Multi-physics field simulation coupled power meter temperature rise fault early warning method and system

Abstract

The application provides a multi-physical field simulation coupling electric energy meter temperature rise fault early warning method and system, relates to the electric energy meter detection technical field, and the method comprises the following steps: building a multi-physical field hierarchy system, and simulating and modeling a target electric energy meter; collecting electric energy meter historical temperature rise fault data, mapping to an electric-thermal three-dimensional coupling simulation model to perform fault prediction identification and fault tracing search; fusing and optimizing the temperature rise fault prediction identification branch and the electric energy meter fault feature tracing branch; based on an electric energy meter temperature rise fault diagnosis device, performing fault diagnosis on real-time operation data of the electric energy meter, outputting electric energy meter temperature rise fault diagnosis results, and performing hierarchical early warning decision. The application solves the technical problems that in the prior art, due to the fact that a single physical field cannot represent the multi-field coupling mechanism and the implicit fault evolution law, the temperature rise fault early warning is lagging, the root cause positioning is difficult, and false positives and false negatives frequently occur, and the temperature rise fault diagnosis accuracy is improved by constructing an electric-thermal three-dimensional coupling simulation model.

Description

Technical Field

[0001] This application relates to the field of electricity meter testing technology, and in particular to a method and system for early warning of temperature rise faults in electricity meters using multi-physics simulation coupling. Background Technology

[0002] With the advancement of smart grid construction, electricity meters, as the core terminal for electricity metering and status sensing, directly affect the accuracy and security of grid metering through their long-term operational reliability. Electricity meters integrate multiple heat sources; during long-term operation under load, loose terminals, drifting contact resistance, harmonic overload, or blockage of heat dissipation channels can all cause abnormal temperature rises, which can lead to meter burnout, inaccurate metering, or even electrical fires in severe cases.

[0003] Most existing methods for early warning of temperature rise faults in electricity meters are based on a single physical field and fail to take into account the coupling mechanism of multiple physical fields. As a result, they cannot accurately describe the evolution of latent faults in electricity meters, leading to a serious lag in temperature rise fault early warning. Often, the alarm is only triggered when the fault has developed to a relatively serious stage. At the same time, it is difficult to trace the root cause of the fault from the apparent temperature anomaly, making root cause location extremely difficult. This makes it difficult to balance false alarms and missed alarms, thus affecting the operation and maintenance efficiency of electricity meters and the safe and stable operation of the power grid.

[0004] In summary, existing technologies suffer from technical problems such as delayed early warning of temperature rise faults, difficulty in root cause location, and frequent false alarms and missed alarms because a single physical field cannot characterize the multi-field coupling mechanism and the evolution law of latent faults, which further affect the operation and maintenance efficiency of electricity meters. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for early warning of temperature rise faults in electricity meters based on multi-physics field simulation coupling, in order to solve the technical problems in the prior art where the single physical field cannot characterize the multi-field coupling mechanism and the evolution law of hidden faults, resulting in delayed early warning of temperature rise faults, difficulty in root cause location, and frequent false alarms and missed alarms, which further affect the operation and maintenance efficiency of electricity meters.

[0006] In view of the above problems, this application provides a method and system for early warning of temperature rise faults in electricity meters based on multiphysics simulation coupling.

[0007] Firstly, this application provides a method for early warning of temperature rise faults in electricity meters based on multi-physics field simulation coupling. This method is implemented through a multi-physics field simulation coupling system for early warning of temperature rise faults in electricity meters. The method includes: constructing a multi-physics field hierarchical system, comprising an electromagnetic field layer, a temperature field layer, and a mechanical field layer; performing simulation modeling of the target electricity meter based on the multi-physics field hierarchical system to construct a three-dimensional coupled electrothermal-mechanical simulation model; and collecting historical temperature data from the electricity meter. The historical temperature rise fault data of the electricity meter is mapped to the electrothermal three-dimensional coupled simulation model for fault prediction, identification, and fault tracing search, generating a temperature rise fault prediction and identification branch and an electricity meter fault feature tracing branch. The temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused and optimized to establish an electricity meter temperature rise fault diagnostic tool. Based on the electricity meter temperature rise fault diagnostic tool, fault diagnosis is performed on the real-time operating data of the electricity meter, outputting the electricity meter temperature rise fault diagnosis result, and a graded early warning decision is made based on the electricity meter temperature rise fault diagnosis result.

[0008] Optionally, key attribute information of the target electricity meter is obtained, including structural dimension parameters, material property parameters, and boundary constraint parameters; electromagnetic boundary conditions, heat source information, and load information are determined based on the actual operating status of the electricity meter; the key attribute information is solved and modeled by sequentially loading the electromagnetic boundary conditions, heat source information, and load information based on the multiphysics hierarchical system to generate an electromagnetic field layer model, a temperature field layer model, and a mechanical field layer model; coupled simulation modeling is performed on the electromagnetic field layer model, the temperature field layer model, and the mechanical field layer model to construct a three-dimensional coupled electrothermal-mechanical simulation model.

[0009] Optionally, the historical temperature rise fault data of the electricity meter is standardized and preprocessed, and the fault type and degree are labeled to obtain electricity meter temperature rise fault data samples; the electricity meter temperature rise fault data samples are mapped to the electrothermal three-dimensional coupled simulation model for fault prediction and identification, generating electricity meter temperature rise fault prediction and identification branches; graph vertices and graph connecting edges are defined, and the target electricity meter is subjected to graph abstraction transformation using the graph vertices and graph connecting edges to construct a multi-physics coupling relationship graph of the electricity meter; based on the historical temperature rise fault data of the electricity meter, fault tracing search is performed on the multi-physics coupling relationship graph of the electricity meter to generate electricity meter fault feature tracing branches.

[0010] Optionally, the sample data of the electricity meter temperature rise fault is mapped to the electrothermal three-dimensional coupled simulation model for sample operating condition inversion to obtain electricity meter fault operating condition inversion data; the electrothermal three-dimensional coupled simulation model is forward-engineered according to the electricity meter fault operating condition inversion data to obtain electricity meter fault simulation fingerprint data; fault features are extracted and associated samples are labeled from the electricity meter fault simulation fingerprint data to form an electricity meter fault fingerprint sample library; the electricity meter fault fingerprint sample library is used to drive the electrothermal three-dimensional coupled simulation model for fault prediction and recognition training to generate an electricity meter temperature rise fault prediction and recognition branch.

[0011] Optionally, the operating condition features of the historical temperature rise fault data of the electricity meter are extracted to obtain the fault operating condition feature data of the electricity meter; the fault operating condition feature data of the electricity meter is projected onto the multi-physics coupling relationship diagram of the electricity meter to mark the fault vertices, thereby obtaining a set of fault vertices of the electricity meter; the vertex with the highest temperature in the set of fault vertices of the electricity meter is selected as the target fault point; based on the target fault point, a fault tracing search is performed in the multi-physics coupling relationship diagram of the electricity meter to generate a fault feature tracing branch of the electricity meter.

[0012] Optionally, taking the target fault point as the starting point, a reverse depth search is performed within the multi-physics coupling relationship graph of the electricity meter to obtain a multi-condition fault root cause tracing chain; based on the multi-condition fault root cause tracing chain, the multi-condition fault root cause vertex is determined; based on the multi-condition fault root cause tracing chain and the multi-condition fault root cause vertex, the graph path is accumulated to obtain a multi-condition fault accumulation tracing graph; the multi-condition fault accumulation tracing graph is updated, classified, and predicted to generate a power meter fault feature tracing branch.

[0013] Optionally, the fault path edges and root cause node weights of the multi-condition fault accumulation tracing graph are updated according to the number of times the operating conditions are accumulated, to obtain a multi-condition fault update tracing graph; the multi-condition fault update tracing graph is then subjected to pattern classification and prediction to generate a power meter fault feature tracing branch.

[0014] Optionally, the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused in parallel to generate an initial temperature rise fault diagnostic tool; the initial temperature rise fault diagnostic tool is tested, evaluated and iteratively optimized to establish an electricity meter temperature rise fault diagnostic tool.

[0015] Optionally, a hierarchical early warning strategy library is constructed, and the hierarchical early warning strategy library is used to perform strategy matching on the temperature rise fault diagnosis results of the electricity meter to determine the hierarchical early warning strategy of the electricity meter; the hierarchical early warning strategy of the electricity meter is used to make hierarchical early warning decisions for the target electricity meter.

[0016] Secondly, this application also provides a multi-physics field simulation-coupled electricity meter temperature rise fault early warning system, used to execute the multi-physics field simulation-coupled electricity meter temperature rise fault early warning method as described in the first aspect. The multi-physics field simulation-coupled electricity meter temperature rise fault early warning system includes: a simulation modeling module for building a multi-physics field hierarchical system, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer; and a simulation modeling module for simulating and modeling the target electricity meter based on the multi-physics field hierarchical system to construct a three-dimensional coupled electrothermal-mechanical simulation model; and a fault identification and tracing module for collecting historical temperature rise fault data of the electricity meter. The historical temperature rise fault data of the electricity meter is mapped to the electrothermal three-dimensional coupled simulation model for fault prediction, identification, and fault tracing search, generating a temperature rise fault prediction and identification branch and an electricity meter fault feature tracing branch. A branch fusion optimization module is used to fuse and optimize the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch to establish an electricity meter temperature rise fault diagnostic tool. A fault diagnosis module is used to perform fault diagnosis on the real-time operating data of the electricity meter based on the electricity meter temperature rise fault diagnostic tool, output the electricity meter temperature rise fault diagnosis result, and make graded early warning decisions based on the electricity meter temperature rise fault diagnosis result.

[0017] One or more technical solutions provided in this application have at least the following beneficial effects:

[0018] By constructing a multi-physics hierarchical system, including an electromagnetic field layer, a temperature field layer, and a mechanical field layer, a simulation model of the target electricity meter is built based on this system, constructing a three-dimensional coupled electrothermal-mechanical simulation model. Historical temperature rise fault data of the electricity meter is collected and mapped to the electrothermal-mechanical three-dimensional coupled simulation model for fault prediction, identification, and fault tracing search, generating a temperature rise fault prediction and identification branch and an electricity meter fault feature tracing branch. These branches are then fused and optimized to establish an electricity meter temperature rise fault diagnostic tool. Based on this tool, fault diagnosis is performed on the real-time operating data of the electricity meter, outputting the temperature rise fault diagnosis results. The results are then used for tiered early warning decisions. In other words, by constructing a three-dimensional coupled electrothermal simulation model and using historical fault data to map and generate a temperature rise fault prediction and identification branch and a fault feature tracing branch, the two branches are then fused and optimized into a unified electricity meter temperature rise fault diagnostic tool. Finally, based on the diagnostic results, a graded early warning decision is made. Even with a small number of real fault samples, it can still achieve early warning of temperature rise faults, identify the root cause of the fault, and output graded handling suggestions, thereby significantly improving the operation and maintenance efficiency of electricity meters.

[0019] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the multiphysics simulation coupling method for early warning of temperature rise faults in electricity meters, as described in this application.

[0022] Figure 2 This is a graph showing the temperature rise and fault probability monitoring of the energy meter in the multiphysics simulation coupled energy meter temperature rise fault early warning method of this application.

[0023] Figure 3 This is a schematic diagram of the structure of the energy meter temperature rise fault early warning system based on multi-physics simulation coupling of this application.

[0024] Figure labeling: Simulation modeling module 11, Fault identification and tracing module 12, Branch fusion optimization module 13, Fault diagnosis module 14. Detailed Implementation

[0025] This application provides a method and system for early warning of temperature rise faults in electricity meters based on multi-physics field simulation coupling. This addresses the technical problems in existing technologies where a single physical field cannot characterize the multi-field coupling mechanism and the evolution law of latent faults, leading to delayed early warning of temperature rise faults, difficulties in root cause location, and frequent false alarms and missed alarms, further impacting the efficiency of electricity meter operation and maintenance. By constructing a three-dimensional coupled electrothermal simulation model and using historical fault data to map and generate a temperature rise fault prediction and identification branch and a fault feature tracing branch, the two branches are then fused and optimized into a unified electricity meter temperature rise fault diagnostic tool. Finally, based on the diagnostic results, a graded early warning decision is made. Even with a small number of real fault samples, it can still achieve advanced early warning of temperature rise faults, clearly identify the root cause of the fault, and output graded handling suggestions, thereby significantly improving the efficiency of electricity meter operation and maintenance.

[0026] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0027] Example 1, please refer to the appendix. Figure 1 This application provides a method for early warning of temperature rise faults in electricity meters based on multi-physics simulation coupling. The method is executed by a multi-physics simulation coupled electricity meter temperature rise fault early warning system, and specifically includes the following steps: A multi-physics field hierarchical system is constructed, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer. Based on the multi-physics field hierarchical system, a simulation model of the target electricity meter is performed, and a three-dimensional coupled electrothermal-mechanical simulation model is constructed.

[0028] Furthermore, this application also includes the following steps: obtaining key attribute information of the target electricity meter, the key attribute information including structural dimension parameters, material property parameters, and boundary constraint parameters; determining electromagnetic boundary conditions, heat source information, and load information based on the actual operating conditions of the electricity meter; solving and modeling the key attribute information by sequentially loading the electromagnetic boundary conditions, heat source information, and load information based on the multiphysics hierarchical system, generating an electromagnetic field layer model, a temperature field layer model, and a mechanical field layer model; and performing coupled simulation modeling on the electromagnetic field layer model, the temperature field layer model, and the mechanical field layer model to construct a three-dimensional coupled electrothermal-mechanical simulation model.

[0029] Specifically, a three-dimensional geometric model of the target electricity meter is obtained, and unnecessary small features are removed. Three solution domains are defined on the same geometric model: the electromagnetic field layer includes all conductive components and the surrounding air; the temperature field layer includes the entire solid components of the electricity meter, the internal air, and the external ambient air; and the mechanical field layer includes only all solid components.

[0030] Complete the module configuration and parameter assignment for the three field layers sequentially. In the electromagnetic field layer, enable the electric and magnetic field coupling solution module and set the frequency domain solution type. Assign conductivity, relative permeability, and relative permittivity to each conductive material. Apply the rated operating voltage (e.g., 220V) to the input terminal of the energy meter, apply the rated operating current (e.g., 5A) to the current loop, and set the output terminal to zero potential. Set an electromagnetic insulation boundary on the outer shell surface. After completing the mesh generation, run the electromagnetic field solver to obtain the Joule heat source density distribution throughout the solution domain and save this data as a spatial field variable. In the temperature field layer, enable the heat conduction, heat convection, and heat radiation modules. Input the thermal conductivity, density, specific heat capacity, and surface emissivity for each solid and fluid material. Set the convective heat transfer coefficient and far-field ambient temperature on the outer shell surface. Establish the connection with the electromagnetic field layer, and map the Joule heat source density calculated by the electromagnetic field layer to the corresponding heating element in the temperature field layer according to the node or element position through the heat source coupling interface. Set the solution type to transient, and the time step to the thermal diffusivity. After completing the thermal mesh generation, run the temperature field solver to obtain the temperature distribution over time. Enable the linear elasticity and thermal expansion modules in the mechanical field layer. Input the elastic modulus, Poisson's ratio, thermal expansion coefficient, and reference temperature for each solid material. Apply zero-displacement constraints at the meter's mounting holes or fixed supports. Establish a connection with the temperature field layer, using the nodal temperature difference (current temperature minus reference temperature) calculated by the temperature field layer as the thermal strain input via the thermal load coupling interface. Set the solution type to static. After completing the mechanical mesh generation, run the mechanical field solver to obtain the displacement, strain, and stress distribution caused by temperature changes.

[0031] The solution settings for the three field layers are saved as a multiphysics hierarchical system file, which explicitly records the coupling mapping relationships and data transfer directions between each layer. The multiphysics hierarchical system establishes independent yet interconnected computational layers according to different types of physical phenomena during simulation modeling. Each layer contains a set of physical governing equations, material properties, boundary conditions, and a solution domain. The core is to clearly define the data transfer interfaces and coupling order between layers, allowing the computational results of different physical fields to be exchanged in a predetermined direction, thus forming an organic, holistic simulation framework.

[0032] To obtain key attribute information of the target electricity meter, structural dimensional parameters of all components of the meter, such as length, width, height, and relative position coordinates of each component, are obtained through 3D scanning or CAD drawings. Material property parameters for each material at different temperatures are collected from material supplier manuals, including electrical conductivity, magnetic permeability, thermal conductivity, specific heat capacity, density, elastic modulus, Poisson's ratio, coefficient of thermal expansion, and surface emissivity. Simultaneously, boundary constraint parameters are determined based on the actual installation method of the electricity meter, such as heat dissipation conditions of the outer surface of the casing in contact with air, and whether the mounting feet are fixed.

[0033] Based on the actual operating conditions of the electricity meter, determine the electromagnetic boundary conditions, heat source information, and load information. Electromagnetic boundary conditions include applying the rated operating voltage at the input terminal, applying the rated operating current in the current loop, and setting a zero-potential reference point. Heat source information needs to be determined based on the heating mechanism; the main heat source is Joule heat generated by current passing through a conductor, and its distribution and magnitude are obtained from subsequent electromagnetic field solutions. If a magnetic core element is present, iron losses also need to be estimated. Load information specifically refers to the thermal load in the mechanical field, which is the thermal expansion strain caused by temperature changes. Its magnitude is equal to the coefficient of thermal expansion multiplied by the temperature difference, and this load will be automatically generated after the temperature field solution is obtained.

[0034] Based on a multiphysics hierarchical system, electromagnetic boundary conditions, heat source information, and load information are loaded sequentially to solve and model key attribute information, generating electromagnetic field layer models, temperature field layer models, and mechanical field layer models respectively. The geometric mesh model and material property library of the electricity meter are retrieved from the multiphysics hierarchical system. In the electromagnetic field solution module, pre-determined electromagnetic boundary conditions, such as applying a 220V power frequency voltage to the electricity meter input, applying a 5A current to the current loop, and setting a zero-potential reference point at the output, are loaded onto the corresponding geometric boundaries. Simultaneously, the conductivity and permeability of all conductive materials within the solution domain are specified. The solution type is set to frequency domain, such as 50Hz. The finite element solver is called to calculate the potential distribution, current density distribution, and Joule heat source density distribution throughout the domain. After the solution is completed, the Joule heat source density field is saved as the core result to the electromagnetic field layer model file, preserving the mesh node coordinates, element connection relationships, and voltage and current density values ​​of each node.

[0035] Retrieve the same geometric model from the hierarchical system, but switch to the thermal solution module. Apply thermal boundary conditions, setting the outer surface of the shell as a natural convection boundary, and specifying a convection heat transfer coefficient such as 5.0 W / (m²). 2 The system sets the ambient temperature to 25°C and the surface radiation to an emissivity of 0.85. Next, it loads the heat source information by reading the Joule heat source density field from the electromagnetic field layer model generated in the previous step through the coupling interface. This field is then mapped to the corresponding heating elements in the temperature field layer model according to their spatial locations, such as manganese-copper shunts, terminals, and transformer windings. Any other known heat sources are also added. The solution type is set to transient, with a given time step and total simulation duration. The thermal solver is then called to calculate the temperature distribution over time under the combined effects of heat conduction, convection, and radiation. After the solution is completed, the temperature values ​​of each node at different times are saved as the temperature field layer model.

[0036] Based on the same geometric model, switch to the structural mechanics solver module. Apply displacement boundary conditions, constraining the displacement at the meter mounting feet or fixing holes to zero, i.e., completely fixed. Apply thermal load information, reading the temperature field from the temperature field layer model generated in the previous step through another coupling interface, and calculate the temperature difference of each node relative to the reference temperature of 25°C. Multiply the temperature difference by the material's coefficient of thermal expansion to obtain the thermal strain, which is applied as the initial strain to each material element. Set the solution type to static. Call the structural mechanics solver to solve the thermoelasticity equations, obtaining the displacement, stress, and strain fields throughout the entire solid domain. After solving, save as a mechanical field layer model.

[0037] In the simulation software, create a new multiphysics coupling project, importing the existing electromagnetic field layer model, temperature field layer model, and mechanical field layer model as three components. Create two core coupling modules: an electrothermal coupling module and a thermomechanical coupling module. The electrothermal coupling module connects the electromagnetic field layer model and the temperature field layer model. It maps the Joule heat source density field output by the electromagnetic field layer model to the volume heat source term of the temperature field layer model. The mapping method uses conserved interpolation based on element integration points; that is, for each element in the temperature field mesh, the heat source density value at its corresponding position in the electromagnetic field mesh is searched, and the average heat generation rate of that element is obtained through shape function interpolation. The electrothermal coupling module also controls the solution order: the electromagnetic field is solved first, and after convergence, the heat source is transferred to the temperature field, and then the temperature field is solved. If the effect of temperature on conductivity needs to be considered, bidirectional coupling is enabled in the electrothermal coupling module; that is, the temperature field feeds back the current temperature to the electromagnetic field, updates the conductivity, and then resolves the electromagnetic field, iterating in this way until stability is achieved.

[0038] The thermo-coupling module connects the temperature field model and the mechanical field model. It converts the nodal temperature values ​​output from the temperature field model into the thermal strain load of the mechanical field model, using the formula: thermal strain = coefficient of thermal expansion × (current temperature - reference temperature). The thermo-coupling module also interpolates the mesh node data of the temperature field onto the mechanical field mesh. The solution order is: first solve the temperature field, then transfer the temperature field results to the mechanical field, and finally solve the mechanical field. If the effect of stress on thermal conductivity needs to be considered, such as the change in contact thermal resistance with pressure, reverse coupling can be enabled.

[0039] The entire coupled solver is configured to operate on a sequential iterative coupling mechanism. At the beginning of each time step, the electrothermal coupling module is executed first to solve the electromagnetic field, transferring the heat source to the temperature field and solving the temperature field. Then, the thermomechanical coupling module is executed, transferring the temperature field to the mechanical field and solving the mechanical field. If bidirectional coupling of conductivity versus temperature is enabled, the temperature must be fed back to the electromagnetic field after the temperature field solution is completed, and the electromagnetic field solution is repeated. This sub-loop continues until the temperature change is less than a set threshold, such as 0.1℃, before resuming the mechanical field solution. All field results are saved synchronously at the end of each time step.

[0040] All modules, data mapping relationships, solver settings, and three field layer models are packaged into an executable multiphysics simulation file, namely the electrothermal-mechanical three-dimensional coupled simulation model. The electrothermal-mechanical three-dimensional coupled simulation model is a digital twin with bidirectional data transfer capability. It contains three physics solvers, two coupling modules, and a solver coordinator, which can completely reproduce the physical chain of current-generated heat, heat-induced temperature rise, temperature rise-induced thermal expansion, and thermal stress. It also supports the inverse effects of temperature on conductivity and deformation on contact resistance.

[0041] By constructing a multi-physics field hierarchy system including electromagnetic field layer, temperature field layer and mechanical field layer, and building a three-dimensional electrothermal-mechanical coupling simulation model based on the system, the complete physical process of current generating Joule heat, Joule heat causing temperature rise, temperature rise leading to thermal expansion and thermal stress is quantitatively described. It also supports the reverse effect of temperature on conductivity, thus realizing a complete characterization of the multi-physics field coupling mechanism inside the electricity meter.

[0042] Historical temperature rise fault data of electricity meters are collected, and the historical temperature rise fault data of electricity meters are mapped to the electrothermal three-dimensional coupled simulation model for fault prediction, identification and fault tracing search, generating temperature rise fault prediction and identification branch and electricity meter fault feature tracing branch.

[0043] Furthermore, this application also includes the following steps: standardizing and preprocessing the historical temperature rise fault data of the electricity meter and labeling the fault type and degree to obtain electricity meter temperature rise fault data samples; mapping the electricity meter temperature rise fault data samples to the electrothermal three-dimensional coupled simulation model for fault prediction and identification, generating electricity meter temperature rise fault prediction and identification branches; defining graph vertices and graph connecting edges, using the graph vertices and graph connecting edges to perform graph abstraction transformation on the target electricity meter, constructing a multi-physics coupling relationship graph of the electricity meter; and performing fault tracing search on the multi-physics coupling relationship graph of the electricity meter based on the historical temperature rise fault data of the electricity meter, generating electricity meter fault feature tracing branches.

[0044] Specifically, historical temperature rise fault data of electricity meters is collected from on-site operation databases or experimental records. This includes voltage, current, ambient temperature, meter casing temperature, terminal temperature, and the fault type and severity confirmed by on-site investigation before and after the fault occurred. The historical temperature rise fault data is standardized and preprocessed, removing obviously erroneous records, filling in missing sensor data using linear interpolation, scaling all voltage, current, and temperature values ​​to the range of 0 to 1 based on their maximum and minimum values, and standardizing the time sampling interval of all data sequences to 1 second. Experts label each data point with fault type based on fault reports, such as normal, loose terminals, shunt malfunction, and heat dissipation blockage, and classify the fault severity into mild, moderate, and severe based on percentages, resulting in a structured electricity meter temperature rise fault data sample set. The electricity meter temperature rise fault data sample is a structured dataset formed after preprocessing and labeling; each sample contains input features and output labels.

[0045] Operating parameters are extracted from pre-processed and labeled historical fault data samples and used as boundary conditions for the electrothermal 3D coupled simulation model. For each sample, the simulation model is used to perform operating condition inversion, obtaining fault operating condition inversion data for the electricity meter. Parameter scanning is performed using this fault operating condition inversion data, such as dividing the contact resistance into multiple gradients between normal and fault values. A simulation is run once for each gradient, generating complete temperature and stress field data under that gradient, i.e., fault simulation fingerprint data. Features, such as hotspot temperature, heating rate, and temperature difference, are extracted from the fault simulation fingerprint data and bound to corresponding fault type and severity labels to form a fault fingerprint sample library. A gradient boosting decision tree classifier is trained using the fault fingerprint sample library, taking features as input and labels as output, and iteratively constructing multiple decision trees to minimize classification error. After training, the model parameters are saved, resulting in the electricity meter temperature rise fault prediction and identification branch.

[0046] Key physical entities are extracted as vertices from the 3D geometric model and electrothermal coupling simulation model of the target energy meter. These include all terminals, manganese-copper shunts, voltage transformer cores, current sampling resistors, main control chips, power devices on the circuit board, heat dissipation fins on the inner wall of the casing, and casing mounting feet. Each vertex is recorded with its material, temperature range under normal operating conditions, heat capacity, and spatial coordinates. For any two vertices, if there is a direct heat conduction path or electrical connection between them, a directed edge is established, pointing from the heat source to the heat sink, i.e., from the side with higher heat generation to the side with lower temperature. The initial weight of the edge is set to thermal resistance, obtained through steady-state thermal analysis in the simulation model, i.e., the temperature difference between the two points divided by the heat power flowing through. For electrical connections, contact resistance weights are also recorded. For mechanical contacts, bidirectional edges are established with weights equal to contact thermal resistance. All vertices and edges form a graph, resulting in the multiphysics coupling graph of the energy meter. The multiphysics coupling graph of the energy meter is a weighted directed graph, where vertices represent components, edges represent physical interactions, and weights are thermal resistance or electrical resistance, etc.

[0047] The key temperature values ​​from each historical temperature rise fault data point are projected onto the corresponding vertices of the multiphysics coupling graph, marking all vertices whose temperatures exceed the normal threshold, forming a fault vertex set. The vertex with the highest temperature is selected as the target fault point. Starting from this target fault point, a depth-first search is performed in the reverse direction of the graph, prioritizing edges with high historical frequency during the search. At each search step, the current vertex is recorded until a pre-defined adjustable parameter vertex is encountered, such as a terminal block, solder joint, or heat dissipation hole. This vertex is marked as the root cause vertex, and the path from the target fault point to the root cause vertex is saved as a tracing chain. This process is repeated for all historical fault data, resulting in multiple tracing chains. These tracing chains are superimposed on the multiphysics coupling graph, accumulating the number of times each edge and each vertex is traversed, and weighted according to the operating condition weights. The weights of the edges and the root cause vertex are updated. Cluster analysis is performed on the updated graph, classifying similar path patterns into several typical fault modes. Construct a real-time tracer based on graph matching. Input real-time temperature data, find hotspot vertices, and perform a heuristic backward search on a weighted graph, prioritizing edges with higher weights to quickly locate the root cause vertex and match the most similar fault mode template. Output the tracer chain and root cause. This tracer is the branch for tracing the fault characteristics of electricity meters.

[0048] Furthermore, this application also includes the following steps: mapping the electricity meter temperature rise fault data samples to the electrothermal three-dimensional coupled simulation model for sample operating condition inversion to obtain electricity meter fault operating condition inversion data; performing forward deduction on the electrothermal three-dimensional coupled simulation model according to the electricity meter fault operating condition inversion data to obtain electricity meter fault simulation fingerprint data; extracting fault features and labeling associated samples from the electricity meter fault simulation fingerprint data to form an electricity meter fault fingerprint sample library; driving the electrothermal three-dimensional coupled simulation model to perform fault prediction and recognition training based on the electricity meter fault fingerprint sample library to generate an electricity meter temperature rise fault prediction and recognition branch.

[0049] Specifically, the sample data of the electricity meter temperature rise fault is mapped onto a pre-constructed three-dimensional coupled electrothermal simulation model for sample operating condition inversion. Steady-state operating parameters before the fault are extracted from the electricity meter temperature rise fault data sample and applied as boundary conditions to the three-dimensional coupled electrothermal simulation model, causing the model to reach a stable state and obtaining the temperature field under normal operating conditions. The actual fault temperature value recorded in the electricity meter temperature rise fault data sample is compared with the simulated temperature value under normal operating conditions. Using the parameter optimization solver built into the three-dimensional coupled electrothermal simulation model, unknown fault parameters in the model, such as contact resistance, convective heat transfer coefficient, and heat source power factor, are adjusted in reverse to ensure that the key point temperature output by the model matches the actual fault temperature. This reverse adjustment process is called inversion, and the result is the electricity meter fault operating condition inversion data, such as the contact resistance value of the terminal corresponding to a certain fault, the equivalent convective heat transfer coefficient of a certain area, etc., characterizing the changes in the underlying physical parameters that led to the fault.

[0050] Based on the inversion data of the electricity meter fault conditions, a forward extrapolation is performed on the electrothermal three-dimensional coupled simulation model. The fault parameters obtained from the inversion are fixed as inputs, while maintaining the same voltage, current, and ambient temperature as the sample. The electrothermal three-dimensional coupled simulation model is then restarted to calculate from time zero. Running in transient mode, the complete process from energization to the appearance of fault temperature is simulated. Temperature data over time at all key points throughout the process are recorded, such as the six terminals, the manganese-copper shunt, the transformer core, and the inner wall of the casing, as well as the three-dimensional temperature distribution cloud map and thermal stress distribution at the final steady state, yielding the electricity meter fault simulation fingerprint data. The electricity meter fault simulation fingerprint data is the simulation result data generated by the forward extrapolation, including temperature curves over time at each key point under fault conditions, hot spot temperature values, temperature distribution cloud maps, and thermal stress distribution. Like a fingerprint, this data corresponds one-to-one with a specific fault type and severity, possessing uniqueness and repeatability.

[0051] A feature extraction algorithm is designed to calculate the highest temperature value at each key point, the time taken for the temperature to rise from ambient temperature to the highest temperature, the maximum first derivative during the heating process, the temperature difference between different key points, and the non-uniformity coefficient of the temperature field for each fault condition. Frequency domain features are extracted, such as the dominant frequency amplitude obtained after performing a Fast Fourier Transform on the temperature time series. A feature vector is generated for each fault condition. The feature vector is then bound to the corresponding fault type and severity labels to form a complete fingerprint sample. All feature-label pairs generated from all fault conditions are collected to form a power meter fault fingerprint sample library, containing fingerprint features under different fault modes, covering the complete range from minor anomalies to severe faults.

[0052] Fault prediction and identification training is performed using a three-dimensional coupled electrothermal simulation model driven by a fault fingerprint sample library of electricity meters. The feature vectors from the fault fingerprint sample library are used as input, and the corresponding fault type and severity labels are used as output. A suitable classifier model, such as a gradient boosting decision tree, is selected. The specific architecture consists of 100 regression trees, each with a maximum depth of 4, a learning rate of 0.1, and a subsampling ratio of 0.8. To handle multiple outputs, namely fault type and severity, a multi-output regression strategy is adopted, where each tree simultaneously outputs prediction scores for four categories, which are then converted into probabilities using softmax. All samples are extracted from the fault fingerprint sample library, and the library is randomly divided into training and validation sets in a 7:3 ratio. In addition to the original sensor values, two domain-adaptive features are constructed: one is the rate of temperature change, such as the average slope over the last 10 seconds, used to capture early rapid temperature rises; the other is the maximum temperature difference between the hot spot and the casing, such as the difference between the highest temperature among the six terminals and the lowest temperature of the casing, which is highly sensitive to changes in contact resistance. The initial predicted values ​​of all samples are set as the average probability of each category. Calculate the residual of the current model on the training set, i.e., the difference between the true label and the current predicted probability; use the residual as the new objective to train a regression tree with a maximum depth of 4; the regression tree selects the optimal splitting feature and threshold by minimizing the squared error; add the new tree to the model with a learning rate of 0.1; every 5 trees trained, calculate the multi-class log loss on the validation set; if the loss does not decrease after 3 consecutive validations, the training terminates early. To prevent overfitting, the early termination rounds are set to 10, the minimum number of leaf node samples is 5, and the feature subsampling ratio is 0.8. Physical priors are injected during training. When calculating the splitting gain, for splits that clearly violate physical laws, such as predicting an overload fault when the voltage is below 100V, a penalty factor is applied, multiplying the gain by 0.5. This is achieved through a regularization term in a custom loss function. Different cost matrices are set for different fault types: the cost of misclassifying a moderate fault as a severe fault is set to 1, and the cost of misclassifying a severe fault as a mild fault is set to 5. Cost-sensitive learning is used during training to adjust sample weights. After training, various metrics are calculated on the validation set, including overall accuracy, precision and recall for each fault class, and confusion matrix. Hyperparameters are adjusted based on the validation results, such as increasing the maximum depth from 4 to 6, the learning rate from 0.1 to 0.05, and the number of trees from 100 to 150. The model with the best performance on the validation set is selected as the final prediction branch.

[0053] The electricity meter temperature rise fault prediction and identification branch includes an input preprocessing module, a decision tree integration module, an accumulation and softmax module, and a decision output module. The input preprocessing module receives raw sensor data streams, calculates the temperature change rate and temperature difference, forming an 8-dimensional feature vector. The decision tree integration module performs parallel computation on 100 decision trees, with each tree traversing branches to a leaf node based on its feature values, outputting a score. The accumulation and softmax module multiplies the outputs of all trees by the learning rate and accumulates them to obtain the raw scores for four categories, then converts them into a probability distribution using the softmax function. The decision output module selects the category with the highest probability as the fault type prediction and outputs the severity level of that category. If the maximum probability is below 0.6, the output is uncertain, and manual review is recommended. Upon receiving a new set of real-time data at 1-second intervals, the model performs one forward inference. The inference time is less than 0.5 milliseconds, meeting real-time requirements. The model does not save its internal state; each inference is independent. To adapt to long-term operational drift, a weekly incremental learning approach is designed: newly confirmed fault data from the current week is added to the fingerprint database, and the model is retrained. The electricity meter temperature rise fault prediction and identification branch can receive real-time monitoring data on voltage, current, ambient temperature, and key point temperature, and output in real time whether there is a potential temperature rise fault, as well as the specific type and severity of the fault.

[0054] By using a three-dimensional coupled electrothermal simulation model for inversion and forward deduction, a large amount of physically consistent simulation fingerprint data can be generated from a small amount of historical fault data. Through sample operating condition inversion, the apparent temperature anomaly can be traced back to the changes in underlying physical parameters, effectively distinguishing different types of faults. The electricity meter temperature rise fault prediction and identification branch focuses on fault type and fault root cause.

[0055] Furthermore, this application also includes the following steps: extracting operating condition features from the historical temperature rise fault data of the electricity meter to obtain fault operating condition feature data of the electricity meter; projecting the fault operating condition feature data of the electricity meter onto the multi-physics coupling relationship diagram of the electricity meter to mark fault vertices, thereby obtaining a set of fault vertices of the electricity meter; selecting the vertex with the highest temperature in the set of fault vertices of the electricity meter as the target fault point; and performing a fault tracing search within the multi-physics coupling relationship diagram of the electricity meter based on the target fault point to generate a fault feature tracing branch of the electricity meter.

[0056] Furthermore, this application also includes the following steps: taking the target fault point as the starting point, performing a reverse depth search within the multi-physics coupling relationship graph of the electricity meter to obtain a multi-condition fault root cause tracing chain; determining the multi-condition fault root cause vertex based on the multi-condition fault root cause tracing chain; performing graph path accumulation based on the multi-condition fault root cause tracing chain and the multi-condition fault root cause vertex to obtain a multi-condition fault accumulation tracing graph; updating, classifying, and predicting the multi-condition fault accumulation tracing graph to generate an electricity meter fault feature tracing branch.

[0057] Furthermore, this application also includes the following steps: updating the fault path edges and root cause node weights of the multi-condition fault accumulation tracing graph according to the number of times the operating conditions are accumulated, to obtain a multi-condition fault update tracing graph; performing pattern classification and prediction on the multi-condition fault update tracing graph to generate a power meter fault feature tracing branch.

[0058] Specifically, for each historical fault record in the historical temperature rise fault data of the electricity meter, the effective value of current, effective value of voltage, ambient temperature, temperature change rate within 30 seconds before the fault, and the highest temperature value of each key point recorded by infrared or sensor are extracted at the time of the fault occurrence to obtain the fault condition characteristic data of the electricity meter, which is used to locate the vertices that may be affected by the fault on the graph.

[0059] The fault condition characteristic data of the electricity meter is projected onto the previously constructed multiphysics coupling relationship diagram of the electricity meter. Based on the association between the fault condition characteristics and the vertices, the fault is marked on the corresponding vertex in the multiphysics coupling relationship diagram. For example, if the point with the highest temperature corresponds to vertex two of the terminal block, then that vertex is marked as a fault vertex. If the temperature of multiple vertices exceeds the normal threshold, such as the normal upper limit of 55℃, then all vertices are marked. The marking method is to add an attribute to each vertex: fault flag = yes, and record the measured temperature value of that point. All marked vertices constitute the set of fault vertices of the electricity meter for that record.

[0060] In the set of fault vertices of the electricity meter, the measured temperature value of each vertex is compared, and the vertex with the highest temperature is selected as the target fault point. Starting from the target fault point, a depth-first traversal is performed along the reverse direction of the edges, that is, from the heat sink to the heat source. When constructing the graph, the direction of the edges is from the heat source to the heat sink, that is, from the heat-generating component to the heat-dissipating component. Therefore, the reverse search is to backtrack from the current vertex along all edges pointing to it to the predecessor vertex. At each backtracking step, it is checked whether the predecessor vertex may be the root cause of the fault. If the predecessor vertex is an electrical connection point and its contact resistance is significantly greater than the normal value, it is marked as a candidate root cause; if the predecessor vertex is a heat source, the backtracking continues. The search depth is limited to a maximum of 5 steps to prevent infinite loops. For each historical fault record, a reverse depth search is run once to obtain a path chain, for example, terminal 2 (high temperature), terminal and copper busbar crimping point, copper busbar, shunt output terminal, shunt body. This path is saved as the root cause tracing chain for this operating condition.

[0061] The terminal vertex of the tracing chain is the root cause vertex, which is the vertex containing the initial cause of the entire failure chain. Tracing chains obtained from all historical operating conditions are superimposed onto the same graph to obtain a tracing chain graph. An empty cumulative graph is initialized, with the same structure as the tracing chain graph, but with a counter attached to each edge and each vertex. For each tracing chain, the tracing count of each edge and the occurrence count of each vertex are incremented by 1. After traversing all operating conditions, a multi-operating-condition fault cumulative tracing graph is obtained, where the counter for each edge indicates how many tracing chains traverse that edge, and the counter for each vertex indicates how many failures that vertex is involved in.

[0062] By overlaying the tracing chains obtained under multiple operating conditions onto the same graph, a multi-operating-condition fault cumulative tracing graph is obtained. The multi-operating-condition fault cumulative tracing graph is a graph obtained after multiple accumulations, in which each edge and each vertex has an occurrence counter, reflecting the frequency with which the path is repeatedly traced under different operating conditions.

[0063] Based on the cumulative count and operating condition weights, the weights of each edge and vertex in the graph are recalculated. Considering the varying importance of different operating conditions—for example, faults occurring at high current and high ambient temperatures are more likely to reveal design weaknesses—the operating condition weight factor w is defined as (current / rated current) × (ambient temperature / 25°C). For each tracing chain, the count of its contributing edges and vertices is increased by w instead of 1. The new weights of all edges are equal to the original thermal resistance weights multiplied by (1 + α × edge counter / total number of operating conditions), where α is an adjustment coefficient of 2.0. Similarly, the weight of the root cause node is equal to the initial weight multiplied by (1 + β × vertex counter / total number of operating conditions), where β is 1.5. The resulting updated multi-operating condition fault tracing graph is then obtained, where the thickness of the edges (visually) and the color depth of the nodes are proportional to their weights.

[0064] A graph clustering algorithm is used to group frequent path patterns in the multi-condition fault update tracing graph. Specifically, all root cause vertices are extracted, and each root cause vertex and its downstream subgraphs are treated as a subgraph. The structural similarity between any two subgraphs is calculated using the longest common subpath length. A similarity threshold of 0.7 is set, and subgraphs with similarity higher than the threshold are grouped into the same class. For example, all subgraphs with terminal blocks as root cause vertices and paths containing crimped points are grouped into the terminal loosening class; all subgraphs with ventilation holes as root cause vertices are grouped into the ventilation blockage class; and all subgraphs with solder joints as root cause vertices are grouped into the solder joint aging class. Each class corresponds to a predictive pattern. Finally, the classifier and these pattern templates are packaged together to form the electricity meter fault feature tracing branch.

[0065] The input to the electricity meter fault feature tracing branch is real-time monitoring data, and the output is the root cause vertex location and a complete tracing chain. A set of fault vertices is filtered out using a temperature threshold, with the highest temperature vertex selected as the starting point. Then, a weight-guided shortest path search is performed on the multi-condition fault update tracing graph, i.e., reverse depth-first search, prioritizing edges with higher weights to represent common fault paths, quickly finding the root cause vertex, and finally outputting the tracing chain.

[0066] The temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused and optimized to establish an electricity meter temperature rise fault diagnostic tool.

[0067] Furthermore, this application also includes the following steps: merging the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch in parallel to generate an initial temperature rise fault diagnostic tool; and testing, evaluating, and iteratively optimizing the initial temperature rise fault diagnostic tool to establish an electricity meter temperature rise fault diagnostic tool.

[0068] Specifically, the pre-trained temperature rise fault prediction and identification branch model and the electricity meter fault feature tracing branch model are loaded into the same diagnostic framework. A unified input interface is defined, receiving a data structure containing the RMS voltage value, RMS current value, ambient temperature, temperatures at three measuring points on the casing, temperatures at six measuring points on the terminals, and an optional timestamp. A unified output interface is defined, outputting a diagnostic report containing the fault type and severity labels given by the prediction branch, and the root cause vertex name and tracing path given by the tracing branch. The two branches run independently and in parallel, without waiting for each other. The initial fusion strategy uses direct concatenation, merging the outputs of the prediction and tracing branches into a single JSON message without additional conflict handling or confidence weighting, generating the initial temperature rise fault diagnostic tool. This tool receives a test sample and simultaneously returns the original results from both branches. The initial temperature rise fault diagnostic tool can output both prediction and tracing results simultaneously.

[0069] Prepare an independent test dataset containing at least 100 real-world fault records or high-fidelity simulation-generated fault cases that were not used in any training. Each test sample has a clearly defined real fault type, real severity, and real root cause vertex. Run the initial diagnostic tool to diagnose each test sample, recording the prediction accuracy, root cause localization accuracy, overall consistency rate, false alarm rate, false negative rate, average diagnostic time, and branch conflict rate. Prediction accuracy is the proportion of samples where the fault type output by the predicted branch matches the real type; root cause localization accuracy is the proportion of samples where the root cause vertex output by the trace branch matches the real root cause vertex; overall consistency rate is the proportion of samples where the outputs of two branches simultaneously match the real situation; false alarm rate is the proportion of normal samples (no fault) where the predicted branch outputs a fault; false negative rate is the proportion of fault samples where the predicted branch outputs a normal result; average diagnostic time is the total time from input to output completion; branch conflict rate is the proportion where the predicted branch identifies a certain type of fault, but the root cause vertex output by the trace branch does not match the typical root cause of that type of fault.

[0070] Based on the evaluation results, the areas requiring optimization are determined. This involves adjusting the two branches individually or jointly based on the deficiencies identified in the testing and evaluation. This includes adjusting the model hyperparameters of the prediction branch, increasing or optimizing the graph weight update strategy of the tracing branch, and modifying the fusion logic. For example, if the prediction accuracy is low, below 85%, the diversity of the fingerprint sample database is increased; the hyperparameters of the prediction branch are adjusted, such as increasing the number of decision trees from 100 to 150 and the maximum depth from 4 to 5. If the conflict rate between the two branches is high, exceeding 20%, a fusion decision logic is designed instead of simple splicing. A confidence threshold is set: when the probability of a certain fault type output by the prediction branch is greater than 0.7 and the root cause vertex output by the tracing branch belongs to the typical root cause set of that type of fault, a high-confidence diagnosis is output; otherwise, the output requires manual review.

[0071] After each round of optimization, testing and evaluation are rerun to compare whether the indicators have improved. Typically, 3 to 5 iterations are performed until all key indicators reach preset targets, such as prediction accuracy ≥90%, root cause localization accuracy ≥85%, overall consistency ≥80%, average latency ≤50 milliseconds, and conflict rate ≤10%. Finally, the optimized diagnostic tool is saved as a single executable file. When the final diagnostic tool is working, it starts the calculation of two branches in parallel for each set of real-time data received. The prediction branch outputs the fault type and severity within milliseconds; the tracing branch is slightly slower but still completes within tens of milliseconds. The fusion module waits for both branches to complete and, according to preset fusion rules, outputs the final conclusion directly if the prediction branch probability >0.8 and the tracing root cause matches a typical pattern; if the probability is between 0.6 and 0.8, it outputs a suspected fault with a tracing path; if the probability <0.6, it outputs a manual review result. The final diagnostic result is sent to the operations and maintenance platform via a message queue.

[0072] By employing parallel fusion, the physical mechanism of the feature tracing branch constrains the data-driven bias of the prediction and identification branch. When the prediction branch makes a misjudgment due to environmental interference, the tracing branch provides a physical-level veto. This dual-branch fusion mechanism allows the diagnostic tool to move beyond simple data fitting and incorporate physical topology reasoning. Even with missing or interfered sensor data, it can maintain a certain level of diagnostic capability through graph tracing logic, enhancing its survivability in harsh field environments.

[0073] The energy meter temperature rise fault diagnostic device performs fault diagnosis on the real-time operating data of the energy meter, outputs the energy meter temperature rise fault diagnosis result, and makes graded early warning decisions based on the energy meter temperature rise fault diagnosis result.

[0074] Furthermore, this application also includes the following steps: constructing a hierarchical early warning strategy library, using the hierarchical early warning strategy library to perform strategy matching on the temperature rise fault diagnosis results of the electricity meter, and determining the hierarchical early warning strategy for the electricity meter; and making hierarchical early warning decisions for the target electricity meter through the hierarchical early warning strategy for the electricity meter.

[0075] Specifically, based on the severity, development speed, and operational resource response capabilities of electricity meter temperature rise faults, early warnings are divided into three levels. Level 1 warnings correspond to mild faults with a low temperature rise rate; at this stage, the fault is still in its early stages and will not immediately affect metering safety. The recommended action is to add it to the observation list and schedule monthly planned maintenance. Level 2 warnings correspond to moderate faults or moderate temperature rise rates; at this stage, the fault may develop into a serious fault within several days to a week. The recommended action is to arrange on-site investigation within a week. Level 3 warnings correspond to severe faults or high temperature rise rates, such as a rise of more than 15°C per hour, or the hot spot temperature is approaching the material's tolerance limit, such as exceeding 85°C. At this stage, there is a risk of meter burnout or fire; the recommended action is to immediately disconnect power remotely and dispatch emergency repair personnel for emergency handling. The tiered early warning strategy library is stored in tabular or decision tree form. Each record includes matching conditions, the corresponding warning level, recommended measures, and notification recipients. The tiered early warning strategy library is typically divided into several levels, each corresponding to different notification methods, response time limits, and operational guidelines.

[0076] A tiered early warning strategy library is used to match the diagnostic results of electricity meter temperature rise faults. When the diagnostic tool outputs a diagnostic result, this result is used as a query condition to traverse the rules in the strategy library in priority order. The most urgent rule is matched first, that is, the matching is performed in reverse order from the highest level to the lowest level. Once a match is found, the tiered early warning strategy corresponding to the diagnostic result is determined.

[0077] A tiered early warning strategy is implemented for target electricity meters. Different actuators are invoked based on the warning level: for a Level 1 warning, it is only displayed as a yellow icon on the maintenance platform interface and recorded in the log, without sending an immediate message; for a Level 2 warning, an orange notification is pushed to the on-duty maintenance personnel's mobile app, and a work order is automatically generated and dispatched; for a Level 3 warning, a red alarm is immediately sent to the dispatch center via the 4G module, while the meter's buzzer sounds continuously, and a remote trip command is triggered. All warning actions are timestamped and their results are recorded, forming a closed-loop feedback system.

[0078] Different response strategies are used for faults of varying severity to avoid wasting resources or causing insufficient response due to a one-size-fits-all approach to alarms. For example, minor faults are only logged to avoid disturbing maintenance personnel, while severe faults trigger an emergency response immediately to minimize the risk of fire.

[0079] For example, a certain model of three-phase smart energy meter, with a rated current of 5A and a rated voltage of 220V / 380V, operates under high temperature and high load conditions, and provides early warning for temperature rise faults. The structural dimensions include a meter casing length × width × height of 160mm × 112mm × 58mm, and a current transformer core cross-section of 25mm². 2 The shunt resistor has a resistance of 0.5 mΩ, and the PCB board thickness is 1.6 mm. Material properties include: ABS plastic for the casing (0.17 W / (m·K) thermal conductivity), copper busbar (398 W / (m·K) thermal conductivity), silicon steel core (25 W / (m·K) thermal conductivity, relative permeability 5000), and PCB substrate (0.3 W / (m·K) thermal conductivity). Boundary constraints include an ambient temperature range of 25℃ to 70℃ and a natural convection heat transfer coefficient of 5 to 25 W / (m·K). 2• K), radiative emissivity 0.85. Electromagnetic boundaries include a rated frequency of 50Hz, power factor of 0.8 (lagging), and load current range of 0~6A (120% overload). After standardizing and preprocessing historical fault samples, operating condition inversion and forward extrapolation results are used to construct a multi-physics coupling relationship diagram of the energy meter. The fixed points are various functional components of the energy meter, such as power modules, metering chips, CT cores, shunts, terminals, relays, LCD drivers, etc.; the edges are heat conduction paths, electromagnetic coupling relationships, and mechanical stress transmission paths. The target fault point is the CT core joint, with a temperature of 123.7℃. A reverse deep search is performed: Root cause vertex one is the load current of 6.0A, with a weight of 0.45, which is a direct heat source. The upstream is a surge in user load, with a weight of 0.30; Root cause vertex two is the power factor of 0.70, with a weight of 0.28, indicating an increase in reactive current. The upstream is a high proportion of inductive load, with a weight of 0.20; Root cause vertex three is the ambient temperature of 65℃, with a weight of 0.18, indicating deteriorated heat dissipation. The upstream is high summer temperature / poor ventilation, with a weight of 0.12; Root cause vertex four is the decrease in core permeability, with a weight of 0.09, indicating material aging. The upstream is long-term operation magnetization, with a weight of 0.06. The weights of the multi-condition cumulative traceability graph have been updated. The weight of the path from overload to core saturation has been changed from 0.45 to 0.52; the weight of the path from low power factor to reactive power generation has been changed from 0.28 to 0.31; the weight of the path from high temperature environment to limited heat dissipation has been changed from 0.18 to 0.14; and the weight of the path from core aging to increased magnetic loss has been changed from 0.09 to 0.03. With real-time operating data input, the predicted fault probability at approximately 14 hours is 78.3%, with the predicted type being core saturation and overheating. The characteristic traceability branch output indicates that the dominant root cause path is a combination of overload, low power factor, and high temperature environment factors. (See attached image) Figure 2 The figure shows the temperature rise and fault probability change curves obtained by continuously monitoring the target energy meter for 24 hours using an energy meter temperature rise fault diagnostic tool. Figure 2 The horizontal axis represents the monitoring time, ranging from 0 to 23 hours; the left vertical axis represents the temperature rise of the electricity meter relative to the ambient temperature; and the right vertical axis represents the fault probability output by the diagnostic tool. Solid lines marked with squares represent the temperature rise curve, dashed lines marked with triangles represent the fault probability curve, and gray dotted lines represent the tiered warning threshold boundaries. Figure 2 It can be seen that during the low-load period of 0-5 hours at night, both the temperature rise and the probability of failure are at a low level; as the daytime load and ambient temperature rise for 6-11 hours, both curves rise synchronously; during the peak period of 12-14 hours at noon, the temperature rise reaches about 57℃ and the probability of failure reaches about 88%, triggering a Level IV danger warning; in the afternoon, as the load decreases, the temperature rise and the probability of failure fall synchronously, returning to normal levels in the evening.

[0080] In summary, the multi-physics simulation-coupled method for early warning of temperature rise faults in electricity meters provided in this application has the following beneficial effects: By constructing a multi-physics hierarchical system, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer, a simulation model of the target electricity meter is built based on the multi-physics hierarchical system, constructing a three-dimensional coupled electrothermal-mechanical simulation model; historical temperature rise fault data of the electricity meter is collected, and the historical temperature rise fault data of the electricity meter is mapped to the three-dimensional coupled electrothermal-mechanical simulation model for fault prediction, identification, and fault tracing search, generating a temperature rise fault prediction and identification branch and an electricity meter fault feature tracing branch; the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused and optimized to establish an electricity meter temperature rise fault diagnostic tool; based on the electricity meter temperature rise fault diagnostic tool, fault diagnosis is performed on the real-time operating data of the electricity meter, outputting the electricity meter temperature rise fault diagnosis result, and a graded early warning decision is made based on the electricity meter temperature rise fault diagnosis result. In other words, by constructing a three-dimensional coupled electrothermal simulation model and using historical fault data to map and generate a temperature rise fault prediction and identification branch and a fault feature tracing branch, the two branches are then fused and optimized into a unified electricity meter temperature rise fault diagnostic tool. Finally, based on the diagnostic results, a graded early warning decision is made. Even with a small number of real fault samples, it can still achieve early warning of temperature rise faults, identify the root cause of the fault, and output graded handling suggestions, thereby significantly improving the operation and maintenance efficiency of electricity meters.

[0081] Example 2: Based on the same inventive concept as the multiphysics simulation-coupled electricity meter temperature rise fault early warning method in Example 1, this application also provides a multiphysics simulation-coupled electricity meter temperature rise fault early warning system. Please refer to the appendix. Figure 3 The multiphysics simulation-coupled energy meter temperature rise fault early warning system includes: The simulation modeling module 11 is used to build a multi-physics field hierarchical system, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer. Based on the multi-physics field hierarchical system, the target electricity meter is simulated and modeled to construct a three-dimensional coupled electrothermal-mechanical simulation model. The fault identification and tracing module 12 is used to collect historical temperature rise fault data of the electricity meter, map the historical temperature rise fault data of the electricity meter to the three-dimensional coupled electrothermal-mechanical simulation model for fault prediction identification and fault tracing search, and generate a temperature rise fault prediction identification branch and an electricity meter fault feature tracing branch. The branch fusion optimization module 13 is used to fuse and optimize the temperature rise fault prediction identification branch and the electricity meter fault feature tracing branch to establish an electricity meter temperature rise fault diagnostic tool. The fault diagnosis module 14 is used to perform fault diagnosis on the real-time operating data of the electricity meter based on the electricity meter temperature rise fault diagnostic tool, output the electricity meter temperature rise fault diagnosis result, and make graded early warning decisions based on the electricity meter temperature rise fault diagnosis result.

[0082] Furthermore, the simulation modeling module 11 in the multiphysics simulation-coupled electricity meter temperature rise fault early warning system is also used to: acquire key attribute information of the target electricity meter, including structural dimension parameters, material property parameters, and boundary constraint parameters; determine electromagnetic boundary conditions, heat source information, and load information based on the actual operating conditions of the electricity meter; solve and model the key attribute information by sequentially loading the electromagnetic boundary conditions, heat source information, and load information based on the multiphysics hierarchical system, generating an electromagnetic field layer model, a temperature field layer model, and a mechanical field layer model; and perform coupled simulation modeling on the electromagnetic field layer model, the temperature field layer model, and the mechanical field layer model to construct a three-dimensional coupled electrothermal-mechanical simulation model.

[0083] Furthermore, the fault identification and tracing module 12 in the multi-physics simulation coupled electricity meter temperature rise fault early warning system is also used for: standardizing and preprocessing the historical temperature rise fault data of the electricity meter and labeling the fault type and degree to obtain electricity meter temperature rise fault data samples; mapping the electricity meter temperature rise fault data samples to the electrothermal three-dimensional coupled simulation model for fault prediction and identification, generating electricity meter temperature rise fault prediction and identification branches; defining graph vertices and graph connecting edges, using the graph vertices and graph connecting edges to perform graph abstraction transformation on the target electricity meter, constructing an electricity meter multi-physics coupling relationship graph; and performing fault tracing search on the electricity meter multi-physics coupling relationship graph based on the historical temperature rise fault data of the electricity meter, generating electricity meter fault feature tracing branches.

[0084] Furthermore, the fault identification and tracing module 12 in the multiphysics simulation coupled electricity meter temperature rise fault early warning system is also used for: mapping the electricity meter temperature rise fault data samples to the electrothermal three-dimensional coupled simulation model for sample operating condition inversion to obtain electricity meter fault operating condition inversion data; performing forward deduction on the electrothermal three-dimensional coupled simulation model according to the electricity meter fault operating condition inversion data to obtain electricity meter fault simulation fingerprint data; extracting fault features and labeling associated samples from the electricity meter fault simulation fingerprint data to form an electricity meter fault fingerprint sample library; and driving the electrothermal three-dimensional coupled simulation model to perform fault prediction and identification training based on the electricity meter fault fingerprint sample library to generate an electricity meter temperature rise fault prediction and identification branch.

[0085] Furthermore, the fault identification and tracing module 12 in the multi-physics simulation coupled electricity meter temperature rise fault early warning system is also used for: extracting operating condition features from the historical temperature rise fault data of the electricity meter to obtain electricity meter fault operating condition feature data; projecting the electricity meter fault operating condition feature data onto the electricity meter multi-physics coupling relationship graph to mark fault vertices, obtaining an electricity meter fault vertex set; selecting the vertex with the highest temperature in the electricity meter fault vertex set as the target fault point; and performing fault tracing search within the electricity meter multi-physics coupling relationship graph based on the target fault point to generate an electricity meter fault feature tracing branch.

[0086] Furthermore, the fault identification and tracing module 12 in the multi-physics field simulation coupled electricity meter temperature rise fault early warning system is also used for: taking the target fault point as the starting point, performing a reverse depth search within the multi-physics field coupling relationship graph of the electricity meter to obtain a multi-condition fault root cause tracing chain; determining the multi-condition fault root cause vertex based on the multi-condition fault root cause tracing chain; performing graph path accumulation based on the multi-condition fault root cause tracing chain and the multi-condition fault root cause vertex to obtain a multi-condition fault accumulation tracing graph; updating, classifying, and predicting the multi-condition fault accumulation tracing graph to generate electricity meter fault feature tracing branches.

[0087] Furthermore, the fault identification and tracing module 12 in the multi-physics simulation coupled electricity meter temperature rise fault early warning system is also used to: update the fault path edges and root cause node weights of the multi-condition fault accumulation tracing map according to the number of times the operating conditions are accumulated, to obtain the multi-condition fault update tracing map; and perform pattern classification and prediction on the multi-condition fault update tracing map to generate electricity meter fault feature tracing branches.

[0088] Furthermore, the branch fusion optimization module 13 in the multiphysics simulation coupled electricity meter temperature rise fault early warning system is also used to: fuse the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch in parallel to generate an initial temperature rise fault diagnostic tool; and test, evaluate and iteratively optimize the initial temperature rise fault diagnostic tool to establish an electricity meter temperature rise fault diagnostic tool.

[0089] Furthermore, the fault diagnosis module 14 in the multiphysics simulation coupled electricity meter temperature rise fault early warning system is also used to: construct a hierarchical early warning strategy library, use the hierarchical early warning strategy library to perform strategy matching on the electricity meter temperature rise fault diagnosis results, and determine the electricity meter hierarchical early warning strategy; and make hierarchical early warning decisions for the target electricity meter through the electricity meter hierarchical early warning strategy.

[0090] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1The multi-physics simulation coupled electricity meter temperature rise fault early warning method and specific examples in Example 1 are also applicable to the multi-physics simulation coupled electricity meter temperature rise fault early warning system in this embodiment. Through the foregoing detailed description of the multi-physics simulation coupled electricity meter temperature rise fault early warning method, those skilled in the art can clearly understand the multi-physics simulation coupled electricity meter temperature rise fault early warning system in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0091] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0092] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for early warning of temperature rise faults in electricity meters based on multiphysics simulation coupling, characterized in that, include: A multi-physics field hierarchical system is constructed, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer. Based on the multi-physics field hierarchical system, a simulation model of the target energy meter is performed, and a three-dimensional coupled electrothermal-mechanical simulation model is constructed. Collect historical temperature rise fault data of electricity meters, map the historical temperature rise fault data of electricity meters to the electrothermal three-dimensional coupled simulation model to perform fault prediction and identification and fault tracing search, and generate temperature rise fault prediction and identification branch and electricity meter fault feature tracing branch. The temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused and optimized to establish an electricity meter temperature rise fault diagnostic tool. Based on the energy meter temperature rise fault diagnostic device, fault diagnosis is performed on the real-time operating data of the energy meter, the energy meter temperature rise fault diagnosis result is output, and a graded early warning decision is made based on the energy meter temperature rise fault diagnosis result; The generation of the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch includes: The historical temperature rise fault data of the electricity meter is standardized and preprocessed, and the fault type and degree are labeled to obtain the electricity meter temperature rise fault data sample. The energy meter temperature rise fault data samples are mapped to the electrothermal three-dimensional coupled simulation model for fault prediction and identification, generating an energy meter temperature rise fault prediction and identification branch. Define graph vertices and graph connecting edges, and use the graph vertices and graph connecting edges to perform graph abstraction transformation on the target energy meter to construct a multi-physics field coupling relationship graph of the energy meter; Based on the historical temperature rise fault data of the electricity meter, a fault tracing search is performed on the multi-physics coupling relationship diagram of the electricity meter to generate a fault feature tracing branch of the electricity meter.

2. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 1, characterized in that, Based on the aforementioned multiphysics hierarchical system, a simulation model of the target electricity meter is constructed, including: Obtain key attribute information of the target energy meter, including structural dimension parameters, material property parameters, and boundary constraint parameters; Based on the actual operating status of the electricity meter, determine the electromagnetic boundary conditions, heat source information, and load information; Based on the multiphysics hierarchical system, the electromagnetic boundary conditions, heat source information and load information are sequentially loaded to solve and model the key attribute information, generating electromagnetic field layer model, temperature field layer model and mechanical field layer model; The electromagnetic field layer model, temperature field layer model, and mechanical field layer model are coupled and simulated to construct a three-dimensional coupled electrothermal-mechanical simulation model.

3. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 1, characterized in that, The energy meter temperature rise fault data samples are mapped to the electrothermal three-dimensional coupled simulation model for fault prediction and identification, generating an energy meter temperature rise fault prediction and identification branch, including: The sample of the energy meter temperature rise fault data is mapped to the electrothermal three-dimensional coupled simulation model to perform sample working condition inversion, and the energy meter fault working condition inversion data is obtained. Based on the inversion data of the fault conditions of the electricity meter, the electrothermal three-dimensional coupled simulation model is forward deduced to obtain the simulation fingerprint data of the electricity meter fault. Fault features are extracted and associated samples are labeled from the simulated fingerprint data of the electricity meter faults to form an electricity meter fault fingerprint sample library. Based on the fault fingerprint sample library of the electricity meter, the electrothermal three-dimensional coupled simulation model is driven to perform fault prediction and identification training, and a fault prediction and identification branch for electricity meter temperature rise is generated.

4. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 1, characterized in that, Based on the historical temperature rise fault data of the electricity meter, a fault tracing search is performed on the multiphysics coupling relationship diagram of the electricity meter to generate a fault feature tracing branch of the electricity meter, including: The operating condition features of the historical temperature rise fault data of the electricity meter are extracted to obtain the fault operating condition feature data of the electricity meter. The fault condition feature data of the electricity meter is projected onto the multi-physics field coupling relationship diagram of the electricity meter to mark the fault vertices, thereby obtaining the set of fault vertices of the electricity meter. The vertex with the highest temperature in the set of fault vertices of the electricity meter is selected as the target fault point; Based on the target fault point, a fault tracing search is performed within the multiphysics coupling relationship diagram of the electricity meter to generate a fault feature tracing branch of the electricity meter.

5. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 4, characterized in that, Based on the target fault point, a fault tracing search is performed within the multiphysics coupling graph of the electricity meter to generate a fault feature tracing branch for the electricity meter, including: Taking the target fault point as the starting point, a reverse depth search is performed within the multi-physics coupling relationship diagram of the energy meter to obtain the multi-condition fault root cause tracing chain. Based on the multi-condition fault root cause tracing chain, the multi-condition fault root cause vertices are determined; Based on the multi-condition fault root cause tracing chain and the multi-condition fault root cause vertex, the graph path is accumulated to obtain the multi-condition fault accumulation tracing graph. The multi-condition fault accumulation tracing map is updated, classified, and predicted to generate a fault feature tracing branch for the electricity meter.

6. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 5, characterized in that, The multi-condition fault accumulation tracing map is updated, classified, and predicted to generate a fault feature tracing branch for the electricity meter, including: The fault path edges and root cause node weights of the multi-condition fault accumulation tracing graph are updated according to the number of accumulated operating conditions to obtain the multi-condition fault update tracing graph. The multi-condition fault update tracing map is classified and predicted in terms of pattern to generate a fault feature tracing branch for the electricity meter.

7. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 1, characterized in that, The temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch are fused and optimized to establish an electricity meter temperature rise fault diagnostic tool, including: The temperature rise fault prediction and identification branch and the energy meter fault feature tracing branch are fused in parallel to generate an initial temperature rise fault diagnostic tool. The initial temperature rise fault diagnostic device was tested, evaluated, and iteratively optimized to establish a temperature rise fault diagnostic device for electricity meters.

8. The method for early warning of temperature rise faults in energy meters based on multiphysics simulation coupling as described in claim 1, characterized in that, The tiered early warning decision is made based on the fault diagnosis results of the electricity meter temperature rise, including: A hierarchical early warning strategy library is constructed, and the hierarchical early warning strategy library is used to match the fault diagnosis results of the electricity meter temperature rise to determine the hierarchical early warning strategy of the electricity meter. The target electricity meter is given a graded warning decision through the electricity meter graded warning strategy.

9. A multiphysics simulation-coupled early warning system for temperature rise faults in electricity meters, characterized in that, The steps for implementing the multiphysics simulation-coupled energy meter temperature rise fault early warning method according to any one of claims 1 to 8, wherein the multiphysics simulation-coupled energy meter temperature rise fault early warning system comprises: The simulation modeling module is used to build a multi-physics field hierarchical system, which includes an electromagnetic field layer, a temperature field layer, and a mechanical field layer. Based on the multi-physics field hierarchical system, the target energy meter is simulated and modeled to construct a three-dimensional coupled electrothermal-mechanical simulation model. The fault identification and tracing module is used to collect historical temperature rise fault data of the electricity meter, map the historical temperature rise fault data of the electricity meter to the electrothermal three-dimensional coupled simulation model to perform fault prediction identification and fault tracing search, and generate temperature rise fault prediction identification branch and electricity meter fault feature tracing branch. The branch fusion optimization module is used to fuse and optimize the temperature rise fault prediction and identification branch and the electricity meter fault feature tracing branch to establish an electricity meter temperature rise fault diagnostic tool. The fault diagnosis module is used to perform fault diagnosis on the real-time operating data of the electricity meter based on the electricity meter temperature rise fault diagnostic device, output the electricity meter temperature rise fault diagnosis result, and make graded early warning decisions based on the electricity meter temperature rise fault diagnosis result.

Images