Automatic transformer disassembling method and system

Abstract

The invention belongs to the technical field of transformer automatic disassembly, and provides a transformer automatic disassembly method and system.The method comprises the steps that candidate positions of bolts are determined in a preset list including transformer types through segmented images, and bolt position detection is carried out in an RGB image through a preset bolt position detection model; comparing the detected bolt position with a candidate position, correcting the bolt position, and determining a three-dimensional coordinate of the bolt in a preset three-dimensional model by using the corrected bolt position; according to the method, the bolt position of the transformer of the corresponding model is directly called from the preset list, and the bolt position is corrected by comparing the detected bolt position with the candidate position, so that on the basis of ensuring the accuracy of the bolt position, the acquisition and updating of excessive data in the processes of judgment, correction and the like are avoided, and the accuracy of the bolt position is improved. And the influence on the control speed and efficiency is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of automatic transformer disassembly, and in particular relates to a transformer automatic disassembly method and system. Background Art

[0002] The construction of an intelligent retired transformer dismantling production line is of great significance to meeting the growing demand for efficient management of waste material dismantling, promoting transformer dismantling towards information-based, professional and green management, and opening up the last mile of waste material recycling management.

[0003] The traditional intelligent dismantling production line for retired transformers is generally a semi-automatic dismantling production line with human-machine collaboration, which cannot realize the full-process automatic dismantling of transformers. One of the reasons for limiting the full-process automatic dismantling of transformers is that the operating position of the transformer needs to be accurately positioned at different dismantling stages, and the requirements for data updating, processing and control are high. For example, when the bolts are automatically dismantled, the position of the bolts to be dismantled each time will change due to different transformer models and manufacturers. Even for transformers of the same model and manufacturer, the position of the bolts will change due to machining accuracy and deformation during later use. If the position of the bolts cannot be well determined, the bolts will not cooperate with the wrench on the robotic arm when the robotic arm is used for operation, which will cause damage to the force position of the wrench, affect its service life, and increase the difficulty of disassembly. The bolts need to be re-positioned each time to avoid the problem of inaccurate positioning, but each data update and the generation of processing control instructions have high requirements for the control system, which will affect the control speed and efficiency and increase the difficulty of data processing and control. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a transformer automatic disassembly method and system. The present invention directly retrieves the bolt position of the transformer of the corresponding model in a preset list, and corrects the bolt position by comparing the detected bolt position with the candidate position. On the basis of ensuring the accuracy of the bolt position, the acquisition and update of excessive data used in the judgment and correction process are avoided, thereby reducing the impact on the control speed and efficiency.

[0005] In order to achieve the above object, the present invention is implemented through the following technical solutions:

[0006] In a first aspect, the present invention provides a transformer automatic disassembly method, comprising:

[0007] Get the RGB image and depth image of the transformer to form a point cloud;

[0008] The upper surface of the transformer is segmented using the RGB image to obtain a segmented image. The model used for the upper surface segmentation of the transformer is YOLACT (You Only Look At Coefficients).

[0009] Using the segmented image, remove other parts from the scene point cloud, retain only the point cloud corresponding to the upper surface of the transformer, and match it with the pre-formed upper surface template of the transformer to obtain the 6D pose (position X, Y, Z and posture RotX, RotY, RotZ) of the transformer; wherein the transformer upper surface point cloud template matching model is PPF (Point Pair Features);

[0010] Using the 6D pose of the transformer upper surface, determine the candidate positions of the bolts in a preset list; wherein the information in the preset list includes the transformer type and the bolt positions corresponding to the transformer type;

[0011] In the RGB image, the bolt position is detected using a preset bolt position detection model, the detected bolt position is compared with the candidate position, and the bolt position is corrected; wherein the bolt position detection model is WBF (Weighted Box Fusion);

[0012] Using the corrected bolt position, determine the three-dimensional coordinates of the bolt in the preset three-dimensional model;

[0013] The three-dimensional coordinates of the bolts are converted into the coordinates of the preset robot end tool to remove the bolts and realize the disassembly of the transformer cover.

[0014] Furthermore, the upper surface of the transformer is segmented using the YOLACT neural network and only the point cloud of the upper surface of the transformer is retained. Specifically, first, the RGB information and the depth information are converted into point cloud information; then, the upper surface of the transformer is segmented using the YOLACT neural network; then, preprocessing is used to retain only the point cloud of the upper surface of the transformer; wherein the preprocessing includes combining the segmentation results, the connected component method and noise removal; the preprocessing improves the stability of subsequent point cloud matching and speeds up the matching speed.

[0015] Furthermore, the 6D pose (X, Y, Z, RotX, RotY, RotZ) of the transformer is obtained by matching the retained point cloud of the upper surface of the transformer with the pre-formed upper surface model of the transformer, wherein the point cloud template matching model is PPF (Point Pair Features); the obtained 6D pose of the transformer is used to correct the segmentation result of the upper surface of the transformer in the two-dimensional RGB image.

[0016] Furthermore, the bolt positions are detected in the upper surface area of ​​the transformer in the RGB image. Specifically, the upper surface area of ​​the transformer is cut out, and the bolt positions are detected using a WBF (Weighted Box Fusion) neural network.

[0017] Furthermore, when determining the candidate position of the bolt, the bolt position of the corresponding model transformer is directly retrieved from the preset list, and the detected bolt position is compared with the candidate position. Specifically, the candidate bolt position of the corresponding model transformer is first retrieved from the preset list and matched with the previously detected bolt position. For a bolt that is not detected due to external interference, the candidate position is regarded as the bolt position. For a position where the distance between the candidate position and the previously detected position is greater than a set threshold, it is regarded as a false detection. For a bolt where the distance between the candidate position and the previously detected position is less than the threshold, the detected position is regarded as the bolt position.

[0018] Furthermore, after the transformer cover is disassembled, the magnetic suction point cloud is matched in the three-dimensional transformer upper surface point cloud through the transformer 6D posture and the transformer upper surface model to obtain the three-dimensional coordinates of the midpoint of the magnetic suction position; the three-dimensional coordinates of the magnetic suction position are converted into the tool coordinates of the electromagnetic tooling, and the electromagnetic tooling is guided to reach the magnetic suction position, automatically suck the transformer cover, and lift the transformer cover;

[0019] The preset oil pumping robot automatically inserts the oil pumping tool at the end into the transformer housing and starts pumping oil, and stops automatically when the oil pumping is completed;

[0020] The electromagnetic tooling transports the transformer cover to the coil disassembly station, and the mechanical arms preset at both ends automatically use hydraulic shears to separate the connectors between the transformer cover and the winding;

[0021] The RGB image of the transformer is segmented. With the help of the RGB image segmentation results, the three-dimensional coordinates of the position of the yoke and coil on the winding are determined from the pre-built transformer type and winding model; the three-dimensional coordinates are converted into the preset robot end tool coordinates, and the robot end clamping tool is guided to clamp the upper yoke for disassembly;

[0022] Carry out the storage of loose parts, shells and materials.

[0023] In a second aspect, the present invention further provides a transformer automatic disassembly system, comprising:

[0024] The image acquisition module is configured to: obtain an RGB image and a depth image of the transformer;

[0025] The image segmentation module is configured to: segment the upper surface of the transformer using the RGB image and the depth image, obtain the segmented image, and perform 6D pose matching with the upper surface point cloud template of the transformer;

[0026] The candidate position determination module is configured to: determine the candidate position of the bolt in a preset list using the image segmented by the matching result image; wherein the information in the preset list includes the transformer type and the bolt position corresponding to the transformer type;

[0027] The bolt position detection module is configured to: detect the bolt position in the RGB image using a preset bolt position detection model, compare the detected bolt position with the candidate position, and correct the bolt position; wherein the bolt position detection model is WBF (Weighted Box Fusion);

[0028] The correction module is configured to: determine the three-dimensional coordinates of the bolt in a preset three-dimensional model using the corrected bolt position;

[0029] The disassembly module is configured to: convert the three-dimensional coordinates of the bolt into the coordinates of a preset robot end tool to remove the bolt, thereby realizing the disassembly of the transformer cover.

[0030] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the transformer automatic disassembly method described in the first aspect.

[0031] In a fourth aspect, the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein when the processor executes the program, the steps of the transformer automatic disassembly method described in the first aspect are implemented.

[0032] In a fifth aspect, the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of the transformer automatic disassembly method described in the first aspect are implemented.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] The present invention utilizes the image segmented by YOLACT neural network instance segmentation and 6D pose matching with the target matching on the transformer to form a good segmentation, determines the candidate position of the bolt in a preset list including the transformer type, and uses a preset bolt position detection model to detect the bolt position in the RGB image, compares the detected bolt position with the candidate position, corrects the bolt position, and determines the three-dimensional coordinates of the bolt in the preset three-dimensional model using the corrected bolt position; directly retrieves the bolt position of the transformer of the corresponding model in the preset list, and corrects the bolt position by comparing the detected bolt position with the candidate position, which, on the basis of ensuring the accuracy of the bolt position, avoids the acquisition and update of too much data used in the judgment and correction processes, and reduces the impact on the control speed and efficiency.

[0035] The present invention is based on the determination of coordinate positions based on images, and realizes the full-process automation of upper cover disassembly, shell separation, oil extraction, connecting piece shearing, iron yoke splitting, coil splitting, warehousing and outbound delivery, thereby realizing the full-process automatic disassembly of the transformer and improving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings in the specification that constitute a part of this embodiment are used to provide a further understanding of this embodiment. The schematic embodiments of this embodiment and their descriptions are used to explain this embodiment and do not constitute improper limitations on this embodiment.

[0037] Figure 1 This is an overall flow chart of transformer disassembly according to Embodiment 1 of the present invention;

[0038] Figure 2 This is an overall layout diagram of a transformer disassembly line according to Embodiment 1 of the present invention;

[0039] Figure 3 This is a schematic diagram of an AGV forklift and an RGV trolley according to Embodiment 1 of the present invention;

[0040] Figure 4 This is a schematic diagram of a disassembly robot according to Embodiment 1 of the present invention;

[0041] Figure 5 It is a schematic diagram of the example segmentation neural network YOLACT according to Embodiment 1 of the present invention;

[0042] Figure 6 This is a schematic diagram of the electromagnetic tooling of Example 1 of the present invention;

[0043] Figure 7 This is a schematic diagram of an oil pumping robot according to Embodiment 1 of the present invention;

[0044] Figure 8 This is a schematic diagram of a coil disassembling device according to Embodiment 1 of the present invention;

[0045] Fig. 9 This is a schematic diagram of a three-dimensional warehouse according to Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0048] Embodiment 1:

[0049] The present embodiment provides a method for automatic disassembly of a transformer, comprising: acquiring an RGB image and a depth image of the transformer; segmenting the upper surface of the transformer using the RGB image and the depth image to obtain a segmented image; determining candidate positions of bolts in a preset list using the segmented image; wherein the information in the preset list includes the transformer type and the bolt positions corresponding to the transformer type; in the RGB image, detecting the bolt position using a preset bolt position detection model, comparing the detected bolt position with the candidate position, and correcting the bolt position; determining the three-dimensional coordinates of the bolt in a preset three-dimensional model using the corrected bolt position; and converting the three-dimensional coordinates of the bolt into the coordinates of a preset robot end tool to perform bolt disassembly, thereby realizing disassembly of the transformer upper cover.

[0050] Specifically, the bolt position of the corresponding transformer model is directly retrieved from the preset list, and the bolt position is corrected by comparing the detected bolt position with the candidate position. On the basis of ensuring the accuracy of the bolt position, the acquisition and update of too much data in the judgment and correction process is avoided, reducing the impact on control speed and efficiency. Figure 1 and Figure 2 As shown, the steps of the method in this embodiment may include:

[0051] S1, RGV transfer:

[0052] Optionally, the AGV forklift places the retired transformer that needs to be disassembled on the RGV transfer vehicle. After the RGV transfer vehicle detects that the transformer is in place, it transports the transformer to the upper cover disassembly station.

[0053] like Figure 3As shown, the AGV forklift and RGV transfer vehicle can be products currently on the market. The RGV transfer vehicle is provided with a material tray, an image sensor and / or a weight device for carrying the transformer; when the RGV transfer vehicle detects whether the transformer is in place, it can use image recognition technology or weight recognition technology. For example, when the preset image sensor recognizes that there is a transformer on the RGV transfer vehicle, or the weight value recognized by the weighing device preset on the RGV transfer vehicle reaches a preset value, it is determined that the transformer is placed on the RGV transfer vehicle.

[0054] After the RGV transfer vehicle detects that the transformer is in place, it can move the transformer to the upper cover disassembly station and other disassembly stations behind it according to a preset path or a guide line set on the road surface.

[0055] S2. Remove the upper cover:

[0056] Optional, such as Figure 4 As shown, after the RGV transporter transports the material tray carrying the transformer to the upper cover disassembly station, the preset disassembly robot is used to remove the screws of the transformer upper cover with the pneumatic tool at the end under visual guidance. The disassembly robot can be implemented by a multi-axis robotic arm or other structure that can move in any direction, and the pneumatic tool can be a pneumatic wrench or other tool that can remove bolts, and the pneumatic tool is set at the end of the disassembly robot by bolt connection or clamping.

[0057] S2.1. After receiving the instruction to start removing the bolts, the RGB image and depth image, i.e., the 3D point cloud, are obtained as input through the 3D RGB-D camera pre-installed on the upper part of the transformer.

[0058] The distance value from the RGB-D camera to each point on the transformer is used as a pixel value to obtain the depth image, which reflects the geometric shape of the visible surface of the transformer; the depth image can be calculated as point cloud data after coordinate transformation.

[0059] S2.2. Use the pre-trained transformer upper surface segmentation model to segment the upper surface of the transformer in the RGB image.

[0060] Optionally, the transformer surface segmentation model is YOLACT neural network instance segmentation; Optionally, such as Figure 5 As shown, the YOLACT neural network instance segmentation model is used to segment the upper surface of the transformer in the RGB image. The point cloud other than the segmented upper surface of the transformer is removed through the correspondence between the RGB image and the point cloud.

[0061] S2.3. By using connected component labeling algorithm and noise removal methods, only the point cloud of the upper surface of the transformer is retained.

[0062] S2.4. Estimate the 6D pose of the transformer by matching it with the pre-formed transformer upper surface model.

[0063] Optionally, the matching process includes the following three steps to improve performance and speed up: Approximate matching, Sparse pose refinement, and Dense pose refinement. In the approximate matching step, the approximate pose of the transformer upper surface model instance in the scene is searched; the point cloud is sampled and a subset of key points is defined; for each key point, the best pose of the transformer upper surface model is calculated under the assumption that the key point is on the surface of the object; then the pose with the highest score is selected from all key points and used as the approximate pose. In the sparse pose refinement step, the approximate pose found in the previous step is refined by minimizing the distance between the sampled scene points and the plane of the transformer upper surface model points; this increases the accuracy of the pose and the importance of the score. In the dense pose refinement step, the pose found in the previous step is accurately refined; this step is similar to the sparse pose refinement, minimizing the distance between all scene points and the nearest plane of the transformer upper surface model points. .

[0064] S2.5. Using the obtained 6D pose of the transformer, correct the segmentation result of the transformer upper surface in the two-dimensional RGB image.

[0065] Optionally, the transformer upper surface segmentation results are corrected in the two-dimensional RGB image using the correspondence between the RGB map and the point cloud; this can avoid segmentation errors previously made by the YOLACT instance segmentation neural network, thereby ensuring the accuracy of the transformer upper surface segmentation results.

[0066] S2.6. Obtain bolt location candidates from a pre-built list of transformer types and bolt locations.

[0067] Optionally, the type of the transformer is known through the transformer upper surface identification stage above, and bolt position candidates are obtained from a pre-built transformer type and bolt position list. This candidate position will be used in the subsequent bolt position detection stage to correct false detections and missed detections and adjust the detection position.

[0068] S2.7. In the two-dimensional RGB image, the bolt position detection model is used to detect the bolt position, and the false detection and missed detection are corrected by comparing with the candidate position.

[0069] Optionally, bolt detection is performed using a WBF (Weighted Box Fusion) model; WBF is a technique that combines models with different levels of bounding box prediction accuracy; By assigning weights to each model variant within the framework, WBF calculates the proposed bounding box coordinates using the weighted sum derived from the predictions of the model variants; Unlike NMS, which selects only one box from a set of bounding boxes, WBF fuses all boxes based on an IoU threshold; Specifically, WBF ranks predictions from different model variants according to weights and iteratively fuses bounding boxes from higher-ranked models with bounding boxes from lower-ranked models in a continuous update process; predictions that do not overlap sufficiently with the currently running prediction IoU are discarded; This method is particularly suitable for detecting small objects because predicting the exact bounding box location is challenging because it uses information from each model variant. In addition, false detections and missed detections are corrected by comparison with candidate positions; if the distance between the detected position and the candidate position exceeds a certain threshold, it is determined to be a false detection and the detection result is removed; if there is no detection result within the range of a candidate position, it is determined to be a missed detection and the candidate position is regarded as a detection result; the candidate position of each bolt is adjusted to the detection position.

[0070] S2.8. Using the three-dimensional bolt point cloud model constructed in the preparation stage, the bolt point cloud is matched in the three-dimensional transformer upper surface point cloud, and finally the three-dimensional coordinates of all bolts are output.

[0071] S2.9. Convert the three-dimensional coordinates of the bolt into the coordinates of the robot end tool, and guide the robot end tool to reach the bolt position.

[0072] S2.10, the robot automatically starts a pneumatic wrench or other tool, clamps the bolt head and rotates it to remove the bolt, and then moves to the next bolt position to perform the clamping and rotation actions.

[0073] S2.11. After all bolts / screws are removed, the robot returns to the standby position and waits for the next operation task.

[0074] S3, separation shell:

[0075] S3.1, if Figure 6 As shown, after the screws are removed, the hanging electromagnetic fixture on the top fixes the upper cover of the transformer, lifts it upward, and separates the upper cover from the shell; optionally, the hanging electromagnetic fixture may include a lifting device fixed above the factory building, and an electromagnetic device arranged at the lower end of the lifting device; the electromagnetic device can switch between the presence or absence of magnetic force by powering on or off.

[0076] S3.2, after receiving the task start instruction, the 6D position of the transformer given in step S2 and the transformer upper surface model constructed in the preparation stage are used to match the magnetic attraction point cloud in the three-dimensional transformer upper surface point cloud, and finally the three-dimensional coordinates of the midpoint of the magnetic attraction position are output. The transformer upper surface model can be understood as a 3D structural model constructed using a three-dimensional tool.

[0077] S3.4, converting the three-dimensional coordinates of the magnetic suction position into the tool coordinates of the electromagnetic tooling 5, and guiding the electromagnetic tooling to reach the magnetic suction position.

[0078] S3.5. When the magnetic suction tooling is powered on, it generates a strong magnet, which automatically attracts the upper cover of the transformer, and the electromagnetic tooling rises to lift the upper cover of the transformer.

[0079] Waiting for the next operation instruction.

[0080] S4, automatic oil pumping:

[0081] like Figure 7 As shown, a preset oil pumping robot is used to automatically insert the oil pumping tool at the end into the transformer housing, start pumping oil, and stop automatically after the oil pumping is completed; optionally, the oil pumping robot may include a multi-degree-of-freedom robotic arm, and oil pumping tooling such as an oil pumping pipeline arranged at the end of the robotic arm, and the oil pumping pipeline is connected to power equipment such as an oil pump.

[0082] S4.1. After receiving the task start command, the pumping robot lifts the pumping tool and extends the pumping tool into the transformer housing along a pre-set trajectory.

[0083] S4.2. After the action is completed, the oil pump is automatically started to pump oil.

[0084] S4.3. When the oil level drops to the lowest pumping level, the oil pump automatically stops pumping oil and simultaneously feeds back a pumping end signal.

[0085] S4.4. The pumping robot lifts the pumping tool and pulls the pumping tool out of the transformer housing along a pre-set trajectory.

[0086] Waiting for the next operation instruction.

[0087] S5, shear connector:

[0088] The electromagnetic tooling transports the upper cover of the transformer to the coil disassembly device. Optionally, the coil disassembly device includes two multi-axis robotic arms and equipment such as hydraulic shears at both ends of the robotic arms. The electromagnetic tooling can be set up in the factory through movable equipment such as overhead cranes; the robotic arms at both ends automatically use hydraulic shears to shear and separate the connectors between the upper cover of the transformer and the winding, and the electromagnetic tooling transports the upper cover of the transformer back to the transformer casing.

[0089] S5.1, Cutting connection process:

[0090] S5.1.1. The RGB image and depth image, i.e., the 3D point cloud, are obtained as input by installing a 3D RGBD camera at both ends of the coil disassembly device.

[0091] S5.1.2, by comparing with the pre-formed transformer winding model, the 6D posture of the transformer winding is estimated, which can be achieved by the same method as in step S1.

[0092] S5.1.3, using the obtained 6D posture of the transformer winding, segment the side view of the transformer winding in the two-dimensional RGB image. This can be achieved by using the same method as in step S1.

[0093] S5.1.4, combining the segmentation results in the two-dimensional RGB graphics, and obtaining the three-dimensional coordinates of the screw position from the pre-built transformer type and winding connection screw position list. This can be achieved by the same method as in step S1.

[0094] S5.1.5. In the two-dimensional RGB image, the position of the wire contour is detected using the semantic segmentation method to obtain the three-dimensional coordinates of the wire position at the predetermined cutting position.

[0095] S5.1.6. Convert the three-dimensional coordinates of the screw and the wire into the coordinates of the robot end tool, and guide the robot end tool to the corresponding position.

[0096] S5.1.7. The robot automatically starts the hydraulic shears to cut the screw and wire.

[0097] S5.1.8. After all screws and wires on both sides are cut, the robot returns to the standby position and waits for the next operation task.

[0098] S5.2, Transportation process:

[0099] S5.2.1, continue the action in step 3, the electromagnetic tooling still acts on the upper cover of the transformer. After receiving the task of transferring the upper cover, the electromagnetic tooling carries the upper cover of the transformer along the pre-planned trajectory and returns to the top of the transformer shell.

[0100] S5.2.2. The electromagnetic tooling is lowered to place the upper cover of the transformer on the transformer housing.

[0101] S5.2.3. The electromagnetic tooling is powered off, the magnetism disappears, and the electromagnetic tooling rises and separates from the upper cover of the transformer.

[0102] S5.2.4. The electromagnetic tooling returns to the standby position and waits for the next task.

[0103] S6. Split the iron yoke coil:

[0104] like Figure 8 As described, the coil disassembling device clamps the coil horizontally, the upper fork teeth enter the gap between the upper iron yoke and the coil, lift it upward to remove the upper iron yoke, and the mechanical arm puts the iron yoke and the coil into special material boxes respectively.

[0105] S6.1. Using the segmentation result of the two-dimensional RGB image in step S5, determine the three-dimensional coordinates of the iron yoke and coil positions on the winding from the pre-constructed transformer type and winding model.

[0106] S6.2. Convert the above three-dimensional coordinates into the robot end tool coordinates, and guide the robot end clamping tool to clamp the upper iron yoke.

[0107] S6.3. The horizontal fork teeth in the coil disassembling device are adjusted in height according to the gap height between the upper iron yoke and the coil, and then extended into the space between the upper iron yoke and the coil.

[0108] S6.4. The fork teeth of the system coil disassembly device and the end clamping tool of the robot move upward synchronously to remove the winding from the iron yoke on the winding.

[0109] S6.5. The system instructs the robot to hold the upper iron yoke and place it in the dedicated recycling box next to it.

[0110] S6.6. The robot returns to the top of the transformer winding and, according to the three-dimensional coordinates of the coils given in step S6.1, picks up the three coils respectively and places them in a dedicated recycling box nearby.

[0111] S6.7. The robot returns to the standby position and waits for the next operation task.

[0112] S7, Bulk Parts Storage:

[0113] Optional, such as Fig. 9 As shown, after the coil is disassembled, the material box is transported along the conveyor line to the stacker docking position, and the stacker automatically places the material box on the shelf of the stereoscopic warehouse.

[0114] S8, Shell storage:

[0115] Optionally, the disassembled housing and upper cover are transported to an outdoor storage area via an AGV forklift;

[0116] S9. Materials out of warehouse:

[0117] Optionally, when materials are shipped out of the warehouse, the material boxes can be automatically forked by the stacker to the outbound conveyor line to complete automatic shipment.

[0118] Embodiment 2:

[0119] This embodiment provides a transformer automatic disassembly system, including:

[0120] The image acquisition module is configured to: obtain an RGB image and a depth image of the transformer;

[0121] The image segmentation module is configured to: segment the upper surface of the transformer using the RGB image and the depth image to obtain a segmented image;

[0122] The candidate position determination module is configured to: determine the candidate position of the bolt in a preset list using the segmented image; wherein the information in the preset list includes the transformer type and the bolt position corresponding to the transformer type;

[0123] The bolt position detection module is configured to: detect the bolt position in the RGB image using a preset bolt position detection model, compare the detected bolt position with the candidate position, and correct the bolt position; wherein the bolt position detection model is WBF (Weighted Box Fusion);

[0124] The correction module is configured to: determine the three-dimensional coordinates of the bolt in a preset three-dimensional model using the corrected bolt position;

[0125] The disassembly module is configured to: convert the three-dimensional coordinates of the bolt into the coordinates of a preset robot end tool to remove the bolt, thereby realizing the disassembly of the transformer cover.

[0126] The working method of the system is the same as the transformer automatic disassembly method in Example 1, and will not be repeated here.

[0127] Embodiment 3:

[0128] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the transformer automatic disassembly method described in Embodiment 1 are implemented.

[0129] Embodiment 4:

[0130] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the program, the steps of the transformer automatic disassembly method described in Example 1 are implemented.

[0131] Embodiment 5:

[0132] This embodiment provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of the transformer automatic disassembly method described in Example 1 are implemented.

[0133] The above description is only a preferred embodiment of the present embodiment and is not intended to limit the present embodiment. For those skilled in the art, the present embodiment may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present embodiment shall be included in the protection scope of the present embodiment.

Claims

1. A transformer automatic disassembly method, characterized in that: include: Get the RGB image and depth image of the transformer; Use the RGB image and the depth image to segment the upper surface of the transformer and obtain the segmented image; Using the segmented image, determine candidate locations of the bolts in a preset list; wherein the information in the preset list includes the transformer type and the bolt location corresponding to the transformer type; In the RGB image, the bolt position is detected using a preset bolt position detection model, the detected bolt position is compared with the candidate position, and the bolt position is corrected; wherein the bolt position detection model is a weighted box fusion model; Using the corrected bolt position, determine the three-dimensional coordinates of the bolt in the preset three-dimensional model; The three-dimensional coordinates of the bolts are converted into the coordinates of the preset robot end tool to remove the bolts and realize the disassembly of the transformer cover.

2. The automatic transformer disassembly method according to claim 1, characterized in that: The transformer upper surface is segmented using a long short-term memory artificial neural network. Specifically, first, the depth information is converted into three channel information: disparity, surface normal, and height. Then, the context information of the transformer upper surface in different directions is extracted using a recurrent event network, and bidirectional propagation is performed in both directions. At the same time, for the RGB channel information in the transformer upper surface image, the convolution structure in the long short-term memory artificial neural network is used to extract features, and interpolation is used to restore the features of each level to the same resolution and cascade them. Finally, a recurrent event network is used to obtain contextual information.

3. The automatic transformer disassembly method according to claim 1, characterized in that: Through the connected component labeling algorithm and noise removal, only the point cloud of the upper surface of the transformer is retained to obtain the upper surface model of the transformer; by comparing with the pre-formed upper surface model of the transformer, the 6D posture of the transformer is obtained; using the obtained 6D posture of the transformer, the segmentation result of the upper surface of the transformer is corrected in the two-dimensional RGB image.

4. The automatic transformer disassembly method according to claim 1, characterized in that: The RGB information and the depth information are converted into point cloud information to segment the upper surface of the transformer; only the point cloud of the upper surface of the transformer is retained through preprocessing; the 6D pose of the transformer is obtained by matching the retained point cloud of the upper surface of the transformer with a preformed upper surface model of the transformer, wherein the upper surface model of the transformer is a point pair feature model; and the segmentation result of the upper surface of the transformer is corrected in the two-dimensional RGB image using the obtained 6D pose of the transformer.

5. The automatic transformer disassembly method according to claim 1, characterized in that: When determining the candidate positions of the bolts, the bolt positions of the corresponding model transformers are directly retrieved from the preset list, and the detected bolt positions are compared with the candidate positions. Specifically, the candidate positions of the bolts of the corresponding model transformers are first retrieved from the preset list and matched with the previously detected bolt positions. For bolts that are not detected due to external interference, the candidate positions are regarded as the bolt positions. For positions where the distance between the candidate positions and the previously detected positions is greater than a set threshold, the positions are regarded as false detections. For bolts where the distance between the candidate positions and the previously detected positions is less than the threshold, the detected positions are regarded as the bolt positions.

6. The automatic transformer disassembly method according to claim 1, characterized in that: After the transformer cover is disassembled, the magnetic suction point cloud is matched in the three-dimensional transformer upper surface point cloud through the transformer 6D posture and the transformer upper surface model to obtain the three-dimensional coordinates of the midpoint of the magnetic suction position; the three-dimensional coordinates of the magnetic suction position are converted into the tool coordinates of the electromagnetic tooling, and the electromagnetic tooling is guided to the magnetic suction position to automatically suck the transformer cover and lift the transformer cover; The preset oil pumping robot automatically inserts the oil pumping tool at the end into the transformer housing and starts pumping oil, and stops automatically when the oil pumping is completed; The electromagnetic tooling transports the transformer cover to the coil disassembly station, and the mechanical arms preset at both ends automatically use hydraulic shears to separate the connectors between the transformer cover and the winding; The RGB image of the transformer is segmented. With the help of the RGB image segmentation results, the three-dimensional coordinates of the position of the yoke and coil on the winding are determined from the pre-built transformer type and winding model; the three-dimensional coordinates are converted into the preset robot end tool coordinates, and the robot end clamping tool is guided to clamp the upper yoke for disassembly; Carry out the storage of loose parts, shells and materials.

7. Transformer automatic disassembly system, characterized in that: include: The image acquisition module is configured to: obtain an RGB image and a depth image of the transformer; The image segmentation module is configured to: segment the upper surface of the transformer using the RGB image and the depth image to obtain a segmented image; The candidate position determination module is configured to: determine the candidate position of the bolt in a preset list using the segmented image; wherein the information in the preset list includes the transformer type and the bolt position corresponding to the transformer type; The bolt position detection module is configured to: detect the bolt position in the RGB image using a preset bolt position detection model, compare the detected bolt position with the candidate position, and correct the bolt position; wherein the bolt position detection model is a weighted box fusion model; The correction module is configured to: determine the three-dimensional coordinates of the bolt in a preset three-dimensional model using the corrected bolt position; The disassembly module is configured to: convert the three-dimensional coordinates of the bolt into the coordinates of a preset robot end tool to remove the bolt, thereby realizing the disassembly of the transformer cover.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the transformer automatic disassembly method according to any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the program, the steps of the transformer automatic disassembly method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the transformer automatic disassembly method according to any one of claims 1 to 6 are implemented.

Images