right guide cylinders is the same. The fine-tuning process of the left guide cylinder is as follows:

[0066] S51: Use an object detection algorithm to identify the restoration container in the current image and obtain the width W of the detection rectangle for the restoration container. B and height H B ;

[0067] S52: Calculate the x-coordinate of the center point of the current reduction vessel after fine-tuning:

[0068] X = (F w / 2-F w / 2*W B / W A )+X A *(W B / W A ), where F w W is the image pixel width. A Indicates the height of the calibration rectangle;

[0069] S53: Calculate the x-coordinate Y of the center point of the current reduction vessel after fine-tuning.

[0070] Y = (F h / 2-F h / 2*H B / H A )+Y A *(H B / H A ), where F h H is the image pixel height. A This indicates the width of the calibration rectangle;

[0071] S54: Based on the finely adjusted coordinates (X, Y), the fine positioning algorithm is invoked to finally complete the docking of the guide cylinder and the reduction tank.

[0072] The beneficial effects of this invention compared to the prior art are as follows: This invention provides an automatic docking system and method for a reduction tank based on machine vision. By calibrating the guide cylinder and the reduction tank, and then finely adjusting the position of the guide cylinder according to coarse positioning and fine positioning after calibration, the automatic docking of the guide cylinder and the reduction tank is realized. Attached Figure Description

[0073] The present invention will be further described below with reference to the accompanying drawings:

[0074] Fig. 1 This is a schematic diagram of the system structure of the present invention;

[0075] Fig. 2 This is a flowchart of the method of the present invention;

[0076] Fig. 3 This is a schematic diagram of the calibration of the reduction vessel of the present invention;

[0077] Fig. 4 This is a coordinate diagram of the fine positioning algorithm of the present invention. Detailed Implementation

[0078] like Figs. 1 to 4As shown, the automatic docking system for restoration tanks based on machine vision provided by the present invention includes a coarse positioning subsystem, a fine positioning subsystem, and a PLC control subsystem.

[0079] The coarse positioning system mainly consists of RFID electronic tags, card readers, a mobile trolley, and coarse positioning software. The mobile trolley is installed on a workshop track and moves along it. The system primarily controls the position of the mobile trolley, ensuring it stops near the reduction tanks to be loaded. To achieve this, a numbered, high-temperature resistant RFID electronic tag is first installed above each reduction tank. Then, a card reader for the electronic tag is installed at a suitable location on the mobile trolley. During the trolley's movement, the card reader reads the electronic tag information in real time. If the reading meets the requirements, the coarse positioning software sends a command to the PLC control system, instructing the mobile trolley to stop. Otherwise, the mobile trolley continues moving until it reads an electronic tag that meets the requirements.

[0080] The precision positioning subsystem of this invention consists of a camera, a guide cylinder, a moving platform, and precision positioning software. The camera is mounted above and behind the guide cylinder to capture video images of the reduction tank. The guide cylinder, primarily used for loading after docking with the reduction tank, is integrated with the moving platform to ensure a fixed relative position. The moving platform can move within a certain range in both horizontal and vertical directions, driving the guide cylinder to move in the same direction. The main function of the precision positioning subsystem is to achieve precise docking between the guide cylinder and the reduction tank using machine vision technology after coarse positioning of the reduction tank, ensuring that the guide cylinder can smoothly enter the reduction tank.

[0081] To achieve the above objectives, this invention also proposes an automatic docking method for restoration tanks based on machine vision, the main steps of which are as follows:

[0082] (1) Calibration of the center point of the restoration tank: Calibration is to obtain the coordinates of the center point of the restoration tank in the image coordinate system when the guide tube enters the restoration tank to be calibrated. Specifically, during calibration, a restoration tank is selected, and the guide tube is manually guided into the restoration tank. The position of the guide tube at this time is calibrated to obtain the calibration point of the guide tube. Then, the restoration tank in the current image is detected and identified using deep learning-based object detection technology. If the identification is successful, the center point coordinates of the restoration tank can be calculated through the detection box. In order to further improve the accuracy of calibration and avoid interference from other factors such as camera shake, the median of the restoration tank detection and identification results of n (n=50) restoration tanks is calculated to obtain the final calibration coordinates. As shown in Figure 3, C(x,y) is the coordinate of the center point of the calibrated restoration tank.

[0083] (2) Fine-tuning of the moving trolley: Since the horizontal and vertical ranges of the moving platform are within a certain range, after coarse positioning, the moving trolley may stop at a position that causes the horizontal position of the current reduction tank to be more than the range of the calibration point due to inertia and other reasons. This will inevitably prevent the guide cylinder from entering the reduction tank. Therefore, in order to solve the above problem, an algorithm for fine-tuning the position of the moving trolley is proposed. Before the left and right guide cylinders are simultaneously connected to the reduction tank, the position of the moving trolley is controlled so that the distance between the guide cylinder and its corresponding calibration point is within the range. The specific algorithm flow is as follows.

[0084] Input: The ranges of the two movable platforms are C1 and C2, and it is assumed that C1≤C2;

[0085] X A : X-coordinate of the calibration point of the left guide tube; B : The x-coordinate of the calibration point of the right guide tube;

[0086] X' A : The x-coordinate of the current position of the left guide tube; X' B : The horizontal coordinate of the current position of the right guide tube;

[0087] process:

[0088] ① Calculate the difference between the x-coordinate of the left guide tube calibration point and its current x-coordinate: a = X A -X' A ;

[0089] ② Calculate the difference between the x-coordinate of the right guide tube calibration point and its current x-coordinate: b = X B -X' B ;

[0090] ③ Determine whether the left and right guide cylinders are within the range: a≤C1 and b≤C2;

[0091] ④ If within the range: no adjustment is needed, the algorithm ends;

[0092] ⑤ If it is not within the range, then calculate:

[0093] When |ab|>2C1, it means that the left and right guide tubes cannot enter at the same time and need to enter one tube at a time, and the algorithm ends.

[0094] When |ab| < 2C1, it means that the left and right guide cylinders can advance simultaneously by adjustment. The process of adjusting the moving trolley is as follows:

[0095] When a > 0 and b > 0, the moving cart moves to the left by |a + b| / 2.

[0096] When a < 0 and b < 0, the moving cart moves to the right by |a + b| / 2.

[0097] When a+b>0, the moving cart moves to the left by |a+b| / 2;

[0098] When a+b<0, the moving cart moves to the right by |a+b| / 2;

[0099] ⑥ Send commands to the control subsystem to control the mobile trolley to move to the adjusted position;

[0100] ⑦ After the trolley moves, check whether the positions of the left and right guide cylinders are within the range of C2;

[0101] ⑧ If present, adjustment stops, algorithm ends;

[0102] ⑨ Otherwise, proceed to step (3) to continue adjusting.

[0103] (3) Precision positioning of the guide cylinder: After the fine adjustment is completed, the precision positioning system is started. This system can start two guide cylinders at the same time to dock with their respective reduction tanks. The algorithm process is described below using the left guide cylinder as an example:

[0104] Input: (X) A Y A ): Represents the coordinates of the calibration point of the left guide cylinder; ε: Represents the allowable error value between the coordinates of the center point of the reduction tank detected in the video image and the coordinates of the calibration point;

[0105] process:

[0106] ① Calculate the difference between the current horizontal position of the left guide tube to be docked with the reduction tank and the x-coordinate of its calibration point:

[0107] c = X' A -X A ;

[0108] Where X' A X is the x-coordinate of the current position of the left guide tube. A The x-coordinate of the calibration point of the left guide tube;

[0109] ②When c>ε, the mobile platform moves c pixels horizontally to the right;

[0110] Convert pixel distance to physical distance;

[0111] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0112] ③ When c < -ε, the mobile platform moves c pixels horizontally to the left;

[0113] Convert pixel distance to physical distance;

[0114] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0115] ④ When |c|<ε, it means that the horizontal fine positioning is completed, and jump to S45 to complete the vertical fine positioning;

[0116] ⑤ Calculate the difference between the current vertical position of the left guide tube to be docked with the reduction tank and the ordinate of its calibration point:

[0117] d=Y' A -Y A ;

[0118] Among them, Y' A Y is the ordinate of the current position of the left guide tube; A The ordinate of the calibration point of the left guide tube;

[0119] ⑥ When d > ε, the mobile platform moves d pixels downwards in the vertical direction;

[0120] Convert pixel distance to physical distance;

[0121] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0122] ⑦ When d < -ε, the mobile platform moves d pixels vertically upwards;

[0123] Convert pixel distance to physical distance;

[0124] The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform;

[0125] ⑧ When |d|<ε, it means that the vertical precise positioning is completed and the algorithm ends.

[0126] (4) Fine-tuning system: Due to the inability to guarantee the standard and uniformity of the longitudinal and transverse positions of the reduction tank during installation or production tank replacement, coupled with tank deformation caused by high temperatures during production, these factors pose challenges to continuous automatic docking. In order to ensure "one-time calibration, arbitrary docking", regardless of the front-to-back relative position of the current reduction tank and the calibration reduction tank, or whether the vertical position is biased upwards or downwards, the algorithm has good adaptability. A fine-tuning algorithm is proposed. The specific process of the fine-tuning algorithm for the left guide cylinder (the right guide cylinder is similar) is given below:

[0127] Input: (X) A Y A ): Represents the coordinates of the calibration point of the left guide tube; (W A H A ): Indicates the width and height of the bounding rectangle;

[0128] process:

[0129] ① Use an object detection algorithm to identify the restoration container in the current image and obtain the width W of the detection rectangle for the restoration container. B and height H B ;

[0130] ② Calculate the x-coordinate of the center point of the current reduction vessel after fine-tuning:

[0131] X = (F w / 2-F w / 2*W B / W A )+X A *(W B / W A ), where F w The width of the image in pixels;

[0132] ③ Calculate the x-coordinate Y of the center point of the current reduction vessel after fine-tuning:

[0133] Y = (F h / 2-F h / 2*H B / H A )+Y A *(H B / H A ), where F h Image pixel height;

[0134] ④ Based on the finely adjusted coordinates (X, Y), the fine positioning algorithm is called to finally complete the docking of the guide cylinder and the reduction tank.

[0135] Because the standardization and uniformity of the longitudinal and transverse positions of the reduction tank cannot be guaranteed during installation or production tank replacement, coupled with tank deformation caused by high temperatures during production, the docking system faces challenges. In order to solve the problems existing in the charging method of reduction tanks in magnesium smelting, this invention proposes an automatic docking system and method for horizontal reduction tanks and guide cylinders in magnesium smelting. This system utilizes technologies such as machine vision and artificial intelligence, and only requires one calibration. It can then use the left and right guide cylinders to work simultaneously to achieve continuous automatic docking of all reduction tanks in the workshop. It can also better adapt to the creep of various deformed reduction tanks, reduce the labor intensity of furnace workers and achieve staff reduction, improve labor productivity and create greater profits.

[0136] Regarding the specific structure of this invention, it should be noted that the connection relationships between the various component modules used in this invention are definite and achievable. Except as specifically described in the embodiments, their specific connection relationships can bring about corresponding technical effects and solve the technical problems proposed by this invention without relying on the execution of corresponding software programs. The models of the components, modules, and specific components appearing in this invention, the connection methods between them, and the conventional usage methods and expected technical effects brought about by the above technical features, unless specifically described, are all publicly disclosed content in patents, journal articles, technical manuals, technical dictionaries, and textbooks that can be obtained by those skilled in the art before the application date, or belong to conventional technology, common knowledge, and other existing technologies in this field. There is no need to elaborate, which makes the technical solution provided in this case clear, complete, and achievable, and can reproduce or obtain corresponding physical products based on this technical means.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A machine vision-based automatic docking system for reduction tanks, characterized in that: It includes a coarse positioning subsystem, a fine positioning subsystem, and a control subsystem. The coarse positioning subsystem is used to achieve coarse positioning of the reduction tank to be filled. The fine positioning subsystem achieves precise docking between the guide cylinder and the reduction tank to be filled through machine vision. The control subsystem is used to control the electrical components in the coarse positioning subsystem and the fine positioning subsystem. The fine positioning subsystem integrates a fine adjustment algorithm, a fine positioning algorithm, and a fine adjustment algorithm. The fine adjustment algorithm is used to control the position of the moving trolley after the coarse positioning of the reduction tank is completed and before the left and right guide cylinders are simultaneously connected to the reduction tank, so that the distance between the guide cylinders and the corresponding calibration points is within the range. The precise positioning algorithm is used to simultaneously activate two guide cylinders to dock with their respective reduction tanks; The fine-tuning algorithm is used to achieve the final docking of the guide cylinder and the reduction tank; The coarse positioning subsystem includes RFID electronic tags, card readers, mobile trolleys, and coarse positioning software. The mobile trolleys are installed on the tracks in the workshop and move along the tracks. Each reduction tank is equipped with a numbered, high-temperature resistant RFID electronic tag. The mobile trolleys are equipped with card readers. The coarse positioning software sends commands to the control subsystem to control the movement and stopping of the mobile trolleys based on the information of the RFID electronic tags read by the card readers, so that the mobile trolleys stop near the reduction tanks to be loaded. The precision positioning subsystem includes a camera, a guide cylinder, a moving platform, and precision positioning software. The camera is mounted on the guide cylinder to capture video images of the reduction tank. The guide cylinder is mounted on the moving platform for loading materials after the reduction tank is docked. The moving platform can move within a set range in the horizontal and vertical directions and drive the guide cylinder to move in the same direction. The precision positioning software has built-in computer programs for fine-tuning, precision positioning, and fine-tuning algorithms. The fine-tuning algorithm keeps the position of the moving trolley within the calibrated range, the precision positioning algorithm accurately positions the guide cylinder with the corresponding reduction tank, and the fine-tuning algorithm enables continuous automatic docking between the guide cylinder and the reduction tank.

2. The automatic docking system for reduction tanks based on machine vision according to claim 1, characterized in that: The camera is specifically installed at the rear and top of the guide tube.

3. A machine vision-based automatic docking method for restoration tanks, employing the machine vision-based automatic docking system for restoration tanks as described in claim 1 or 2, characterized in that: Includes the following steps: S1: Coarse positioning of the guide tube: During the movement of the mobile trolley, the information of the RFID electronic tag installed on the reduction tank is read in real time using a card reader. If the reading result meets the requirements, the coarse positioning software sends a command to the control subsystem to require the mobile trolley to stop moving. Otherwise, the mobile trolley continues to move until it reads an RFID electronic tag that meets the conditions and finds the position of the reduction tank to be loaded. S2: Center point calibration of the reduction tank to be loaded: Select the reduction tank with the RFID electronic tag that meets the conditions installed in step S1, and then guide the guide cylinder into the reduction tank manually. Calibrate the coordinates of the guide cylinder at this time. Then, use deep learning-based target detection technology to detect and identify the reduction tank in the current image. If the identification is successful, calculate the center point coordinates of the reduction tank through the detection box, and obtain the final calibration center point coordinates of the reduction tank by calculating the median of the nth reduction tank detection and identification structure. S3: Fine-tuning the moving trolley through a fine-tuning algorithm: Before the left and right guide cylinders are simultaneously connected to the reduction tank, the position of the moving trolley is controlled by the control subsystem so that the distance between the guide cylinders and the corresponding calibration points is within the range. S4: Precise positioning of the guide cylinder is achieved through a precision positioning algorithm: the movement of the mobile platform is controlled based on the allowable error between the coordinates of the center point of the restoration tank detected in the video image and the coordinates of the calibration point of the guide cylinder, thereby completing the precise positioning of the guide cylinder and the restoration tank. S5: The guide cylinder position is precisely adjusted through a fine-tuning algorithm to achieve docking with the restoration tank: The target detection algorithm is used to identify the restoration tank in the current image, calculate the coordinates of the center point of the current restoration tank after fine-tuning, and call the fine positioning algorithm based on the coordinates of the center point of the restoration tank after fine-tuning to finally complete the docking of the guide cylinder and the restoration tank.

4. The automatic docking method for restoration tanks based on machine vision according to claim 3, characterized in that: The specific steps of the fine-tuning algorithm are as follows: Let the ranges of the two moving platforms mounted on the two guide cylinders be C1 and C2, respectively, and assume that C1 ≤ C2. Let the x-coordinate of the calibration point on the left guide cylinder be X. A The horizontal coordinate of the current position of the left guide tube is X' A The x-coordinate of the calibration point of the right guide tube is X. B The x-coordinate of the current position of the right guide tube is X' B ; The calculation process of the fine-tuning algorithm is as follows: S31: Calculate the difference between the x-coordinate of the left guide tube calibration point and its current x-coordinate: a = X A -X' A ; S32: Calculate the difference between the x-coordinate of the right guide tube calibration point and its current x-coordinate: b=X B -X' B ; S33: Determine whether the left and right guide cylinders are within the range: a≤C1 and b≤C2; S34: If within the range: no adjustment is needed, the algorithm ends; S35: If not within the range, then calculate: When |ab|>2C1, it means that the left and right guide tubes cannot enter at the same time and need to enter one tube at a time, and the algorithm ends. When |ab| < 2C1, it means that the left and right guide cylinders can advance simultaneously by adjustment. The process of adjusting the moving trolley is as follows: When a > 0 and b > 0, the moving cart moves to the left by |a + b| / 2. When a < 0 and b < 0, the moving cart moves to the right by |a + b| / 2. When a+b>0, the moving cart moves to the left by |a+b| / 2; When a+b<0, the moving cart moves to the right by |a+b| / 2; S36: Send a command to the control subsystem to control the mobile trolley to move to the adjusted position; S37: After the trolley moves, check whether the positions of the left and right guide cylinders are within the range of C2; S38: If present, adjustment stops, algorithm ends; S39: Otherwise, proceed to step S4 to continue the adjustment.

5. The automatic docking method for restoration tanks based on machine vision according to claim 4, characterized in that: The specific steps of the precise localization algorithm are as follows: The precision positioning algorithm includes simultaneous precision positioning of the left guide cylinder and the right guide cylinder. The principle of precision positioning of the left and right guide cylinders is the same. The adjustment process for precision positioning of the left guide cylinder is as follows: Let the coordinates of the calibration point of the left guide tube be (X... A Y A ), ε represents the allowable error value between the coordinates of the center point of the restoration tank and the coordinates of the calibration point detected in the video image; S41: Calculate the difference between the current horizontal position of the left guide tube to be docked with the reduction tank and the x-coordinate of its calibration point: c=X’ A -X A ; Where X' A X is the x-coordinate of the current position of the left guide tube. A The x-coordinate of the calibration point of the left guide tube; S42: When c > ε, the mobile platform moves c pixels horizontally to the right. Convert pixel distance to physical distance; The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform; S43: When c < -ε, the mobile platform moves c pixels horizontally to the left. Convert pixel distance to physical distance; The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform; S44: When |c|<ε, it means that the horizontal fine positioning is completed, and jump to S45 to complete the vertical fine positioning; S45: Calculate the difference between the current vertical position of the left guide tube to be docked with the reduction tank and the longitudinal coordinate of its calibration point: d=Y’ A -Y A ; Among them, Y' A Y is the ordinate of the current position of the left guide tube; A The ordinate of the calibration point of the left guide tube; S46: When d > ε, the mobile platform moves d pixels downwards in the vertical direction. Convert pixel distance to physical distance; The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform; S47: When d < -ε, the mobile platform moves d pixels vertically upwards; Convert pixel distance to physical distance; The converted physical distance is sent to the control subsystem to complete the movement of the mobile platform; S48: When |d|<ε, it means that the vertical precision positioning is complete and the algorithm ends.

6. The automatic docking method for restoration tanks based on machine vision according to claim 5, characterized in that: The specific steps of the fine-tuning algorithm are as follows: The fine-tuning algorithm includes simultaneous fine-tuning of the left guide cylinder and the right guide cylinder. The principle of fine-tuning the left and right guide cylinders is the same. The fine-tuning process of the left guide cylinder is as follows: S51: Use an object detection algorithm to identify the restoration container in the current image and obtain the width W of the detection rectangle for the restoration container. B and height H B ; S52: Calculate the x-coordinate of the center point of the current reduction vessel after fine-tuning: X=(F w / 2-F w / 2 W B / W A )+X A (W B / W A ), where F w W is the image pixel width. A Indicates the height of the calibration rectangle; S53: Calculate the x-coordinate Y of the center point of the current reduction vessel after fine-tuning. Y=(F h / 2-F h / 2 H B / H A )+Y A (H B / H A ), where F h H is the image pixel height. A This indicates the width of the calibration rectangle; S54: Based on the finely adjusted coordinates (X, Y), the fine positioning algorithm is invoked to finally complete the docking of the guide cylinder and the reduction tank.

Images