Renewable electric power material recovery system and method

Abstract

The invention belongs to the field of waste electric power material recycling control, and provides a renewable electric power material recycling system and method. The renewable electric power material recycling system comprises a disassembling dispatching center, waste and old material storage equipment, recycling storage logistics equipment and a plurality of disassembling production lines. The disassembly dispatching center is used for generating a matched waste electric power material warehouse-out feeding instruction according to the waste electric power material disassembly plan in combination with the waste electric power material storage information and the positions of the disassembly production lines and issuing the matched waste electric power material warehouse-out feeding instruction to the waste and old material storage equipment; after the disassembling production line executes the disassembling process, a recycling warehouse logistics equipment dispatching instruction is generated according to the current state information of the recycling warehouse logistics equipment, the type of the disassembled material and the warehouse position of the material; and the recycling warehouse logistics equipment receives and responds to the recycling warehouse logistics equipment dispatching instruction, and the corresponding recycling materials are conveyed to the corresponding warehouses.

Description

Technical Field

[0001] The present invention belongs to the field of waste electric power material recycling control, and in particular relates to a renewable electric power material recycling system and method. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] There are many types of waste power materials, including but not limited to waste cables, waste transformers and waste electricity meters. In order to meet the requirements for the reuse of waste power materials, waste power materials are currently dismantled and the usable materials are recycled and reused. However, the traditional technology for dismantling and recycling waste power materials has the following technical problems:

[0004] (1) The dismantling control center only dismantles and recycles one type of waste power materials, and is not suitable for the dismantling and recycling of multiple types of waste power materials;

[0005] (2) The traditional dismantling process of waste power materials is complicated, and the storage of the parts after dismantling is relatively chaotic, which is not conducive to subsequent recycling and utilization;

[0006] (3) In the process of transporting recycled materials, there are problems such as information asymmetry between vehicle loading, cargo transportation and vehicles, and mismatched itineraries, which increases logistics costs and reduces recycling efficiency. Summary of the invention

[0007] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a renewable electric power material recycling system and method, which can be applicable to the disassembly and recycling control of various waste electric power materials, and can orderly store some of the disassembled parts, and also make the loading and transportation of recycled and reused materials vehicles information symmetrical and itinerary matching between vehicles, thereby reducing logistics costs and improving recycling efficiency.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] A first aspect of the present invention provides a renewable electricity material recovery system.

[0010] A renewable power material recycling system, comprising: a dismantling and dispatching center, waste material storage equipment, recycling storage and logistics equipment and several dismantling production lines;

[0011] The dismantling dispatching center is used to: generate matching instructions for loading and unloading waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the locations of each dismantling production line, and issue them to the waste power materials storage equipment;

[0012] The waste material storage equipment is used to: according to the waste power material outbound loading instruction, take out the corresponding type of waste power materials from the set warehouse and transport them to the corresponding disassembly production line;

[0013] The disassembly production line is used to: automatically trigger the disassembly process when detecting that the corresponding waste power materials are in place, generate disassembled materials and feed back its own disassembly progress to the disassembly dispatching center;

[0014] The disassembly scheduling center is further used to: after the disassembly production line completes the disassembly process, generate a recycling storage logistics equipment scheduling instruction according to the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location;

[0015] The recycling storage logistics equipment is used to: receive and respond to the recycling storage logistics equipment scheduling instructions, and transport the corresponding recycled materials to the corresponding warehouse.

[0016] As an implementation mode, the dismantling dispatching center is also used to: predict the generation amount and type of waste power materials in the future period based on historical waste power material dismantling data and pre-trained machine learning algorithms, and then optimize the waste power material dismantling plan.

[0017] As an implementation mode, the process of the disassembly dispatching center generating a dispatching instruction for recycling storage logistics equipment includes:

[0018] When the disassembly production line completes the disassembly process, it identifies the packing volume of the disassembled materials, retrieves the warehouse location that matches the disassembled materials, and retrieves the current status information of the storage and logistics equipment;

[0019] Determine the quantity of recycled storage and logistics equipment based on the packing volume of disassembled materials and the conveying capacity of recycled storage and logistics equipment;

[0020] Prioritize the idle recycling storage and logistics equipment, and then select all recycling storage and logistics equipment for this transportation task according to the closest end time to the previous transportation task;

[0021] With the shortest transportation path as the goal, combined with the packing starting point of the disassembled materials and the warehouse location that matches the disassembled materials, the recycling storage logistics equipment scheduling instructions are generated.

[0022] As an implementation method, in each of the dismantling production lines, automatic loading and unloading target detection is also performed on the corresponding waste power materials, and the process is as follows:

[0023] Obtain the number of the current process of the waste electrical materials and the corresponding process actual image set;

[0024] According to the number of the current process, a target detection model corresponding to the current process is retrieved from the set power material equipment model library, and the target detection model is used to identify the actual image set of the process to identify whether the actual image set of the current process and the standard image set of the process are consistent with each other;

[0025] Among them, each process in the power material equipment model library is set to be associated with the target detection model through the process number;

[0026] The target detection model corresponding to each process is trained using the process standard image set corresponding to each process of waste electrical materials.

[0027] As an implementation method, the target detection model is a CrossKD target detection model.

[0028] As an implementation method, the process standard image set corresponding to each process includes the standard initial state image of the waste electrical materials in each process and the standard component images that need to be assembled or disassembled in each process.

[0029] As an implementation mode, the standard initial state image and the standard component image both include a first sample image with a label and a second sample image without a label; using the process standard image set of each process to train the CrossKD target detection model of each process, including:

[0030] A process standard image set of a process is arbitrarily selected, and based on the first sample image, the teacher model is used to perform knowledge distillation training on the student model, and the trained student model is determined as the initial target detection model of each process;

[0031] Based on the second sample image, using the teacher model to perform knowledge distillation training processing on the initial target detection model of each process, and determining the trained initial target detection model as the CrossKD target detection model of each process;

[0032] The teacher model is a model pre-trained based on the first sample image, and the knowledge distillation training process includes feature layer distillation training and logic layer distillation training.

[0033] As an implementation mode, the teacher model includes a first feature extraction module and a first segmentation module;

[0034] The student model includes a second feature extraction module, and based on the first sample image, performs knowledge distillation training processing on the student model using the teacher model, including:

[0035] Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain the first image feature output by the teacher model and the second image feature output by the student model;

[0036] Inputting the first image feature into the first segmentation module to obtain a first target detection result output by the teacher model;

[0037] Inputting the second image feature into the first segmentation module to obtain a second object detection result output by the teacher model with respect to the student model;

[0038] Calculating a first loss between the first image feature and the second image feature with respect to feature layer distillation training, and calculating a second loss between the first object detection result and the second object detection result with respect to logic layer distillation training;

[0039] Inversely update the model parameters of the student model based on the first loss and the second loss.

[0040] As an implementation manner, the first sample image is respectively input into the first feature extraction module and the second feature extraction module to obtain the first image feature output by the teacher model and the second image feature output by the student model, including:

[0041] Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain a first feature map output by the first feature extraction module and a second feature map output by the second feature extraction module;

[0042] Normalizing the pixel values ​​in the first feature map and the second feature map respectively to obtain the first image feature and the second image feature.

[0043] As an implementation manner, in the process of reversely updating the model parameters of the student model based on the first loss and the second loss, the method further includes:

[0044] Determine whether the first loss and the second loss have reached convergence;

[0045] If so, it is determined that the training of the student model is completed, and the trained student model is determined as the initial target detection model.

[0046] As an implementation mode, the actual process image set of the current process is consistent with the process standard image set as follows:

[0047] The actual parts that need to be assembled or disassembled in the current process identified are consistent with the standard parts images that need to be assembled or disassembled in the current process, and the actual initial state image of the waste electrical materials in the current process is consistent with the standard initial state of the current process;

[0048] The actual process image set of the current process does not match the process standard image set:

[0049] The actual parts identified to be assembled or disassembled in the current process are inconsistent with the standard parts images to be assembled or disassembled in the current process, and / or the actual initial state image of the waste electrical materials in the current process is inconsistent with the standard initial state of the current process.

[0050] As an implementation mode, in the disassembly production line, when the waste power material is a waste transformer, the corresponding disassembly process is as follows:

[0051] Get the RGB image and depth image of the transformer;

[0052] Use the RGB image and the depth image to segment the upper surface of the transformer and obtain the segmented image;

[0053] 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;

[0054] 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;

[0055] Using the corrected bolt position, determine the three-dimensional coordinates of the bolt in the preset three-dimensional model;

[0056] 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.

[0057] As an implementation mode, in the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, the upper surface of the transformer is segmented using a long short-term memory artificial neural network. Specifically, first, the depth information is converted into three channel information of disparity, surface normal and height; then, the context information of the upper surface of the transformer in different directions is extracted using a recurrent event network, and bidirectional propagation is performed in two directions; at the same time, for the RGB channel information in the upper surface image of the transformer, 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 at all levels to the same resolution and cascade; finally, the context information is obtained using a recurrent event network.

[0058] As an implementation mode, in the disassembly production line, when the waste electric power materials are waste transformers, in the corresponding disassembly process, only the point cloud of the upper surface of the transformer is retained through the connected component labeling algorithm and noise removal 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; and 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.

[0059] As an implementation mode, in the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, the RGB information and the depth information are converted into point cloud information, and the upper surface of the transformer is segmented; 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 pre-formed upper surface model of the transformer, wherein the upper surface model of the transformer is a point pair feature model; and 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.

[0060] As an implementation mode, in the disassembly production line, when the scrap electric power material is a scrap transformer, in the corresponding disassembly process, 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 the bolt that is not detected due to external interference, the candidate position is regarded as the bolt position, for the 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, and for the 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.

[0061] As an implementation mode, in the disassembly production line, when the waste power material is a waste transformer, in the corresponding disassembly process, 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, and the three-dimensional coordinates of the midpoint of the magnetic suction position are obtained; 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;

[0062] 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;

[0063] 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;

[0064] 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 to guide the robot end clamping tooling to clamp the upper yoke for disassembly.

[0065] As an implementation method, when the waste power materials are waste transformers, the corresponding disassembly production line includes a conveyor line, and a weighing mechanism and a code scanning mechanism arranged on the conveyor line;

[0066] A screw removal robot arm, an oil pumping robot arm, a power-assisted cantilever crane, a coil disassembly robot arm, a sorting robot arm and a three-dimensional warehouse are arranged near the conveyor line; the screw removal robot arm, the oil pumping robot arm, the coil disassembly robot arm and the sorting robot arm are all provided with rotating parts through adjusting mechanisms and connecting rods, and the rotating parts are provided with connecting heads for connecting various tools.

[0067] As an implementation method, when the waste power materials are waste cables, the corresponding disassembly production line includes: an automatic feeding device, an automatic picking and grabbing device, an automatic disassembly device and a crushing and screening device arranged in sequence along the main transmission line;

[0068] The automatic feeding device is used to automatically transport the waste electric meters to the entrance end of the automatic picking and grabbing device;

[0069] The automatic picking and grabbing device is used to automatically pick up a single waste electric meter and transfer it to the workstation where the automatic disassembly device is located;

[0070] The automatic disassembly device is used to automatically disassemble the waste electric meter on its workstation into set parts;

[0071] The crushing and screening device is used to crush the disassembled parts and separate and screen the recyclable materials according to their properties and store them in corresponding material receiving boxes.

[0072] As an implementation method, when the waste power materials are waste cables, the corresponding disassembly production line includes: an automatic cable feeding device, a primary stripping device, a wire core separation and feeding device, and an automatic crushing and sorting device arranged in sequence along the main transmission line; a secondary stripping device and an automatic packaging device are arranged in sequence on the branch transmission line connected to the rear end of the wire core separation and feeding device;

[0073] The automatic cable feeding device is used to separate the piled cables into individual cables by step transmission, and to sequentially transmit each individual cable to the primary stripping device;

[0074] The primary stripping device is used to adaptively clamp the cable according to the cable diameter and perform primary stripping to obtain a primary stripping product; the primary stripping product includes inner and outer sheaths, fillers, steel armor and wire cores;

[0075] The wire core separation and feeding device is used to transmit and feed the primary stripping product in layers;

[0076] The secondary stripping device is used to perform secondary stripping on the wire core to obtain the metal wire core and transmit it to the automatic packaging device for packaging;

[0077] The automatic crushing and sorting device is used for automatically crushing and sorting the inner and outer sheaths, fillers and steel armor.

[0078] As an implementation method, when the waste power materials are composite insulators, the corresponding disassembly production line includes: a feeding conveyor line for transporting the insulators to be recycled;

[0079] The insulator fixing mechanism receives the insulators from the feeding conveyor line and fixes the insulators to be recovered with the cooperation of the feeding and handling mechanism;

[0080] The cutting manipulator uses the cutting tool carried at the end to cut and separate the shed part along the axial direction of the insulator to be recovered in cooperation with the insulator fixing mechanism, and then cuts and separates the glass fiber rod and metal head in the insulator after the shed is separated;

[0081] The sorting conveyor line uses a sorting robot to sort the separated sheds, fiberglass rods and metal heads into corresponding recycling containers.

[0082] A second aspect of the present invention provides a method for recycling renewable electricity materials.

[0083] A method for recycling renewable electricity materials, comprising:

[0084] The dismantling dispatch center generates matching instructions for the delivery and loading of waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the location of each dismantling production line, and sends them to the waste material storage equipment;

[0085] The waste material storage equipment takes out the corresponding type of waste power materials from the set warehouse and transports them to the corresponding disassembly production line according to the waste power material outbound loading instructions;

[0086] When the corresponding waste power materials are detected to be in place, the disassembly process of the disassembly production line is automatically triggered, the disassembly materials are generated and the disassembly progress is fed back to the disassembly dispatch center;

[0087] When the disassembly production line completes the disassembly process, the disassembly dispatch center also generates a dispatch instruction for the recycling storage logistics equipment based on the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location;

[0088] The recycling storage logistics equipment receives and responds to the recycling storage logistics equipment dispatching instruction, and transports the corresponding recycling materials to the corresponding warehouse.

[0089] The beneficial effects of the present invention are:

[0090] (1) The present invention innovatively provides a waste power material disassembly and recycling control technology. According to the waste power material disassembly plan, disassembly production line and status information of recycling storage and logistics equipment, a disassembly scheduling center generates corresponding scheduling and disassembly strategies, thereby realizing orderly storage of some disassembled parts. Through the dynamic and static combined scheduling planning of recycling storage and logistics equipment, the logistics cost is reduced and the recycling efficiency is improved.

[0091] (2) The present invention innovatively provides a target detection technology for automatic loading and unloading of waste electric power materials. By identifying the actual image set of the process, it is determined whether the actual image set of the current process is consistent with the standard image set of the process, thereby solving the shortcomings of the prior art in terms of recognition accuracy, computing resource requirements and environmental adaptability, and realizing target detection of waste electric power materials with complex and diverse structures and changeable operating environments, thereby improving the accuracy and stability of waste electric power material detection.

[0092] (3) The present invention innovatively proposes a transformer automatic disassembly strategy for determining bolt positions based on a pre-stored list. The candidate positions of the bolts are determined in a preset list including transformer types by using the segmentation of the YOLACT neural network instance and the 6D pose matching image with the target matching on the transformer. In the RGB image, the bolt position is detected by using a preset bolt position detection model. The detected bolt position is compared with the candidate position, and the bolt position is corrected. The corrected bolt position is used to determine the three-dimensional coordinates of the bolt in the preset three-dimensional model. The bolt position of the corresponding model transformer 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 excessive data used in the judgment and correction process are avoided, thereby reducing the impact on the control speed and efficiency.

[0093] (4) The present invention creatively proposes a human-machine collaborative method for dismantling batch waste transformers. During the transportation of the waste transformers on the conveyor line, the weighing mechanism and the scanning mechanism are used to weigh and scan to obtain information; the waste transformers are dismantled by the screw removal robot arm, the oil extraction robot arm, the coil dismantling robot arm and the sorting robot arm, thereby solving the problem of dismantling large quantities of waste transformers and improving the dismantling efficiency of batch waste transformers.

[0094] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0096] Figure 1 is a schematic diagram of a renewable electricity material recovery system according to an embodiment of the present invention;

[0097] Figure 2 is a teacher-student guidance framework diagram of the CrossKD target detection model according to an embodiment of the present invention;

[0098] Figure 3 is a flowchart of the training process of the CrossKD target detection model according to an embodiment of the present invention;

[0099] Figure 4 This is a schematic diagram of a waste transformer disassembly production line according to an embodiment of the present invention;

[0100] Figure 5 This is a schematic diagram of the structure of a waste electric energy meter disassembly production line according to an embodiment of the present invention;

[0101] Figure 6 This is a schematic structural diagram of an automatic disassembly device for a waste electric energy meter disassembly production line according to an embodiment of the present invention;

[0102] Figure 7 The present invention is a schematic diagram of the structure of a crushing and screening device for a waste electric energy meter disassembly production line according to an embodiment of the present invention.

[0103] Among them, 1-1, conveyor line; 1-2, weighing mechanism; 1-3, code scanning mechanism; 1-4, screw removal mechanical arm; 1-5, oil pumping mechanical arm; 1-6, power cantilever crane; 1-7, coil disassembly mechanical arm;

[0104] 2-1. Main transmission line; 2-2. Automatic picking and grabbing device; 2-3. Automatic disassembling device; 2-4. Material separation operation table; 2-5. Double-layer conveyor line; 2-6. Crushing and screening device; 2-7. Material receiving box; 2-8. First spindle milling cutter; 2-9. Coordinate robot; 2-10. Vision system; 2-11. Second spindle milling cutter; 2-12. Shredder; 2-13. Magnetic separator; 2-14. Air separator; 2-15. Eddy current separator; 2-16. Conveyor belt. DETAILED DESCRIPTION

[0105] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0106] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. 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 invention belongs.

[0107] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0108] Figure 1 Schematic diagram of the renewable electricity material recovery system according to an embodiment of the present invention. Figure 1 , a renewable power material recycling system according to an embodiment of the present invention comprises: a dismantling and dispatching center, waste material storage equipment, recycling storage and logistics equipment and a number of dismantling production lines;

[0109] The dismantling dispatching center is used to: generate matching instructions for loading and unloading waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the locations of each dismantling production line, and issue them to the waste power materials storage equipment;

[0110] The waste material storage equipment is used to: according to the waste power material outbound loading instruction, take out the corresponding type of waste power materials from the set warehouse and transport them to the corresponding disassembly production line;

[0111] The disassembly production line is used to: automatically trigger the disassembly process when detecting that the corresponding waste power materials are in place, generate disassembled materials and feed back its own disassembly progress to the disassembly dispatching center;

[0112] The disassembly scheduling center is further used to: after the disassembly production line completes the disassembly process, generate a recycling storage logistics equipment scheduling instruction according to the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location;

[0113] The recycling storage logistics equipment is used to: receive and respond to the recycling storage logistics equipment scheduling instructions, and transport the corresponding recycled materials to the corresponding warehouse.

[0114] During the specific implementation process, the dismantling and dispatching center is also used to: predict the amount and type of waste power materials generated in the future based on historical waste power material dismantling data and pre-trained machine learning algorithms, and then optimize the waste power material dismantling plan.

[0115] The machine learning algorithm here can be implemented using an existing neural network structure model, such as CNN, etc., which will not be described in detail here.

[0116] In one or more embodiments, the process of the disassembly scheduling center generating a scheduling instruction for recycling storage logistics equipment includes:

[0117] When the disassembly production line completes the disassembly process, it identifies the packing volume of the disassembled materials, retrieves the warehouse location that matches the disassembled materials, and retrieves the current status information of the storage and logistics equipment;

[0118] Determine the quantity of recycled storage and logistics equipment based on the packing volume of disassembled materials and the conveying capacity of recycled storage and logistics equipment;

[0119] Prioritize the idle recycling storage and logistics equipment, and then select all recycling storage and logistics equipment for this transportation task according to the closest end time to the previous transportation task;

[0120] With the shortest transportation path as the goal, combined with the packing starting point of the disassembled materials and the warehouse location that matches the disassembled materials, the recycling storage logistics equipment scheduling instructions are generated.

[0121] In this way, by designing the packaging of recycling warehousing logistics equipment and the static and dynamic scheduling planning scheme of recycling warehousing logistics equipment, the problems of loading and transporting recycling warehousing logistics equipment, information symmetry, and itinerary matching between recycling warehousing logistics equipment can be solved, thereby reducing logistics costs and improving recycling efficiency.

[0122] As an implementation method, in each of the dismantling production lines, automatic loading and unloading target detection is also performed on the corresponding waste power materials, and the process is as follows:

[0123] S101, obtaining the number of the current process of the waste electrical materials and the corresponding actual process image set.

[0124] In this embodiment, the process actual image set of the current process of the scrap electrical materials includes: the actual initial state image of the scrap electrical materials in the current process and the images of the actual parts that need to be assembled or disassembled in the current process.

[0125] S102, according to the number of the current process, retrieve the corresponding target detection model from the model library, use the target detection model to identify the actual image set of the process, and identify whether the actual image set of the current process and the standard image set of the process are consistent with each other;

[0126] Among them, each process in the model library is associated with the target detection model through the process number;

[0127] The target detection model corresponding to each process is trained using the process standard image set corresponding to each process of waste electrical materials.

[0128] In the specific implementation process, the actual process image set of the current process is consistent with the process standard image set:

[0129] The actual parts that need to be assembled or disassembled in the current process identified are consistent with the standard parts images that need to be assembled or disassembled in the current process, and the actual initial state image of the waste electrical materials in the current process is consistent with the standard initial state of the current process;

[0130] The actual process image set of the current process does not match the process standard image set:

[0131] The actual parts identified to be assembled or disassembled in the current process are inconsistent with the standard parts images to be assembled or disassembled in the current process, and / or the actual initial state image of the waste electrical materials in the current process is inconsistent with the standard initial state of the current process.

[0132] In one or more embodiments, it is also determined based on the judgment result whether to output a prompt signal regarding the actual initial state of the waste electrical materials in the current process and whether the actual parts to be assembled or disassembled are detected normally.

[0133] If the actual process image set of the current process matches the process standard image set, a prompt signal is outputted indicating that the actual initial state of the waste electrical materials in the current process and the actual parts to be assembled or disassembled are normal;

[0134] If the actual process image set of the current process does not match the process standard image set, an alarm signal is output regarding the actual initial state of the waste electrical materials in the current process and abnormal detection of actual parts that need to be assembled or disassembled.

[0135] In this embodiment, the target detection model is a CrossKD target detection model, such as Figure 2 shown.

[0136] The process standard image set corresponding to each process includes the standard initial state image of the waste electrical materials in each process and the standard component images that need to be assembled or disassembled in each process.

[0137] Specifically, the standard initial state image and the standard component image both include a first sample image with a configuration label and a second sample image without a configuration label; the CrossKD target detection model of each process is trained using the process standard image set of each process, such as Figure 3 As shown, including:

[0138] A process standard image set of a process is arbitrarily selected, and based on the first sample image, the teacher model is used to perform knowledge distillation training on the student model, and the trained student model is determined as the initial target detection model of each process;

[0139] Based on the second sample image, using the teacher model to perform knowledge distillation training processing on the initial target detection model of each process, and determining the trained initial target detection model as the CrossKD target detection model of each process;

[0140] The teacher model is a model pre-trained based on the first sample image, and the knowledge distillation training process includes feature layer distillation training and logic layer distillation training.

[0141] It should be noted that the acquisition module of the present invention can simultaneously acquire a first sample image with configured labels and a second sample image without configured labels; based on the first sample image, the teacher model is used to perform knowledge distillation training on the student model, and the trained student model is determined as the initial target detection model; based on the second sample image, the teacher model is used to perform knowledge distillation training on the initial target detection model to obtain the target detection model. By training the target detection model using knowledge distillation, the feature extraction capability of the target detection model can be improved, so that the target detection model can ensure the accuracy of target detection while being successfully applied on the vehicle platform. In addition, during the training process, by adopting a semi-supervised training method, the unlabeled image data can be accurately used to improve the model performance, and there is no need to label a large amount of image data, which can reduce the labeling cost of the training samples.

[0142] Specifically, the process standard image sets corresponding to each process include multiple standard initial state images and multiple standard component images. The number of standard initial state images and standard component images can be selected according to actual needs. In the present invention, the number of standard initial state images and standard component images is preferably 200. Among the 200 standard initial state images, the shooting angle of each standard initial state image varies.

[0143] Among them, a preset proportion of images are labeled images, recorded as the first sample images, and the rest are unlabeled images, recorded as the second sample images. Preferably, the preset proportion is 50%. After obtaining the sample images, a small number of sample images can be pre-labeled. Specifically, the classification label of each pixel in the sample image can be marked. The classification label can be the number of each process. Among the 200 standard parts images, the shooting angle of each standard part image varies. Among them, a preset proportion of images are labeled images, recorded as the first sample images, and the rest are unlabeled images, recorded as the second sample images.

[0144] Preferably, the preset ratio is 50%. After acquiring the sample images, a small number of sample images can be pre-labeled. Specifically, the classification labels of each pixel in the sample images can be labeled. The classification labels can be the names of various components. Then, the labeled sample images are determined as the first sample images, and the remaining unlabeled sample images are determined as the second sample images. In the subsequent training of the target detection model, semi-supervised training can be performed based on the first sample images and the second sample images, and the unlabeled image data can be accurately used to improve the model performance. There is no need to label a large amount of image data, which can reduce the labeling cost of training samples.

[0145] Specifically, the teacher model includes a first feature extraction module and a first segmentation module;

[0146] The student model includes a second feature extraction module, and based on the first sample image, performs knowledge distillation training processing on the student model using the teacher model, including:

[0147] Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain the first image feature output by the teacher model and the second image feature output by the student model;

[0148] Inputting the first image feature into the first segmentation module to obtain a first target detection result output by the teacher model;

[0149] Inputting the second image feature into the first segmentation module to obtain a second object detection result output by the teacher model with respect to the student model;

[0150] Calculating a first loss between the first image feature and the second image feature with respect to feature layer distillation training, and calculating a second loss between the first object detection result and the second object detection result with respect to logic layer distillation training;

[0151] Inversely update the model parameters of the student model based on the first loss and the second loss.

[0152] In some optional embodiments, the CWD method can be used to input the first sample image into the first feature extraction module in the teacher model and the second feature extraction module in the student model respectively. The first feature extraction module will output multiple first feature maps under the preset feature layer number, and then the softmax normalization processing can be performed on each pixel value in each of the multiple first feature maps, so that each pixel value is distributed between 0-1, and the first image feature T1 is obtained; at the same time, the second feature extraction module will output multiple second feature maps under the preset feature layer number, and then the softmax normalization processing can be performed on each pixel value in each of the multiple second feature maps, so that each pixel value is distributed between 0-1, and the second image feature S1 is obtained. Among them, the preset feature layer number can be set according to the actual application scenario, such as 256 layers, 512 layers, etc.

[0153] The CrossKD method can be used to forward propagate the teacher model, input the first image feature into the first segmentation module, and generate a category probability map for each pixel, that is, the first target detection result; the second image feature output by the student model is taken out, and the second image feature is input into the segmentation head (that is, the first segmentation module) of the teacher model. The first segmentation module is used to process the second image feature into the number of categories to be segmented. For example, the number of segmentation categories is 19, and the shape of the feature vector at this time is [2,256,224,224]. After being processed by the segmentation head, the vector shape is [2,19,224,224], where 2 is the number of batches, 19 is the number of channels, and 224 is the width and height of the image. Finally, the category probability map for each pixel output by the first segmentation module is obtained, that is, the second target detection result of the student model output by the teacher model.

[0154] The first sample image is input into the first feature extraction module and the second feature extraction module respectively to obtain the first image feature output by the teacher model and the second image feature output by the student model, including:

[0155] Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain a first feature map output by the first feature extraction module and a second feature map output by the second feature extraction module;

[0156] Normalizing the pixel values ​​in the first feature map and the second feature map respectively to obtain the first image feature and the second image feature.

[0157] In some other optional embodiments, the DIST method can be used. After obtaining the first image feature and the second image feature, the Pearson correlation coefficient can be used as a loss function to calculate the first loss of the second image feature and the first image feature, and the student model parameters are updated by back propagation based on the first loss. The first loss is used to represent the completion of the feature layer distillation training under the model training with labeled data. The closer the first loss is to 0, the higher the completion of the feature layer distillation training. After obtaining the first target detection result and the second target detection result, the Pearson correlation coefficient can be used as a loss function to calculate the second loss of the second target detection result and the first target detection result, and the student model parameters are updated by back propagation based on the second loss. The second loss is used to represent the completion of the logic layer distillation training under the model training with labeled data. The closer the second loss is to 0, the higher the completion of the logic layer distillation training.

[0158] In the process of reversely updating the model parameters of the student model based on the first loss and the second loss, the method further includes:

[0159] Determine whether the first loss and the second loss have reached convergence;

[0160] If so, it is determined that the training of the student model is completed, and the trained student model is determined as the initial target detection model.

[0161] In some other optional embodiments, the teacher model includes a first feature extraction module and a first segmentation module, the initial target detection model includes a third feature extraction module and a second segmentation module, and the initial target detection model is subjected to the knowledge distillation training process by using the teacher model based on the second sample image, including:

[0162] Inputting the second sample image into the first feature extraction module and the third feature extraction module respectively to obtain the third image feature output by the teacher model and the fourth image feature output by the initial object detection model;

[0163] Inputting the third image feature into the first segmentation module to obtain a third object detection result output by the teacher model;

[0164] Inputting the fourth image feature into the second segmentation module to obtain a fourth object detection result output by the initial object detection model;

[0165] Calculating a third loss between the third image feature and the fourth image feature with respect to feature layer distillation training, and calculating a fourth loss between the third object detection result and the fourth object detection result with respect to logic layer distillation training;

[0166] The model parameters of the initial target detection model are reversely updated based on the third loss and the fourth loss.

[0167] Specifically, the method further includes:

[0168] The DIST method can be used. After obtaining the third image feature and the fourth image feature, the Pearson correlation coefficient can be used as the loss function to calculate the third loss of the fourth image feature and the third image feature, and the initial target detection model parameters can be updated by back propagation based on the third loss. The third loss is used to represent the completion of the feature layer distillation training under the model training of unlabeled data. The closer the third loss is to 0, the higher the completion of the feature layer distillation training. After obtaining the third target detection result and the fourth target detection result, the Pearson correlation coefficient can be used as the loss function to calculate the fourth loss of the fourth target detection result and the third target detection result, and the initial target detection model parameters can be updated by back propagation based on the fourth loss. The fourth loss is used to represent the completion of the logic layer distillation training under the model training of unlabeled data. The closer the fourth loss is to 0, the higher the completion of the logic layer distillation training.

[0169] Accordingly, when back propagation is used to update the student model parameters, the model gradient corresponding to the third loss or the fourth loss can be calculated based on the back propagation algorithm, and the model parameters of the initial target detection model can be updated according to the model gradient. In a specific application scenario, the number of training rounds of the initial target detection model can be set according to the data size of the second sample image, and after it is determined that the iterative training of the initial target detection model reaches the number of training rounds, it is determined whether the third loss and the fourth loss have reached convergence.

[0170] In summary, the technical solution in this application can improve the feature extraction capability of the target detection model by training the target detection model through the method of knowledge distillation, so that the target detection model can have the same performance as the large model when it can be successfully applied to the target monitoring scheme of automatic loading and unloading of waste power materials, and ensure the accuracy of target detection. In addition, during the training process, by adopting a semi-supervised training method, the unlabeled image data is accurately used to improve the model performance, and a large amount of image data annotation is not required, which can reduce the annotation cost of training samples.

[0171] When the third loss and the fourth loss reach a convergence state, it can be determined that the initial target detection model training is completed, and the trained initial target detection model is determined as the final target detection model. Among them, the preset thresholds corresponding to the feature layer distillation training and the logic layer distillation training can be the same or different, and can be a value between 0 and 1. The closer the preset threshold is to 0, the higher the training accuracy requirement of the corresponding feature layer distillation training or logic layer distillation training. The specific value can be set according to the actual application scenario, and no specific limitation is made here.

[0172] Specifically, the step of inputting the second sample image into the first feature extraction module and the third feature extraction module respectively to obtain the third image feature output by the teacher model and the fourth image feature output by the initial target detection model includes:

[0173] Inputting the second sample image into the first feature extraction module and the third feature extraction module respectively to obtain a third feature map output by the first feature extraction module and a fourth feature map output by the third feature extraction module;

[0174] Normalization is performed on the pixel values ​​in the third feature map and the fourth feature map, respectively, to obtain the third image feature and the fourth image feature.

[0175] Specifically, the method further includes:

[0176] Determine whether the third loss and the fourth loss have reached convergence; if so, determine that the initial target detection model training is completed, and determine the trained initial target detection model as the target detection model.

[0177] In order to deal with the problem of target conflict, this paper proposes a cross-head knowledge distillation method (CrossKD), which aims to optimize the knowledge transfer process and enhance the performance of the student model in the automatic loading and unloading task of waste power materials. The core idea of ​​CrossKD is to generate cross-head predictions by passing the intermediate feature maps of the student model to the detection head of the teacher model. This method increases the consistency between the teacher and student models and effectively reduces the impact of target conflict.

[0178] Specifically, for a dense detector (such as RetinaNet), its detection head consists of a series of convolutional layers. Assume that the detection head has a total of n convolutional layers, of which the first n-1 layers generate the intermediate feature map f i , and the last layer generates the prediction result p. The operation of CrossKD is to pass the feature map of the middle layer of the student model to the corresponding layer of the teacher detection head, for example, the i-th layer feature map of the student Passed to the teacher's i+1th convolutional layer Thus generating cross-head predictions This process can be expressed as:

[0179]

[0180] Among them, S(·) and They are the region selection principle and the normalization factor. This loss calculation method avoids the trap of direct prediction differences between teachers and students, allowing some of the student's detection heads to better optimize the real target. This distillation method is particularly suitable for complex task scenarios, such as automatic loading and unloading and disassembly of waste power materials, because it can effectively handle multi-target and multi-scale detection requirements.

[0181] Through this cross-head prediction approach, CrossKD effectively reduces the inconsistency between the teacher and student models, thereby alleviating the negative impact of target conflict on the distillation effect. Compared with traditional prediction imitation methods, CrossKD is unique in that it does not simply minimize the direct prediction difference between the teacher and the student, but enhances the collaborative work between the teacher and the student through the sharing of cross-head features, so that the student model can better learn the deep features and prediction strategies of the teacher model.

[0182] CrossKD is designed to be particularly suitable for complex task scenarios, such as automatic loading and unloading and disassembly of waste electrical materials. Such tasks usually have multi-target and multi-scale detection requirements, and are also accompanied by problems of equipment diversity and environmental complexity. By sharing feature maps, CrossKD can better handle multi-target and multi-scale detection requirements and improve the generalization and robustness of the model.

[0183] In addition, since the automated operation of waste power materials requires high-precision target positioning and classification, CrossKD can show significant performance improvement in such scenarios. It can effectively reduce the prediction error caused by target conflict, thereby improving the accuracy and efficiency of the system in practical applications.

[0184] The overall training loss function consists of the weighted sum of the detection loss and the distillation loss:

[0185]

[0186] in, and Represents classification loss and regression loss, respectively, calculated between the student model prediction and the true target. The additional CrossKD loss and Used for distillation between cross-head predictions and teacher predictions.

[0187] In the automatic loading and unloading task of waste power materials, the classification loss can use the quality focus loss to process the soft label prediction, and the regression loss uses GIoU loss or KL divergence according to the regression form of the detector. Through the optimization of CrossKD, the model can complete the detection, positioning and classification of transformers more accurately and efficiently, thereby improving the accuracy and robustness of the automatic loading and unloading system.

[0188] In the automatic loading and unloading task of waste electrical materials, the classification loss can use Quality Focal Loss to process soft label prediction. This loss function can better balance the ratio between positive and negative samples, and is particularly suitable for dense detection tasks. The regression loss uses GIoU loss or KL divergence according to the regression form of the detector. These loss functions can provide more accurate bounding box adjustments during the regression process and improve the model's prediction accuracy for the target position.

[0189] CrossKD's distillation loss aligns the cross-head predictions with the teacher model's predictions, thus avoiding the pitfall of direct teacher-student prediction differences in traditional distillation methods. This design can better maintain the prediction strategy of the teacher model while allowing the student model to more accurately capture the true target through the optimization of part of the detection head.

[0190] This optimization method is particularly suitable for multi-target and multi-scale detection in complex scenarios. For example, the automated operation system for waste power materials needs to accurately detect, locate and classify targets in different environments and equipment types. CrossKD improves the design of distillation loss to enable the model to better cope with these challenges, thereby improving the accuracy and robustness of the automated system.

[0191] In one or more embodiments, in the disassembly production line, when the waste power material is a waste transformer, the corresponding disassembly process is:

[0192] Get the RGB image and depth image of the transformer;

[0193] Use the RGB image and the depth image to segment the upper surface of the transformer and obtain the segmented image;

[0194] 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;

[0195] 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;

[0196] Using the corrected bolt position, determine the three-dimensional coordinates of the bolt in the preset three-dimensional model;

[0197] 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.

[0198] Among them, in the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, the long short-term memory artificial neural network is used to segment the upper surface of the transformer. Specifically, first, the depth information is converted into three channel information of disparity, surface normal and height; then, the recurrent event network is used to extract the context information in different directions of the upper surface of the transformer, and bidirectional propagation is performed in two directions; at the same time, for the RGB channel information in the upper surface image of the transformer, 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 at all levels to the same resolution and cascade; finally, the recurrent event network is used to obtain the context information.

[0199] In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, only the point cloud of the upper surface of the transformer is retained through the connected component labeling algorithm and noise removal 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; and 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.

[0200] In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, RGB information and depth information are converted into point cloud information, and the upper surface of the transformer is segmented; 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 pre-formed upper surface model of the transformer, wherein the upper surface model of the transformer is a point pair feature model; and 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.

[0201] In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, 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, and for bolts that are not detected due to external interference, the candidate positions are regarded as the bolt positions, and for positions where the distance between the candidate positions and the previously detected positions is greater than a set threshold, they are regarded as false detections, and 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.

[0202] In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, 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, and the three-dimensional coordinates of the midpoint of the magnetic suction position are obtained; 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;

[0203] 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;

[0204] 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;

[0205] 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 to guide the robot end clamping tooling to clamp the upper yoke for disassembly.

[0206] In another embodiment, when the waste power material is a waste transformer, Figure 4 As shown, the corresponding disassembly production line includes a conveyor line 1-1, and a weighing mechanism 1-2 and a code scanning mechanism 1-3 arranged on the conveyor line;

[0207] A screw removal robot arm 1-4, an oil pumping robot arm 1-5, a power-assisted cantilever crane 1-6, a coil disassembly robot arm 1-7, a sorting robot arm and a stereoscopic warehouse are arranged near the conveyor line 1-1; the 1-4, the oil pumping robot arm 1-5, the power-assisted cantilever crane 1-6, the coil disassembly robot arm 1-7 and the sorting robot arm are all provided with rotating parts through adjusting mechanisms and connecting rods, and the rotating parts are provided with connecting heads for connecting various tools.

[0208] Specifically, screw removal robotic arms 1-4, oil extraction robotic arms 1-5, power-assisted cantilever cranes 1-6, coil disassembly robotic arms 1-7, sorting robotic arms and stereoscopic warehouses are arranged near the conveyor line, so that separate operations of shell disassembly, oil extraction and coil disassembly are realized on one conveyor line. It is not necessary to disassemble the current transformer and classify the parts after disassembly before disassembling the subsequent transformer. The disassembly of the previous and next transformers is not affected, which can be suitable for the disassembly of large quantities of waste transformers. At the same time, the weighing mechanism 1-2 and the code scanning mechanism 1-3 on the conveyor line 1-1, combined with the sorting robotic arms, can track and classify the entire disassembly process of transformers of different models, solve the problem of chaotic management of parts after disassembly, and is conducive to later recycling and other treatments.

[0209] Optional, such as Figure 1 As shown, the conveyor line 1-1 includes a track and an AGV car arranged on the track; the AGV car is used to transport the waste transformers to various workstations such as weighing, code scanning, screw removal and oil extraction. Two sides of each waste transformer are provided with a QR code label; when receiving a work task, the QR code on the transformer is scanned by a mobile terminal, and after confirming that it is consistent with the task list, the AGV car transports the transformer to the disassembly line.

[0210] Optionally, the track includes a loading conveyor line and a unloading conveyor line; wherein the loading conveyor line is used for loading waste transformers; and the unloading conveyor line is used for unloading waste transformers. Optionally, the weighing mechanism 1-2 is used to weigh the disassembled materials; optionally, the weighing mechanism 1-2 can be a floor scale or a weight scale. The code scanning mechanism 1-3 is used to scan the label code on the waste transformer to record the information of the disassembled materials.

[0211] like Figure 5-Figure 7 As shown, when the waste power materials are waste cables, the corresponding disassembly production line includes: an automatic feeding device, an automatic picking and grabbing device 2-2, an automatic disassembly device 2-3 and a crushing and screening device 2-6 arranged in sequence along the main transmission line 2-1;

[0212] The automatic feeding device is used to automatically transport the waste electric meters to the entrance end of the automatic picking and grabbing device;

[0213] The automatic picking and grabbing device 2-2 is used to automatically pick a single waste electric meter and transfer it to the station where the automatic disassembly device is located;

[0214] The automatic disassembly device 2-3 is used to automatically disassemble the waste electric meter on its workstation into set parts;

[0215] The crushing and screening device 2-6 is used to crush the disassembled components, separate and screen the recyclable materials according to their properties, and classify and store them in corresponding material receiving boxes.

[0216] In one or more embodiments, a material separation operation table 2-4 and a double-layer conveyor line 2-5 are sequentially arranged behind the automatic disassembly device, wherein the material separation operation table 2-4 is a manual auxiliary position, and the circuit board and the terminal seat are manually taken out from the lower shell, and the wire between the circuit board and the terminal seat is cut with scissors, and the circuit board is placed in the material basket. The upper and lower shells are manually placed in the lower layer of the double-layer conveyor line 2-5, and the remaining materials are placed in the upper layer.

[0217] In this embodiment, the main transmission line is a double-sided transmission line.

[0218] In a specific implementation process, the automatic loading device may be a forklift.

[0219] The used electricity meters are placed in the crossbeam shelves. When the production task is issued, the used electricity meters are transported by forklift to the double-sided conveyor line next to the robot, unpacked, and transported on the double-sided conveyor line.

[0220] In the specific implementation process, the automatic picking and grasping device is a loading robot arm.

[0221] like Figure 6 As shown, the automatic disassembly device includes a main controller, a first spindle milling cutter 2-8, a coordinate manipulator 2-9, a visual system 2-10 and a second spindle milling cutter 2-11;

[0222] The first spindle milling cutter 2-8, the coordinate manipulator 2-9, the visual system 2-10 and the second spindle milling cutter 2-11 are all connected to the main controller;

[0223] The first spindle milling cutter 2-8 is used to mill off the locking screws and the tail cover screws of the electric energy meter under the action of the main controller;

[0224] The main controller is also used to send a screw removal completion signal to the coordinate manipulator 2-9 to control the coordinate manipulator to take out the upper shell and place it on the belt line body of the main transmission line;

[0225] The visual system 2-10 is used to determine the material category and model based on the material characteristics by calling the backend database data, and then visually identify the position of the circuit board card fixing screws and transmit it to the main controller;

[0226] The main controller is used to control the second spindle milling cutter 2-11 to perform the screw milling action according to the position of the circuit board card fixing screws visually identified.

[0227] In this way, the clips or screws connecting the circuit board and the lower shell can be accurately milled off, so as to facilitate manual assistance and smoothly remove the circuit board from the lower shell.

[0228] like Figure 7 As shown, the crushing and screening device includes a shredder 2-12, a magnetic separator 2-13, a wind separator 2-14 and an eddy current separator 2-15 arranged in series in sequence; one end of the conveyor belt 2-16 is connected to the output end of the automatic disassembly device, and the other end is connected to the input end of the shredder 2-12.

[0229] The shredder 2-12 is used to shred the received housing and terminal seat;

[0230] The magnetic separator 2-13 is used to separate the iron components from the output material of the shredder;

[0231] The wind separator 2-14 is used to select the light impurities in the material after the iron component is selected by wind, and enter the material receiving box 7, and the remaining material components continue to be transported;

[0232] The eddy current separator 2-15 is used to separate copper metal and plastic from the debris output from the air separator, and discharge them through different discharge ports into a receiving box.

[0233] In one or more embodiments, the shredder is further connected to a start controller, and the start controller is connected to a material in place sensor, and the material in place sensor is used to detect whether there is material in place at the input port of the shredder and transmit a corresponding signal to the start controller.

[0234] The automatic feeding device is used to automatically transport the waste electric meters to the entrance end of the automatic picking and grabbing device;

[0235] The automatic picking and grabbing device is used to automatically pick up a single waste electric meter and transfer it to the workstation where the automatic disassembly device is located;

[0236] The automatic disassembly device is used to automatically disassemble the waste electric meter on its workstation into set parts;

[0237] The crushing and screening device is used to crush the disassembled parts and separate and screen the recyclable materials according to their properties and store them in corresponding material receiving boxes.

[0238] In one or more embodiments, when the waste power materials are waste cables, the corresponding disassembly production line includes: an automatic cable feeding device, a primary stripping device, a wire core separation and feeding device, and an automatic crushing and sorting device arranged in sequence along the main transmission line; a secondary stripping device and an automatic packaging device are arranged in sequence on the branch transmission line connected to the rear end of the wire core separation and feeding device;

[0239] The automatic cable feeding device is used to separate the piled cables into individual cables by step transmission, and to sequentially transmit each individual cable to the primary stripping device;

[0240] The primary stripping device is used to adaptively clamp the cable according to the cable diameter and perform primary stripping to obtain a primary stripping product; the primary stripping product includes inner and outer sheaths, fillers, steel armor and wire cores;

[0241] The wire core separation and feeding device is used to transmit and feed the primary stripping product in layers;

[0242] The secondary stripping device is used to perform secondary stripping on the wire core to obtain the metal wire core and transmit it to the automatic packaging device for packaging;

[0243] The automatic crushing and sorting device is used for automatically crushing and sorting the inner and outer sheaths, fillers and steel armor.

[0244] In other embodiments, when the waste power materials are composite insulators, the corresponding disassembly production line includes: a loading conveyor line for transporting the insulators to be recycled;

[0245] The insulator fixing mechanism receives the insulators from the feeding conveyor line and fixes the insulators to be recovered with the cooperation of the feeding and handling mechanism;

[0246] The cutting manipulator uses the cutting tool carried at the end to cut and separate the shed part along the axial direction of the insulator to be recovered in cooperation with the insulator fixing mechanism, and then cuts and separates the glass fiber rod and metal head in the insulator after the shed is separated;

[0247] The sorting conveyor line uses a sorting robot to sort the separated sheds, fiberglass rods and metal heads into corresponding recycling containers.

[0248] Based on the above renewable electricity material recycling system, the renewable electricity material recycling method includes:

[0249] Step 1: The dismantling dispatch center generates matching instructions for the delivery and loading of waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the location of each dismantling production line, and sends them to the waste power materials storage equipment;

[0250] Step 2: The waste material storage equipment takes out the corresponding type of waste power materials from the set warehouse and transports them to the corresponding disassembly production line according to the waste power material outbound loading instruction;

[0251] Step 3: When the corresponding waste power materials are detected to be in place, the disassembly process of the disassembly production line is automatically triggered to generate disassembled materials and feed back its own disassembly progress to the disassembly dispatching center;

[0252] Step 4: After the disassembly production line completes the disassembly process, the disassembly dispatching center also generates a dispatching instruction for the recycling storage logistics equipment based on the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location;

[0253] Step 5: The recycling storage logistics equipment receives and responds to the recycling storage logistics equipment dispatching instruction, and transports the corresponding recycling materials to the corresponding warehouse.

[0254] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A renewable electricity material recovery system, characterized in that: include: Dismantling dispatch center, waste material storage equipment, recycling storage logistics equipment and several dismantling production lines; The dismantling dispatching center is used to: generate matching instructions for loading and unloading waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the locations of each dismantling production line, and issue them to the waste power materials storage equipment; The waste material storage equipment is used to: according to the waste power material outbound loading instruction, take out the corresponding type of waste power materials from the set warehouse and transport them to the corresponding disassembly production line; The disassembly production line is used to: automatically trigger the disassembly process when detecting that the corresponding waste power materials are in place, generate disassembled materials and feed back its own disassembly progress to the disassembly dispatching center; The disassembly scheduling center is further used to: after the disassembly production line completes the disassembly process, generate a recycling storage logistics equipment scheduling instruction according to the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location; The recycling storage logistics equipment is used to: receive and respond to the recycling storage logistics equipment scheduling instructions, and transport the corresponding recycled materials to the corresponding warehouse.

2. The renewable electricity material recovery system according to claim 1, characterized in that: The dismantling dispatching center is also used to: predict the amount and type of waste power materials generated in the future based on historical waste power material dismantling data and pre-trained machine learning algorithms, and then optimize the waste power material dismantling plan.

3. The renewable electricity material recovery system according to claim 1, characterized in that: The process of the disassembly dispatching center generating a dispatching instruction for recycling storage logistics equipment includes: When the disassembly production line completes the disassembly process, it identifies the packing volume of the disassembled materials, retrieves the warehouse location that matches the disassembled materials, and retrieves the current status information of the storage and logistics equipment; Determine the quantity of recycled storage and logistics equipment based on the packing volume of disassembled materials and the conveying capacity of recycled storage and logistics equipment; Prioritize the idle recycling storage and logistics equipment, and then select all recycling storage and logistics equipment for this transportation task according to the closest end time to the previous transportation task; With the shortest transportation path as the goal, combined with the packing starting point of the disassembled materials and the warehouse location that matches the disassembled materials, the recycling storage logistics equipment scheduling instructions are generated.

4. The renewable electricity material recovery system according to claim 1, characterized in that: In each of the dismantling production lines, automatic loading and unloading target detection is also performed on the corresponding waste power materials, and the process is as follows: Obtain the number of the current process of the waste electrical materials and the corresponding process actual image set; According to the number of the current process, a target detection model corresponding to the current process is retrieved from the set power material equipment model library, and the target detection model is used to identify the actual image set of the process to identify whether the actual image set of the current process and the standard image set of the process are consistent with each other; Among them, each process in the power material equipment model library is set to be associated with the target detection model through the process number; The target detection model corresponding to each process is trained using the process standard image set corresponding to each process of waste electrical materials.

5. The renewable electricity material recovery system according to claim 4, characterized in that: The target detection model is a CrossKD target detection model.

6. The renewable electricity material recovery system according to claim 5, characterized in that: The process standard image set corresponding to each process includes the standard initial state image of the waste electrical materials in each process and the standard component images that need to be assembled or disassembled in each process.

7. The renewable electricity material recovery system according to claim 6, characterized in that: The standard initial state image and the standard component image both include a first sample image with a configuration label and a second sample image without a configuration label; using the process standard image set of each process to train the CrossKD target detection model of each process, including: A process standard image set of a process is arbitrarily selected, and based on the first sample image, the teacher model is used to perform knowledge distillation training on the student model, and the trained student model is determined as the initial target detection model of each process; Based on the second sample image, using the teacher model to perform knowledge distillation training processing on the initial target detection model of each process, and determining the trained initial target detection model as the CrossKD target detection model of each process; The teacher model is a model pre-trained based on the first sample image, and the knowledge distillation training process includes feature layer distillation training and logic layer distillation training.

8. The renewable electricity material recovery system according to claim 7, characterized in that: The teacher model includes a first feature extraction module and a first segmentation module; The student model includes a second feature extraction module, and based on the first sample image, performs knowledge distillation training processing on the student model using the teacher model, including: Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain the first image feature output by the teacher model and the second image feature output by the student model; Inputting the first image feature into the first segmentation module to obtain a first target detection result output by the teacher model; Inputting the second image feature into the first segmentation module to obtain a second object detection result output by the teacher model with respect to the student model; Calculating a first loss between the first image feature and the second image feature with respect to feature layer distillation training, and calculating a second loss between the first object detection result and the second object detection result with respect to logic layer distillation training; Inversely update the model parameters of the student model based on the first loss and the second loss.

9. The renewable electricity material recovery system according to claim 8, characterized in that: Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain the first image feature output by the teacher model and the second image feature output by the student model, including: Inputting the first sample image into the first feature extraction module and the second feature extraction module respectively to obtain a first feature map output by the first feature extraction module and a second feature map output by the second feature extraction module; Normalizing the pixel values ​​in the first feature map and the second feature map respectively to obtain the first image feature and the second image feature.

10. The renewable electricity material recovery system according to claim 8, characterized in that: In the process of reversely updating the model parameters of the student model based on the first loss and the second loss, the method further includes: Determine whether the first loss and the second loss have reached convergence; If so, it is determined that the training of the student model is completed, and the trained student model is determined as the initial target detection model.

11. The renewable electricity material recovery system according to claim 4, characterized in that: The actual process image set of the current process is consistent with the process standard image set: The actual parts that need to be assembled or disassembled in the current process identified are consistent with the standard parts images that need to be assembled or disassembled in the current process, and the actual initial state image of the waste electrical materials in the current process is consistent with the standard initial state of the current process; The actual process image set of the current process does not match the process standard image set: The actual parts identified to be assembled or disassembled in the current process are inconsistent with the standard parts images to be assembled or disassembled in the current process, and / or the actual initial state image of the waste electrical materials in the current process is inconsistent with the standard initial state of the current process.

12. The renewable electricity material recovery system according to claim 1, characterized in that: In the disassembly production line, when the waste power material is a waste transformer, the corresponding disassembly process is: 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.

13. The renewable electricity material recovery system according to claim 12, characterized in that: In the disassembly production line, when the waste power material is a waste transformer, in the corresponding disassembly process, the upper surface of the transformer is segmented using a long short-term memory artificial neural network. Specifically, first, the depth information is converted into three channel information of disparity, surface normal and height; Then, the recurrent event network is used to extract contextual information on the transformer surface in different directions and propagate bidirectionally in both directions. At the same time, for the RGB channel information in the transformer 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 at all levels to the same resolution and cascade them. Finally, a recurrent event network is used to obtain contextual information.

14. The renewable electricity material recovery system according to claim 12, characterized in that: In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, only the point cloud of the upper surface of the transformer is retained through the connected component labeling algorithm and noise removal 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; and 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.

15. The renewable electricity material recovery system according to claim 12, characterized in that: In the disassembly production line, when the waste power material is a waste transformer, in the corresponding disassembly process, 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 pre-formed 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.

16. The renewable electricity material recovery system according to claim 12, characterized in that: In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, 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, and for bolts that are not detected due to external interference, the candidate positions are regarded as the bolt positions, and for positions where the distance between the candidate positions and the previously detected positions is greater than a set threshold, they are regarded as false detections, and 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.

17. The renewable electricity material recovery system according to claim 12, characterized in that: In the disassembly production line, when the waste power materials are waste transformers, in the corresponding disassembly process, 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, and the three-dimensional coordinates of the midpoint of the magnetic suction position are obtained; 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; 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 to guide the robot end clamping tooling to clamp the upper yoke for disassembly.

18. The renewable electricity material recovery system according to claim 1, characterized in that: When the waste power materials are waste transformers, the corresponding disassembly production line includes a conveyor line, and a weighing mechanism and a code scanning mechanism arranged on the conveyor line; A screw removal robot arm, an oil pumping robot arm, a power-assisted cantilever crane, a coil disassembly robot arm, a sorting robot arm and a three-dimensional warehouse are arranged near the conveyor line; the screw removal robot arm, the oil pumping robot arm, the coil disassembly robot arm and the sorting robot arm are all provided with rotating parts through adjusting mechanisms and connecting rods, and the rotating parts are provided with connecting heads for connecting various tools.

19. The renewable electricity material recovery system according to claim 1, characterized in that: When the waste power materials are waste cables, the corresponding disassembly production line includes: an automatic feeding device, an automatic picking and grabbing device, an automatic disassembly device and a crushing and screening device arranged in sequence along the main transmission line; The automatic feeding device is used to automatically transport the waste electric meters to the entrance end of the automatic picking and grabbing device; The automatic picking and grabbing device is used to automatically pick up a single waste electric meter and transfer it to the workstation where the automatic disassembly device is located; The automatic disassembly device is used to automatically disassemble the waste electric meter on its workstation into set parts; The crushing and screening device is used to crush the disassembled parts and separate and screen the recyclable materials according to their properties and store them in corresponding material receiving boxes.

20. The renewable electricity material recovery system according to claim 1, characterized in that: When the waste power materials are waste electric energy meters, the corresponding disassembly production line includes: an automatic cable feeding device, a primary stripping device, a wire core separation and feeding device, and an automatic crushing and sorting device arranged in sequence along the main transmission line; a secondary stripping device and an automatic packaging device are arranged in sequence on the branch transmission line connected to the rear end of the wire core separation and feeding device; The automatic cable feeding device is used to separate the piled cables into individual cables by step transmission, and to sequentially transmit each individual cable to the primary stripping device; The primary stripping device is used to adaptively clamp the cable according to the cable diameter and perform primary stripping to obtain a primary stripping product; the primary stripping product includes inner and outer sheaths, fillers, steel armor and wire cores; The wire core separation and feeding device is used to transmit and feed the primary stripping product in layers; The secondary stripping device is used to perform secondary stripping on the wire core to obtain the metal wire core and transmit it to the automatic packaging device for packaging; The automatic crushing and sorting device is used for automatically crushing and sorting the inner and outer sheaths, fillers and steel armor.

21. The renewable electricity material recovery system according to claim 1, characterized in that: When the waste power materials are composite insulators, the corresponding disassembly production line includes: a feeding conveyor line for transporting the insulators to be recycled; The insulator fixing mechanism receives the insulators from the feeding conveyor line and fixes the insulators to be recovered with the cooperation of the feeding and handling mechanism; The cutting manipulator uses the cutting tool carried at the end to cut and separate the shed part along the axial direction of the insulator to be recovered in cooperation with the insulator fixing mechanism, and then cuts and separates the glass fiber rod and metal head in the insulator after the shed is separated; The sorting conveyor line uses a sorting robot to sort the separated sheds, fiberglass rods and metal heads into corresponding recycling containers.

22. A control method for a renewable electricity material recovery system according to any one of claims 1 to 21, characterized in that: include: The dismantling dispatch center generates matching instructions for the delivery and loading of waste power materials according to the dismantling plan of waste power materials, combined with the storage information of waste power materials and the location of each dismantling production line, and sends them to the waste material storage equipment; The waste material storage equipment takes out the corresponding type of waste power materials from the set warehouse and transports them to the corresponding disassembly production line according to the waste power material outbound loading instructions; When the corresponding waste power materials are detected to be in place, the disassembly process of the disassembly production line is automatically triggered, the disassembly materials are generated and the disassembly progress is fed back to the disassembly dispatch center; When the disassembly production line completes the disassembly process, the disassembly dispatch center also generates a dispatch instruction for the recycling storage logistics equipment based on the current status information of the recycling storage logistics equipment, the type of disassembled materials and their warehouse location; The recycling storage logistics equipment receives and responds to the recycling storage logistics equipment dispatching instruction, and transports the corresponding recycling materials to the corresponding warehouse.

Images