Intelligent disassembly and resource recycling method and system for decommissioned electric energy metering equipment

Abstract

The application discloses a method and system for intelligent disassembly and resource recycling of scrapped electric energy metering equipment. The method comprises the following steps: acquiring basic information data, physical parameter data and environmental cost data of the scrapped electric energy metering equipment to be processed, inputting the data into a pre-constructed multi-dimensional heterogeneous data fusion evaluation model based on a multi-head attention mechanism, outputting a device-level comprehensive value index DVI, a component-level recycling value label and a process difficulty coefficient, and generating a device disassembly priority list and a to-be-disassembled resource recycling priority list based on the output results. The multi-dimensional heterogeneous data fusion evaluation model based on the multi-head attention mechanism can accurately identify high-value devices and components, generate a scientific disassembly priority list, avoid inefficient fragmentation of high-value resources, and improve overall recycling efficiency.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence, specifically relating to a method and system for intelligent dismantling and resource recycling of scrapped electrical energy metering equipment. Background Technology

[0002] Scrapped electricity metering equipment contains precious metals such as gold, silver, and platinum group metals, as well as harmful substances such as lead, mercury, and cadmium. Improper handling not only wastes resources but also causes serious environmental pollution. Currently, the dismantling and recycling of scrapped electricity metering equipment mainly relies on manual dismantling or semi-mechanized crushing and sorting, which has the following technical drawbacks: First, dismantling decisions rely on manual experience and lack quantitative assessment of the intrinsic value of equipment, resulting in high-value equipment being inefficiently crushed while low-value equipment occupies resources for meticulous dismantling. Second, key resource identification methods are limited, either using machine vision to obtain only appearance information or using X-rays to obtain only elemental information, failing to simultaneously obtain the spatial location and material composition of resources, leading to significant blindness in dismantling operations. Third, the dismantling process is fixed and rigid, unable to dynamically adjust the dismantling intensity according to the actual value of the equipment, either sacrificing efficiency to protect components or damaging high-value components in pursuit of efficiency. Fourth, data from each processing stage is fragmented, failing to form a traceable and auditable closed-loop information system, which not only affects the accurate calculation of recycling benefits but also makes it difficult to accumulate data for process optimization.

[0003] Therefore, how to achieve intelligent assessment, accurate identification, adaptive dismantling, and full-process traceability of scrapped energy metering equipment has become a technical challenge that urgently needs to be solved in this field. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an intelligent dismantling and resource recycling method and system for end-of-life electricity metering equipment. The system acquires basic information data, physical parameter data, and environmental cost data of the end-of-life electricity metering equipment to be processed, and inputs this data into a pre-constructed multi-dimensional heterogeneous data fusion evaluation model based on a multi-head attention mechanism. The model outputs an equipment-level comprehensive value index (DVI), a component-level recycling value label, and a process difficulty coefficient. Based on the output results, it generates a priority list for equipment dismantling and a priority list for resource recycling. By constructing a multi-dimensional heterogeneous data fusion evaluation model based on a multi-head attention mechanism, this invention can accurately identify high-value equipment and components, generate a scientific dismantling priority list, avoid inefficient fragmentation of high-value resources, and improve overall recycling efficiency.

[0005] The first aspect of this application discloses a method for intelligent dismantling and resource recycling of scrapped energy metering equipment, which adopts the following technical solution: The system acquires basic information data, physical parameter data, and environmental cost data of the waste electrical energy metering equipment to be processed. After preprocessing into initial fusion feature quantities, it constructs a multi-branch task learning architecture to extract features and conduct joint training, and outputs the equipment-level comprehensive value index, component-level recycling value label, and process difficulty coefficient. A two-stage component identification process is performed on the device to be processed. In the first stage, the key component ROI region of the device to be processed is identified through a machine vision model. In the second stage, the elemental composition data and three-dimensional morphology analysis of the ROI region are performed to generate a voxel grid of chemical element distribution. For the element distribution voxel grid, three-dimensional connected regions are obtained based on element concentration. For each connected region, accessibility analysis and force direction effectiveness calculation are performed to output recommended dismantling entry points. Combined with the equipment-level comprehensive value index, component-level recycling labels and process difficulty coefficients, a differentiated adaptive dismantling process is executed for the scrapped energy metering equipment to be processed. Using the equipment identification code of the electrical energy metering equipment to be disposed of as an index, a full life cycle traceability information is established; based on the historical data in the full life cycle traceability information, the multi-branch task learning architecture and the machine vision model are incrementally optimized periodically.

[0006] Furthermore, the environmental cost data includes: The confidence level of the types of hazardous substances contained in the waste electrical energy metering equipment to be processed; the confidence level of each type of hazardous substance is expressed as the proportion of equipment in which the type of hazardous substance was detected in historical random inspections of similar equipment.

[0007] Furthermore, the environmental cost data also includes: The estimated carbon emissions during the dismantling process of the waste electrical energy metering equipment to be processed are calculated using a weighted empirical formula based on equipment weight, the complexity of the equipment's pre-designed internal structure, and the carbon emissions of hazardous substances. Among them, the carbon emission of hazardous substances is obtained by summing the individual carbon emissions of all types of hazardous substances in the waste electrical energy metering equipment to be treated. The individual carbon emission of each type of hazardous substance is the product of the confidence level of the hazardous substance and the corresponding additional carbon emission coefficient per unit concentration.

[0008] Furthermore, the second stage includes: Line scanning was performed on the ROI region to obtain three-dimensional topographic data, which was then registered with the CAD model of the discarded energy metering equipment to be processed. The ROI region is scanned by a grid to output chemical element composition data, i.e., the intensity count of chemical element characteristic peaks at each scan point; the chemical element composition data is mapped to the registered three-dimensional space to construct the corresponding chemical element distribution voxel grid; each voxel in each grid contains the chemical element concentration vector at that location.

[0009] Furthermore, a voxel grid of chemical element distribution is constructed, including: For each scan point, the characteristic peak area of ​​each chemical element is obtained by integrating the characteristic peak range data. The characteristic peak area is substituted into the linear regression model of spectral intensity and elemental concentration to obtain the concentration estimate of the corresponding element. Based on the element concentration estimates and their three-dimensional coordinates at each scanning point, a three-dimensional voxel grid is constructed by filling the difference, resulting in a voxel grid of chemical element distribution.

[0010] Furthermore, the reachability analysis includes: Select the target element in the voxel grid of chemical element distribution, and perform threshold segmentation on the element concentration of the target element; perform three-dimensional connected component analysis on the segmented binary voxel set to extract the convex hull contour point set of the key components of the equipment to be processed; A set of candidate points is uniformly generated on the convex hull surface at a preset sampling interval. For each candidate point, inverse kinematics solution and collision detection are performed based on the kinematic model of the multi-degree-of-freedom intelligent disassembly robot, and candidate points that satisfy the reachability constraints are retained.

[0011] Furthermore, the calculation of the effectiveness of the force direction includes: Calculate the normal of the connection surface between the contour of the key component and the housing of the device to be processed, calculate the local surface normal at each candidate point, and filter out candidate points where the angle between the local surface normal and the connection surface normal is less than a preset angle threshold.

[0012] Furthermore, the recommended disassembly entry points include: Among the candidate points that satisfy the accessibility constraint and the validity of the force direction, the candidate point that is closest to the current position of the multi-degree-of-freedom intelligent disassembly robot and is located at the edge of the connection surface is selected as the recommended disassembly entry point, and its spatial coordinates and execution direction are recorded.

[0013] The second aspect of this application discloses an intelligent dismantling and resource recycling system for scrapped electricity metering equipment, which implements the intelligent dismantling and resource recycling method described in the first aspect of this application. The system includes: The pre-assessment module is used to acquire basic information data, physical parameter data, and environmental cost data of the waste energy metering equipment to be processed. After preprocessing into initial fusion feature quantities, a multi-branch task learning architecture is constructed to extract features and conduct joint training, outputting the equipment-level comprehensive value index, component-level recycling value label, and process difficulty coefficient. The equipment component analysis module is used to perform two-stage component identification on the equipment to be processed. The first stage identifies the key component ROI region of the equipment to be processed through a machine vision model. The second stage performs elemental composition data and three-dimensional morphology analysis on the ROI region to generate a voxel grid of chemical element distribution. An adaptive dismantling module is used to obtain three-dimensional connected domains based on the element concentration of the element distribution voxel mesh, perform reachability analysis and force direction effectiveness calculation for each connected domain, and output recommended dismantling entry points; combined with the equipment-level comprehensive value index, component-level recycling labels and process difficulty coefficients, a differentiated adaptive dismantling process is executed for the scrapped energy metering equipment to be processed. The full lifecycle management module is used to establish full lifecycle traceability information using the equipment identification code of the electrical energy metering equipment to be processed as an index; and to periodically perform incremental optimization on the multi-branch task learning architecture and the machine vision model based on historical data in the full lifecycle traceability information.

[0014] The beneficial effects of this invention are that, compared with the prior art, 1. This invention constructs a multi-dimensional heterogeneous data fusion evaluation model based on a multi-head attention mechanism. Compared with the traditional extensive sorting that relies on human experience, it can accurately identify high-value equipment and components, generate a scientific dismantling priority list, avoid inefficient crushing of high-value resources, and significantly improve the overall recycling efficiency. 2. This invention adopts a two-layer nested precision identification scheme of "coarse identification + fine identification" to generate a three-dimensional heat map of resource distribution and a dismantling guidance path file. This method overcomes the limitations of a single identification method, achieving both accurate positioning of key resources and taking into account identification efficiency, providing accurate operation guidance for adaptive dismantling. 3. Based on DVI values ​​and component-level recycling value tags, this invention dynamically switches between three dismantling modes: non-destructive priority dismantling, efficiency priority dismantling, and integrated crushing and sorting. This differentiated adaptive dismantling strategy significantly improves dismantling efficiency while maximizing the protection of high-value resources. 4. This invention uses a unique device identification code as its core to construct a full lifecycle traceability archive based on blockchain technology. Data from each stage of evaluation, identification, disassembly, and sorting is stored on the blockchain in real time, ensuring data immutability and end-to-end traceability. Simultaneously, a mechanism for periodically statistically analyzing historical data and generating training datasets for incremental training and parameter optimization of the evaluation and identification models provides data support for subsequent process improvements, enabling the system to possess self-learning and self-optimization capabilities. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the intelligent dismantling and resource recycling method for scrapped electrical energy metering equipment according to the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0017] Please see Figure 1 This embodiment provides a method for intelligent dismantling and resource recycling of scrapped electricity metering equipment, specifically including: Step S1: Obtain basic information data, physical parameter data, and environmental cost data of the waste electrical energy metering equipment to be processed, input them into the pre-constructed multi-dimensional heterogeneous data fusion evaluation model based on multi-head attention mechanism, and output the equipment-level comprehensive value index (DVI), component-level recycling value label, and process difficulty coefficient; and generate a list of equipment dismantling priorities and a list of resources to be dismantled and recycled based on the output results. 1.1: In a further implementation, the following three types of raw data from the scrapped energy metering equipment to be processed are acquired by a data acquisition station deployed at the entrance of the dismantling production line: (1) Basic information data ; Basic information data is obtained by scanning the one-dimensional or two-dimensional barcode on the casing of the waste electrical energy metering equipment to be processed using an industrial barcode reader, or by manually inputting it through a touch screen when the barcode is damaged. The specific fields and data formats are as follows: ,in: This indicates the model number of the electrical energy metering equipment to be scrapped, and is a string type, such as "DDZY102-Z"; This represents the manufacturer code of the electrical energy metering equipment to be disposed of, and is an integer code. The year of manufacture of the electrical metering equipment to be disposed of is indicated by an integer, such as 2020; This indicates the nominal voltage of the electrical energy metering equipment to be disposed of, and is a floating-point number in volts (V), such as 220V. The nominal current of the electrical energy metering equipment to be disposed of is represented as a floating-point number in amperes (A), such as 10. The accuracy class of the electrical energy metering equipment to be scrapped is indicated by an integer, such as 2 representing class 2; (2) Physical parameter data ; The physical parameter data Real-time measurement is achieved through dynamic weighing modules, machine vision inspection modules, and electrical safety testers on the production line; The specific fields and data formats are as follows: ,in: The weight of the electrical energy metering equipment to be disposed of is a floating-point number in kg, obtained from the weighing sensor. The appearance integrity score of the electrical energy metering equipment to be scrapped is represented by an integer and the value range is [0, 10]. It is obtained by querying the corresponding visual inspection results in the database preset form. This indicates the shell material type of the electrical energy metering equipment to be disposed of. It is an integer code, and different integer codes correspond to different preset shell material types. The internal structural complexity of the discarded electrical energy metering equipment to be processed is represented by a floating-point number between 0 and 1, which is obtained by querying the edge density calculation results of the corresponding X-ray transmission image in the database preset form; This represents a vector consisting of the measured resistance values ​​of each functional module in the discarded energy metering equipment, obtained through probe contact measurement. It is a floating-point vector, with units of... ; This represents a vector composed of the measured capacitance values ​​of each functional module of the discarded energy metering equipment to be processed. It is a floating-point vector, with units of [missing information]. ; (3) Environmental cost data ; The environmental cost data The equipment casing and key components can be sampled using a handheld XRF analyzer, or the manufacturer's environmental materials database can be consulted. Specific fields and data format: , in, The probability of lead content in the scrapped energy metering equipment to be processed is represented by a floating-point number between 0 and 1, indicating the confidence level of lead detection. The confidence level is expressed as the historical detection rate. For example, if 85 out of 100 similar equipment that were previously sampled and tested as the scrapped energy metering equipment to be processed detected lead, then the probability of lead content in the scrapped energy metering equipment to be processed is 0.85.

[0018] This represents the probability of mercury content in the discarded electrical energy metering equipment to be processed, and is a floating-point number between 0 and 1; similar to the probability of lead content, it is obtained through historical detection rates. The probability of brominated flame retardants being present in the discarded electrical energy metering equipment to be processed is represented by a floating-point number between 0 and 1; similar to the probability of lead content, it is obtained through historical detection rates. This represents the estimated carbon emissions during the dismantling process of the scrapped energy metering equipment. It is a floating-point number, expressed in kgCO2, and its value is estimated based on the equipment's weight and materials using an empirical formula. For specific calculations, the following empirical formula can be used as a reference:

[0019] in, and These are the weight of the discarded electrical energy metering equipment to be processed and the preset internal structure complexity of the equipment in the preset form of the database; and The confidence levels for each preset hazardous substance type are ( ). For all in the equipment (Index of hazardous substance types) and the additional carbon emission factor per unit concentration corresponding to each hazardous substance type; , and The weighting coefficients for each item can be obtained by fitting the energy consumption and carbon emission data of multiple devices measured and dismantled in historical data.

[0020] 1.2: In a further embodiment, the basic information data, physical parameter data, and environmental cost data are normalized, feature extracted, and data aligned.

[0021] Vectorize textual data, such as converting device model numbers into 128-dimensional dense vectors using a pre-trained Word2Vec word embedding model, and converting manufacturer codes into multi-dimensional vectors using one-hot encoding; normalize numerical data, such as processing production years using Min-Max normalization, and processing continuous values ​​such as nominal voltage, current, and weight using Z-score normalization. For parameters such as accuracy level, appearance integrity score, internal structure complexity, and confidence level that are already in the range of [0,1] or finite integers, keep the original values ​​unchanged; All processed basic information data Physical parameter data and environmental cost data The concatenation results in an initial fused feature vector, denoted as . Its dimension is denoted as ; The feature fusion layer is used to fuse the initial feature vectors preprocessed by the data preprocessing layer. Feature-level fusion is performed through a multi-head attention mechanism to generate a multi-dimensional feature vector for the device. In one reference implementation, an 8-head attention mechanism is used to perform deep interactive fusion of the initial features. The specific calculation process is as follows: The initial fused feature vector Expanding to a sequence matrix, the query matrix is ​​obtained through linear transformation. Key matrix and value matrix ; Based on query matrix Key matrix and value matrix The attention weights of each attention head are calculated, and the outputs of multiple attention heads are concatenated and subjected to a linear transformation to obtain the fused features, i.e., the multidimensional feature vector. .

[0022] Evaluate the output layer by multidimensional feature vectors The three parallel fully connected sub-networks are jointly trained using a multi-task learning architecture, as follows: Branch 1: By adding Sigmoid activation to a single neuron fully connected layer, a continuous value between 0 and 1 is output, which is then mapped to the 0-100 range to obtain the device-level comprehensive value index (DVI); the larger the value, the higher the recyclable value per unit mass of the device and the stronger the urgency of dismantling. Branch 2: The probability distribution vectors of each pre-set category label (including high-value precious metals, reusable components, general materials and hazardous substances) are output by a fully connected layer and softmax. The final label is the category corresponding to the highest probability. Branch 3: Obtained through a single neuron fully connected layer with ReLU activation, output range is [0,1]. The higher the process difficulty coefficient, the higher the disassembly difficulty. A process difficulty coefficient less than or equal to 0.3 indicates easy disassembly (such as screw connection), and a process difficulty coefficient greater than 0.6 indicates difficult disassembly (such as strong adhesive bonding or corrosion). Total loss function Represented as: ; , and These are the mean square error between predicted DVI and actual DVI, the standard cross-entropy loss of component category prediction, and the mean square error between predicted process difficulty coefficient and actual disassembly time (normalized). Based on the multi-task learning architecture described above, the following decision variables are output: Equipment-level Comprehensive Value Index (DVI), Component-level Recycling Value Label, and Process Difficulty Coefficient. A priority list for equipment dismantling and a priority list for recycling resources to be dismantled are then generated based on the output decision variables. The equipment dismantling priority list is an execution sequence for the scrapped energy metering equipment to be processed. It is a list of equipment IDs arranged in descending order of DVI value, with equipment having higher DVI values ​​entering the production line first. The priority list for recycling resources to be dismantled is a list of components arranged according to value tag priority (high value > reusable > general > hazardous); The difficulty level of the process is used in the subsequent selection of disassembly mode.

[0023] Step S2: Based on the priority list of resources to be dismantled and recycled, implement a two-stage component accuracy identification scheme for the scrapped electricity metering equipment to be processed.

[0024] The two-stage component precision identification scheme performs coarse identification through a machine vision fast scanning unit to initially locate the region of interest; then, a high-precision multispectral analysis unit performs fine identification on the initially located region of interest to obtain the corresponding elemental composition data and three-dimensional morphology data; the data from the two identifications are spatially registered and feature fused to generate a three-dimensional heat map of resource distribution and a disassembly guidance path file containing the three-dimensional coordinates, contours, materials of key components and recommended disassembly entry points.

[0025] 2.1: In a further implementation, the first stage performs coarse identification; The overall RGB image and infrared thermal image of the waste electrical energy metering equipment to be processed are input into a fast recognition neural network model pre-trained based on the YOLOv8s architecture. The output includes bounding boxes and confidence scores of six key components: circuit boards, transformers, capacitors, LCD screens, terminal blocks, and batteries. The bounding box detection results of each type of key component with a confidence score higher than the threshold are selected to generate a set of coordinate points corresponding to the preliminary location of the region of interest (ROI).

[0026] 2.2: In a further implementation, the second stage performs fine recognition based on the coarse recognition output; including the following steps; (1) The ROI region is scanned by a three-dimensional laser contour sensor to obtain the three-dimensional topography data corresponding to the ROI region, that is, the point cloud data containing three-dimensional coordinates.

[0027] The 3D topographic data of the ROI area is registered with the CAD model of the decommissioned energy metering equipment to be processed. In one reference implementation, ICP registration (Iterative Closest Point Registration) can be used.

[0028] (2) The ROI region obtained in 2.1 was scanned using an X-ray fluorescence spectrometer (XRF) to output the chemical element composition data of the ROI region, i.e., the intensity count of the characteristic peaks of the chemical elements at each scan point; the chemical element composition data of the ROI region was mapped to the registered three-dimensional space to construct the corresponding chemical element distribution voxel grid, and each voxel in each grid contained the chemical element concentration vector at that location. Specifically: For each scan point, the concentration estimate of each chemical element is calculated based on the spectral data; Integrate the data within the characteristic peak range of each element to obtain the peak area (spectral intensity) for each element. First, a linear regression model representing the relationship between spectral intensity and elemental concentration is constructed using standard samples. Then, the peak area (spectral intensity) corresponding to each element is substituted into the fitted linear regression model to obtain the concentration estimate of the corresponding element. A three-dimensional voxel mesh (voxel size 1mm×1mm×1mm) is constructed, and the element concentration vector of each voxel is filled by interpolation; Based on the element distribution voxel grid, different element concentrations are visualized and rendered to generate a three-dimensional heat map of the resource distribution of the waste energy metering equipment to be processed. 2.3: Select the target element in the obtained element distribution voxel grid and perform threshold segmentation on the element concentration. Then, perform three-dimensional connected component analysis on the segmented binary voxel set, extract the bounding box and convex hull contour point set of each connected component, and obtain the contour of the key components of the scrapped energy metering equipment to be processed; specifically: For example, suppose there is a relay inside the device. After the gold contact area is scanned by XRF, a voxel grid (resolution 1 mm³) is constructed, and each voxel stores the concentration of Au element.

[0029] The Au element concentration in the background voxels (plastic, copper pins) is less than 0.01%; the Au element concentration in the contact voxels (gold plating) is between 2% and 5%; thus, a segmentation threshold of 0.1% can be set to obtain a set of binary voxels marked as potentially gold regions; Furthermore, the 26-neighborhood connectivity rule is applied to the binary voxel set to extract all connected components, and the connected component with the most voxels is selected as the gold contact region; the bounding box of this connected component is calculated and the convex hull contour point set is extracted to obtain a set of boundary points. 2.4: Recommended disassembly entry points are obtained through reachability analysis and force direction effectiveness calculation; Candidate points are uniformly sampled on the convex hull surface (with a spacing of 2 mm); for each sampled point, robot inverse kinematics is used to determine whether a solution exists (i.e. whether the reachability constraint is satisfied: whether the end effector of the robotic arm can reach the position without collision); candidate points that satisfy the reachability constraint are retained.

[0030] Calculate the normal of the connection surface between the key component and the shell (obtained through voxel gradient or CAD model), and calculate the local surface normal at each candidate point. Select candidate points whose local surface normal and connection surface normal are less than 30° to ensure that the disassembly force is effectively applied to the connection. Finally, the candidate point that is closest to the robot's current pose, has been successfully inversely solved, and is located at the edge of the connecting surface is selected as the recommended entry point, and its coordinates and execution direction are recorded. The material information is determined to be "copper-plated gold" material by combining the average element concentration vector of the voxels in the connected domain (e.g., high Au, medium to low Cu, etc.) with the preset form in the knowledge base.

[0031] The material information of key components is determined based on the element composition data, and the coordinate point set corresponding to the three-dimensional coordinates and contour of the key components is extracted based on the element distribution voxel mesh.

[0032] Step S3: Based on the equipment dismantling priority list, the equipment-level comprehensive value index (DVI), and the dismantling guidance path file, a multi-degree-of-freedom intelligent dismantling robot executes a differentiated adaptive dismantling process for the scrapped energy metering equipment to be processed.

[0033] The adaptive dismantling process includes a non-destructive priority dismantling mode, an efficiency priority dismantling mode, and an integrated crushing and sorting mode, which are dynamically switched according to the equipment-level comprehensive value index (DVI) value and the component-level recycling value label. The integrated crushing and sorting mode sends the equipment into a multi-stage crushing and intelligent sorting unit. After crushing, it is sorted through a variety of preset sorting methods to obtain various recycled material components of the waste electrical energy metering equipment to be processed.

[0034] The adaptive disassembly process in step S3 dynamically switches the disassembly mode according to the following logic: When the equipment-level comprehensive value index (DVI) is less than the first preset value, the multi-degree-of-freedom intelligent dismantling robot (whose end effector is a precision pneumatic gripper equipped with a six-dimensional force / torque sensor) is driven to directly execute the integrated crushing and sorting mode. When the equipment-level comprehensive value index (DVI) is greater than or equal to the first preset value, each key component is evaluated: if the component-level recycling value label of the key component is the preset first category label and the corresponding process difficulty coefficient is less than or equal to the second preset value, then the multi-degree-of-freedom intelligent dismantling robot is determined to execute the non-destructive priority dismantling mode; if the component-level recycling value label is the preset second category label and the process difficulty coefficient is less than or equal to the second preset value, then the multi-degree-of-freedom intelligent dismantling robot is determined to execute the efficiency priority dismantling mode; otherwise, the multi-degree-of-freedom intelligent dismantling robot is guided to the integrated crushing and sorting mode.

[0035] The non-destructive priority disassembly mode employs an impedance control algorithm to achieve force-guided disassembly, with the control law being: ; in For the desired contact force, The desired location (the recommended entry point from the disassembly guide file). The actual position (calculated from feedback by the joint encoder); Indicates the desired speed. Indicates actual speed. Here is the stiffness matrix; Here is the damping matrix. ; It serves as a feedforward force to compensate for gravity; the force control accuracy is better than 0.1N. Its execution process is as follows: The robot moves to a position 5mm above the entry point; Activate force control mode and feed slowly along the cutting point direction (-Z) at a speed of 1 mm / s; Real-time monitoring of contact force; feed stops when the force reaches the force control threshold (e.g., 5N). Perform a slight oscillation or rotation to loosen the component; The grippers close and grasp the component, then move in the exit direction (+Z) to complete the separation; The multi-stage crushing and intelligent sorting unit in the integrated crushing and sorting mode includes magnetic separation, eddy current separation, density separation and spectral recognition separation.

[0036] Step S4: Using the unique equipment identification code of the waste electrical energy metering equipment to be processed as the core data node, construct a full life cycle traceability file based on blockchain technology, dynamically record and associate the data of the entire process from Step S1 to S3; periodically perform statistical analysis on the historical data in the full life cycle traceability file to generate a training dataset, which is used for incremental training and parameter optimization of the multi-task learning architecture in Step S1 and the model in the machine vision fast scanning unit in Step S2.

[0037] In step S4, the full lifecycle traceability archive based on blockchain technology adopts the Hyperledger Fabric architecture, which encapsulates the data of each core data node into transactions and writes them into the distributed ledger. The hash value of the data corresponding to the core data node is stored on the chain, and the original data is stored in the distributed file system. The data upload process is as follows: After each workstation completes its operation, it saves the raw data to the distributed ledger; calculates the SHA-256 hash value of the data; constructs a transaction and submits it to the blockchain nodes; after the nodes reach a consensus, the transaction is written into the block and the transaction ID is returned. The data recorded in the full lifecycle traceability archive includes: The detailed report consists of the equipment-level comprehensive value index (DVI), component-level recycling value label, process difficulty coefficient, equipment dismantling priority list, and resource recycling priority list to be dismantled output in step S1. The resource distribution three-dimensional heat map, disassembly guide path file, and overall RGB image, infrared thermal image, elemental composition data, and three-dimensional morphology data of the waste energy metering equipment to be processed are output in step S2. In step S3, the operation logs and mode switching records of the multi-degree-of-freedom intelligent dismantling robot, as well as the key components and various recycled material components of the dismantled waste electrical energy metering equipment to be processed, are recorded.

[0038] The incremental training in step S4 employs mini-batch stochastic gradient descent, with the loss function being: ; in, and These are different balance coefficients; This represents the predicted value of the equipment-level comprehensive value index output by the model in step S1; This is the true value of the equipment-level comprehensive value index, calculated manually based on the actual recycled value. This represents the component-level recycling value label probability distribution vector output by the model in step S1; The results of manual verification of the component-level recycling value tags; This represents the mean squared error loss function; This represents the cross-entropy loss function; the model parameters after training are version-managed through a blockchain smart contract.

[0039] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0040] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention. It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0041] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0042] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0043] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0044] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for intelligent dismantling and resource recycling of scrapped electricity metering equipment, characterized in that, include: The system acquires basic information data, physical parameter data, and environmental cost data of the waste electrical energy metering equipment to be processed. After preprocessing into initial fusion feature quantities, it constructs a multi-branch task learning architecture to extract features and conduct joint training, and outputs the equipment-level comprehensive value index, component-level recycling value label, and process difficulty coefficient. A two-stage component identification process is performed on the device to be processed; the first stage identifies the ROI regions of key components of the device to be processed using a machine vision model; The second stage involves elemental composition data and three-dimensional morphology analysis of the ROI region to generate a voxel grid of chemical element distribution. For the element distribution voxel grid, three-dimensional connected regions are obtained based on element concentration. For each connected region, accessibility analysis and force direction effectiveness calculation are performed to output recommended dismantling entry points. Combined with the equipment-level comprehensive value index, component-level recycling labels and process difficulty coefficients, a differentiated adaptive dismantling process is executed for the scrapped energy metering equipment to be processed. Using the equipment identification code of the electrical energy metering equipment to be disposed of as an index, a full life cycle traceability information is established; based on the historical data in the full life cycle traceability information, the multi-branch task learning architecture and the machine vision model are incrementally optimized periodically.

2. The intelligent dismantling and resource recycling method for scrapped energy metering equipment according to claim 1, characterized in that, The environmental cost data includes: The confidence level of the types of hazardous substances contained in the waste electrical energy metering equipment to be processed; the confidence level of each type of hazardous substance is expressed as the proportion of equipment in which the type of hazardous substance was detected in historical random inspections of similar equipment.

3. The intelligent dismantling and resource recycling method for scrapped energy metering equipment according to claim 1, characterized in that, The environmental cost data also includes: The estimated carbon emissions during the dismantling process of the waste electrical energy metering equipment to be processed are calculated using a weighted empirical formula based on equipment weight, the complexity of the equipment's pre-designed internal structure, and the carbon emissions of hazardous substances. Among them, the carbon emission of hazardous substances is obtained by summing the individual carbon emissions of all types of hazardous substances in the waste electrical energy metering equipment to be treated. The individual carbon emission of each type of hazardous substance is the product of the confidence level of the hazardous substance and the corresponding additional carbon emission coefficient per unit concentration.

4. The intelligent dismantling and resource recycling method for scrapped electricity metering equipment according to claim 1, characterized in that, The second phase includes: Line scanning was performed on the ROI region to obtain three-dimensional topographic data, which was then registered with the CAD model of the discarded energy metering equipment to be processed. The ROI region is scanned by a grid to output chemical element composition data, i.e., the intensity count of chemical element characteristic peaks at each scan point; the chemical element composition data is mapped to the registered three-dimensional space to construct the corresponding chemical element distribution voxel grid; each voxel in each grid contains the chemical element concentration vector at that location.

5. The intelligent dismantling and resource recycling method for scrapped energy metering equipment according to claim 4, characterized in that, Constructing a voxel grid of chemical element distribution, including: For each scan point, the characteristic peak area of ​​each chemical element is obtained by integrating the characteristic peak range data. The characteristic peak area is substituted into the linear regression model of spectral intensity and elemental concentration to obtain the concentration estimate of the corresponding element. Based on the element concentration estimates and their three-dimensional coordinates at each scanning point, a three-dimensional voxel grid is constructed by filling the difference, resulting in a voxel grid of chemical element distribution.

6. The intelligent dismantling and resource recycling method for scrapped electricity metering equipment according to claim 1, characterized in that, The reachability analysis includes: Select the target element in the voxel grid of chemical element distribution, and perform threshold segmentation on the element concentration of the target element; perform three-dimensional connected component analysis on the segmented binary voxel set to extract the convex hull contour point set of the key components of the equipment to be processed; A set of candidate points is uniformly generated on the convex hull surface at a preset sampling interval. For each candidate point, inverse kinematics solution and collision detection are performed based on the kinematic model of the multi-degree-of-freedom intelligent disassembly robot, and candidate points that satisfy the reachability constraints are retained.

7. The intelligent dismantling and resource recycling method for scrapped energy metering equipment according to claim 1, characterized in that, The calculation of the effectiveness of the force direction includes: Calculate the normal of the connection surface between the contour of the key component and the housing of the device to be processed, calculate the local surface normal at each candidate point, and filter out candidate points where the angle between the local surface normal and the connection surface normal is less than a preset angle threshold.

8. The intelligent dismantling and resource recycling method for scrapped energy metering equipment according to claim 1, characterized in that, The recommended disassembly entry points include: Among the candidate points that satisfy the accessibility constraint and the validity of the force direction, the candidate point that is closest to the current position of the multi-degree-of-freedom intelligent disassembly robot and is located at the edge of the connection surface is selected as the recommended disassembly entry point, and its spatial coordinates and execution direction are recorded.

9. An intelligent dismantling and resource recycling system for scrapped electrical energy metering equipment, comprising the intelligent dismantling and resource recycling method as described in any one of claims 1-8, characterized in that, The system includes: The pre-assessment module is used to acquire basic information data, physical parameter data, and environmental cost data of the waste energy metering equipment to be processed. After preprocessing into initial fusion feature quantities, a multi-branch task learning architecture is constructed to extract features and conduct joint training, outputting the equipment-level comprehensive value index, component-level recycling value label, and process difficulty coefficient. The equipment component analysis module is used to perform two-stage component identification on the equipment to be processed. The first stage identifies the key component ROI region of the equipment to be processed through a machine vision model. The second stage performs elemental composition data and three-dimensional morphology analysis on the ROI region to generate a voxel grid of chemical element distribution. An adaptive dismantling module is used to obtain three-dimensional connected domains based on the element concentration of the element distribution voxel mesh, perform reachability analysis and force direction effectiveness calculation for each connected domain, and output recommended dismantling entry points; combined with the equipment-level comprehensive value index, component-level recycling labels and process difficulty coefficients, a differentiated adaptive dismantling process is executed for the scrapped energy metering equipment to be processed. The full lifecycle management module is used to establish full lifecycle traceability information using the equipment identification code of the electrical energy metering equipment to be processed as an index; and to periodically perform incremental optimization on the multi-branch task learning architecture and the machine vision model based on historical data in the full lifecycle traceability information.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements the intelligent dismantling and resource recycling method according to any one of claims 1-8.

11. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent dismantling and resource recycling method according to any one of claims 1-8.

Images