Multi-physics field coupled power transmission and transformation equipment fault prediction method, system, equipment and medium

Abstract

The invention discloses a multi-physics field coupled power transmission and transformation equipment fault prediction method, system, equipment and medium. The method comprises the following steps: cleaning and normalizing operation data to obtain standard data; constructing a corresponding simulation model for the power transmission and transformation equipment; a pre-trained machine learning module is adopted to calibrate the simulation model, secondary rendering is carried out on the calibrated simulation model, and a calibration rendering model is generated; and performing fault prediction on the power transmission and transformation equipment by using the calibrated simulation model, extracting fault feature data from a prediction result, and mapping the fault feature data to the calibration rendering model for visual rendering. According to the method, the simulation model is constructed by setting a plurality of mutually coupled physical fields and control equations, so that the problem that single-physical-field simulation ignores a multi-physical-field coupling effect is solved, and a pre-trained machine learning module is adopted to calibrate the simulation model; the problem that a simulation model lacks online calibration capability and cannot adapt to equipment aging and working condition changes is solved.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault prediction technology, and in particular to a method, system, device and medium for predicting faults in power transmission and transformation equipment using multi-physics coupling. Background Technology

[0002] With the rapid construction of smart grids and the large-scale integration of new energy sources, the requirements for the operational reliability of power transmission and transformation equipment are increasing. As the core hubs of the power grid, the health status of key equipment such as transformers, circuit breakers, and transmission lines directly affects the safe and stable operation of the entire power system. Against this backdrop, condition monitoring and fault prediction technologies for power transmission and transformation equipment have become a research hotspot in the intelligent operation and maintenance of the power industry.

[0003] However, existing fault prediction technologies rely on empirical models or single-physics simulations, neglecting the coupling effects between multiple physical fields such as electromagnetics, heat, and fluids. This leads to insufficient prediction accuracy. Furthermore, most existing simulation models have fixed parameters and lack online calibration capabilities, failing to adapt to equipment aging and dynamic changes in operating conditions. In addition, traditional simulation methods struggle to balance computational efficiency and real-time performance, resulting in excessively long full-scale simulation times that cannot meet the demands of real-time monitoring and rapid response. These issues collectively limit the effective application of existing equipment condition monitoring and fault prediction technologies in smart grid operation and maintenance scenarios. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method, system, device, and medium for predicting faults in power transmission and transformation equipment using multi-physics coupling, which solves the problems in the prior art such as insufficient prediction accuracy due to neglecting the multi-physics coupling effect, lack of online calibration capability of simulation models that cannot adapt to equipment aging and changes in operating conditions, difficulty in balancing computational efficiency and real-time performance in traditional simulation methods, and difficulty in balancing computational efficiency and real-time performance.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a multi-physics coupled fault prediction method for power transmission and transformation equipment, comprising: collecting operating data of power transmission and transformation equipment through sensors, cleaning and normalizing the operating data to obtain standard data, and uploading the standard data to a cloud platform through an edge gateway; constructing a corresponding simulation model for the power transmission and transformation equipment, and driving the simulation model to perform initial rendering by pulling the standard data through the cloud platform; calibrating the simulation model using a pre-trained machine learning module, and performing a second rendering of the calibrated simulation model to generate a calibrated rendering model; using the calibrated simulation model to predict faults in the power transmission and transformation equipment, extracting fault feature data from the prediction results and mapping it to the calibrated rendering model for visualization rendering, and performing risk rating and outputting early warning information based on the visualization rendering results.

[0007] As a preferred embodiment of the multi-physics coupling power transmission and transformation equipment fault prediction method of the present invention, the step of cleaning and normalizing the operating data to obtain standard data includes: identifying and removing abnormal data in the operating data by using the 3σ criterion, and filling in the missing values ​​in the operating data to obtain cleaned data; and normalizing the cleaned data to the [0,1] interval using the maximum-minimum normalization algorithm to obtain the standard data.

[0008] The beneficial effects of this preferred technical solution are as follows: Outliers in the operational data are identified and removed using the 3σ criterion, effectively eliminating distorted data points caused by sensor noise or transmission interference, thus ensuring the reliability of the data source; missing data is filled in using linear interpolation, maintaining the continuity and integrity of the operational data sequence and avoiding deviations in analysis results caused by missing data; the cleaned data is linearly transformed to the [0,1] interval using the max-min normalization algorithm, eliminating the influence of different physical dimensions and magnitudes on subsequent calculations, and making multi-source heterogeneous data comparable.

[0009] As a preferred embodiment of the multi-physics coupling fault prediction method for power transmission and transformation equipment described in this invention, the step of constructing a corresponding simulation model for the power transmission and transformation equipment includes: obtaining a three-dimensional structural diagram of the power transmission and transformation equipment, and generating a corresponding computational grid model based on the three-dimensional structural diagram; setting multiple mutually coupled physical fields and control equations on the computational grid model to simulate the actual operating conditions of the power transmission and transformation equipment, thereby obtaining the simulation model suitable for multi-physics coupling.

[0010] The beneficial effects of this preferred technical solution are as follows: Based on the three-dimensional structural diagram of the power transmission and transformation equipment, a computational grid model is generated, ensuring the consistency between the simulation model and the physical equipment in terms of geometric structure. Then, by setting up multiple mutually coupled physical fields, the interaction relationship between electromagnetic field, thermal field and fluid field is established, eliminating the limitations of single physical field analysis. Then, the evolution law and coupling mechanism of each physical field are described by the control equation, forming a complete mathematical model. Moreover, the coupled simulation of multiple physical fields can reflect the comprehensive state of the power transmission and transformation equipment under actual operating conditions and capture potential problems that cannot be found by single physical field analysis.

[0011] As a preferred embodiment of the multiphysics coupling fault prediction method for power transmission and transformation equipment described in this invention, the step of driving the simulation model to perform the initial rendering by pulling the standard data from the cloud platform includes: obtaining the standard data from the cloud platform in real time, inputting the standard data into the physical field control equations corresponding to the simulation model, driving the simulation model to perform real-time solution calculations, and obtaining the multiphysics distribution results of the power transmission and transformation equipment under the current operating state; updating the visualization state of the simulation model in real time according to the multiphysics distribution results, and completing the initial rendering of the simulation model.

[0012] The beneficial effects of this preferred technical solution are as follows: By acquiring standard data from the cloud platform in real time, the timeliness of the input data is ensured, enabling the simulation model to reflect the latest operating status of the power transmission and transformation equipment; by inputting the standard data into the physical field control equations, the direct coupling between the monitoring data and the simulation model is realized, establishing a synchronous relationship between the virtual model and the physical equipment; the simulation model is driven to perform real-time solution calculations, obtaining comprehensive analysis results containing the distribution of multiple physical quantities, overcoming the limitations of single parameter monitoring, and the obtained multi-physics distribution results can reflect the interaction of various physical fields inside the power transmission and transformation equipment, revealing complete information on the operating status of the power transmission and transformation equipment. At the same time, updating the visualization status of the simulation model based on the multi-physics distribution results allows the invisible physical field distribution to be transformed into an intuitive graphical display. The entire initial rendering process establishes a visual correspondence between the digital model and the physical equipment, enabling maintenance personnel to directly observe the internal status of the power transmission and transformation equipment, thus improving real-time performance.

[0013] As a preferred embodiment of the multi-physics field coupled power transmission and transformation equipment fault prediction method of the present invention, the step of calibrating the simulation model using a pre-trained machine learning module includes: comparing the multi-physics field distribution results with the measured data collected by the sensors at the corresponding time to generate multiple sets of error data, and aggregating the multiple sets of error data to obtain the overall error scalar; iteratively adjusting the parameters inside the simulation model through the optimization algorithm in the machine learning module until the overall error scalar is less than a preset error threshold, thereby completing the calibration of the simulation model.

[0014] The beneficial effects of this preferred technical solution are as follows: by comparing the multiphysics distribution results with the sensor measured data, a quantitative comparison benchmark is established between the simulation model output and the actual state of the power transmission and transformation equipment; the process of generating multiple sets of error data realizes a comprehensive evaluation of the simulation model accuracy in multiple dimensions; the aggregation of multiple sets of error data into a total error scalar simplifies the comprehensive judgment criteria for the accuracy of the simulation model; the optimization algorithm in the machine learning module iteratively adjusts the internal parameters of the simulation model, and the calibration completion condition is that the total error scalar is less than the preset error threshold, ensuring the consistency between the simulation model output and the measured data. It can correct the model parameters according to the actual operating data of the power transmission and transformation equipment, compensate for the model deviation caused by equipment aging or environmental changes, and the parameter iterative adjustment method enables the simulation model to have the ability to continuously optimize, maintaining the correspondence between the state of the simulation model and the actual equipment.

[0015] As a preferred embodiment of the multiphysics coupling fault prediction method for power transmission and transformation equipment described in this invention, the step of extracting fault feature data from the prediction results and mapping it to the calibration rendering model for visualization rendering includes: extracting fault feature data characterizing the health status of the power transmission and transformation equipment from the prediction results, and spatially associating the fault feature data with the corresponding geometric components in the calibration rendering model; converting the spatially associative fault feature data into the visualization attributes of the corresponding geometric components in the calibration rendering model according to preset visual encoding rules; and updating the display state of the calibration rendering model based on the converted visualization attributes to complete the visualization rendering of the fault feature data.

[0016] The beneficial effects of this preferred technical solution are as follows: Fault feature data characterizing the health status of equipment is extracted from the prediction results, transforming abstract prediction conclusions into specific, processable parameters. Furthermore, by establishing a spatial association between the fault feature data and corresponding geometric components in the calibration rendering model, the accurate correspondence between the fault feature data and the components of the calibration rendering model is ensured. Next, based on preset visual encoding rules, the spatially associated feature data is converted into visual attributes, realizing the transformation from data to graphic elements. The converted visual attributes are used to update the display status of the calibration rendering model, allowing the fault feature data to be intuitively displayed at the corresponding position in the simulation model. This mapping mechanism enables accurate positioning and presentation of internal fault features in the 3D model. Moreover, the visualization rendering process transforms numerical analysis results into directly identifiable graphic information, reducing the difficulty of data interpretation. The spatial association between fault features and model components ensures consistency between the displayed content and the actual equipment structure, allowing maintenance personnel to quickly locate problematic components. Simultaneously, standardized display of different types of fault feature data is achieved through unified visual encoding rules, establishing a standardized visual expression system.

[0017] As a preferred embodiment of the multi-physics coupling fault prediction method for power transmission and transformation equipment described in this invention, the step of performing risk rating and outputting early warning information through visualization rendering results includes: determining the comprehensive risk level of the power transmission and transformation equipment based on the risk characteristics displayed in the visualization rendering results through preset risk mapping rules, and generating corresponding early warning information according to the comprehensive risk level; triggering a visual warning mark at the corresponding position of the calibration rendering model according to the early warning information, and pushing an early warning report containing risk decision elements to the operation and maintenance terminal through an interface.

[0018] The beneficial effects of this preferred technical solution are as follows: By converting the risk characteristics in the visualization rendering results into a definite comprehensive risk level through preset risk mapping rules, a standardized mapping relationship from visual information to risk level is established. Based on the comprehensive risk level, corresponding early warning information is generated, forming a complete risk assessment output. At the same time, visual warning markers are triggered at the corresponding positions of the calibration rendering model according to the early warning information, realizing the visual positioning of risk locations. Finally, the early warning report containing risk decision-making elements is pushed to the operation and maintenance terminal through the interface, ensuring the timely transmission of risk information. This risk rating mechanism transforms complex multiphysics analysis results into clear operation and maintenance guidance, reducing decision-making complexity. The collaborative output of visual warning markers and early warning reports establishes a dual prompting mechanism, improving the reliability of risk notification. Meanwhile, the standardized risk level classification unifies the evaluation standards of different devices, facilitating the formulation of corresponding disposal measures.

[0019] Secondly, the present invention provides a multi-physics coupled power transmission and transformation equipment fault prediction system, comprising: Data acquisition and preprocessing module: used to collect operating data of power transmission and transformation equipment through sensors, clean and normalize the operating data to obtain standard data, and upload the standard data to the cloud platform through the edge gateway; Simulation model building module: used to build corresponding simulation models for power transmission and transformation equipment; Real-time rendering module: used to pull the standard data from the cloud platform to drive the simulation model to perform the initial rendering, and to perform a secondary rendering of the calibrated simulation model to generate a calibration rendering model; Model calibration module: integrates a pre-trained machine learning module, used to calibrate the simulation model to obtain a calibrated simulation model; Fault prediction and early warning module: It is used to predict faults using the calibrated simulation model, extract fault feature data from the prediction results, map them onto the calibrated rendering model for visualization rendering, and perform risk rating and output early warning information based on the visualization rendering results.

[0020] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the multi-physics field coupled power transmission and transformation equipment fault prediction method are implemented.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the multi-physics coupled power transmission and transformation equipment fault prediction method.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows: By setting up multiple mutually coupled physical fields and control equations to construct a simulation model, the interaction relationship between electromagnetic field, thermal field and fluid field is established, overcoming the problem of ignoring the multi-physical field coupling effect in single-physical field simulation, making the simulation results closer to the actual operating state of power transmission and transformation equipment, and solving the problem of insufficient prediction accuracy caused by ignoring the multi-physical field coupling effect; A pre-trained machine learning module is used to calibrate the simulation model. Error data is generated by comparing the multiphysics distribution results with the sensor measured data. An optimization algorithm is used to iteratively adjust the parameters inside the simulation model, and an optimization mechanism for the model parameters is established. This enables the simulation model to correct its parameters based on the actual operating data of the power transmission and transformation equipment, thus solving the problem that the simulation model lacks online calibration capabilities and cannot adapt to equipment aging and changes in operating conditions.

[0023] By pulling standard data from the cloud platform to drive the simulation model for real-time calculation, and combining data preprocessing and incremental transmission mechanisms from the edge gateway, data transmission latency is reduced. Real-time rendering technology is used to update the model's visualization status, achieving synchronization between the virtual model and physical devices, thus solving the problem of balancing computational efficiency and real-time performance in traditional simulation methods. Furthermore, by spatially associating fault feature data with geometric components in the calibration rendering model, and then converting the feature data into visual attributes according to visual encoding rules, fault features are accurately located and presented in the 3D model. Finally, based on the visualization rendering results, a comprehensive risk level is determined through risk mapping rules, generating early warning information and triggering visual warning markers. An early warning report containing risk decision-making elements is pushed to the operation and maintenance terminal via an interface, forming a complete risk early warning closed loop. The synergistic effect of these technical features achieves an effective combination of multiphysics coupled simulation and real-time data-driven approaches, improving the accuracy of fault prediction while ensuring computational efficiency. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the overall process of a multi-physics field coupled power transmission and transformation equipment fault prediction method according to an embodiment of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1 As an embodiment of the present invention, a fault prediction method for power transmission and transformation equipment coupled with multi-physics fields is provided, including steps S100 to S400: S100 collects operating data of power transmission and transformation equipment through sensors, cleans and normalizes the operating data to obtain standard data, and uploads the standard data to the cloud platform through the edge gateway.

[0028] S200: Construct a corresponding simulation model for power transmission and transformation equipment, and use the cloud platform to pull standard data to drive the simulation model for initial rendering.

[0029] S300 uses a pre-trained machine learning module to calibrate the simulation model and then performs a second rendering on the calibrated simulation model to generate a calibration rendering model.

[0030] S400 uses the calibrated simulation model to predict faults in power transmission and transformation equipment, extracts fault feature data from the prediction results, maps it onto the calibrated rendering model for visualization rendering, and uses the visualization rendering results to perform risk rating and output early warning information.

[0031] It should be noted that power transmission and transformation equipment (such as transformers and circuit breakers) are in a multi-physical field coupling environment involving electricity, heat, and mechanics. The data types are diverse and easily affected by electromagnetic interference and environmental noise, making data cleaning and normalization difficult. At the same time, traditional fault prediction is mostly based on single-physical field analysis or offline simulation, which is difficult to accurately simulate the dynamic coupling characteristics of multi-physical fields. This leads to deviations between the simulation model and the actual equipment, resulting in insufficient fault prediction accuracy. In addition, fault characteristic data lacks intuitive visualization, and the timeliness and accuracy of risk rating and early warning are limited. This poses a challenge to the early fault identification and safe operation and maintenance of power transmission and transformation equipment. Therefore, fault prediction and visualized early warning based on multi-physical field coupling are crucial.

[0032] Therefore, to address the aforementioned issues of high data processing difficulty, insufficient accuracy in multi-physics coupled simulation, and unintuitive fault visualization and early warning, steps S100-S400 are used to collect and process operational data, construct and drive a simulation model, calibrate the model using a machine learning module for fault prediction, and then map fault features onto the model for visualization rendering and early warning output. This achieves effective cleaning and uploading of operational data, solving the accuracy deviation problem in multi-physics coupled simulation; and by intuitively presenting fault features through visualization rendering, the efficiency and accuracy of risk rating and early warning are improved, thereby achieving accurate prediction, intuitive visualization, and timely early warning of power transmission and transformation equipment faults, ensuring the safe and stable operation of power transmission and transformation equipment.

[0033] Example 2, refer to Figure 1 As an embodiment of the present invention, based on the above embodiment, a method for predicting faults in power transmission and transformation equipment using multi-physics coupling is provided.

[0034] In this embodiment of the application, the step S100 of cleaning and normalizing the running data to obtain standard data includes A1~A2: A1. Identify and remove abnormal data in the running data using the 3σ criterion, and fill in the missing values ​​in the running data to obtain cleaned data; In this embodiment, the arithmetic mean and standard deviation are calculated for different types of operating data, such as electrical parameters, thermal parameters, and mechanical parameters. Then, three times the standard deviation is used as the judgment threshold. Data points that deviate from the arithmetic mean by more than the judgment threshold are judged as outliers and removed. For data missing due to outlier removal, linear interpolation is used to fill in the missing data. First, a linear function is constructed using the valid data points adjacent to the missing point. Then, the value at the missing position is calculated based on the linear function to obtain complete and reliable cleaned data, which is temporarily stored in the edge gateway cache.

[0035] A2. Use the maximum-minimum normalization algorithm to normalize the cleaned data to the [0,1] interval to obtain standard data; For each type of dataset (such as electrical parameters, thermal parameters, and mechanical parameters) in the cleaned data, the global maximum and global minimum values ​​of the dataset are first identified. Then, the range is obtained through the global maximum and global minimum values. Next, the global minimum value is subtracted from the value of each data point in the dataset, and the difference is divided by the range, thereby converting each data point into a dimensionless value between 0 and 1. After all data points in all types of datasets have been processed, standard data is obtained.

[0036] In an optional implementation, the standard data obtained by cleaning and normalizing the operating data in step S100 can also be identified and removed from the operating data using the box plot method. First, the upper quartile, lower quartile, and interquartile range of various types of operating data, such as electrical parameters, thermal parameters, and mechanical parameters, are calculated respectively. Then, the upper quartile is added to 1.5 times the interquartile range as the upper bound for judging abnormal data, and the lower quartile is subtracted from 1.5 times the interquartile range as the lower bound for judging abnormal data. Data points that exceed the upper and lower bounds are judged as abnormal data and removed. For data missing after the removal of abnormal values, the value at the missing position is calculated by constructing a smooth curve function with adjacent valid data points. After the data cleaning is completed, the upper and lower bounds for judging abnormal data are used as the maximum and minimum values, respectively. Max-min normalization processing is performed on the cleaned data to transform the cleaned operating data into the [0,1] interval, and finally standard data is generated.

[0037] In this embodiment of the application, step S200, which involves constructing a corresponding simulation model for the power transmission and transformation equipment, includes steps B1 to B2: B1. Obtain the three-dimensional structural diagram of the power transmission and transformation equipment, and generate the corresponding computational mesh model based on the three-dimensional structural diagram; In this embodiment, the CAD geometric model of the power transmission and transformation equipment is imported into finite element software for geometric structure analysis. Key areas of different types of power transmission and transformation equipment are identified, such as transformer windings, transformer cores, and circuit breaker arc-extinguishing chambers. Subsequently, unstructured mesh generation technology is used to discretize the computational domain. During the discretization process, it is necessary to ensure that the generated mesh meets the preset quality indicators: element skewness is less than 0.8 and aspect ratio is less than 5. At the same time, mesh refinement is implemented for the identified key areas (such as the ends of transformer windings and the corners of transformer cores). After the mesh is generated, mesh independence verification is performed. Specifically, the mesh is gradually refined and the changes in key physical quantities (such as maximum field strength or highest temperature) are compared. When the difference between two adjacent mesh refinement results is less than 5%, the mesh accuracy is considered to meet the simulation requirements. Finally, a computational mesh model consistent with the physical structure of the power transmission and transformation equipment is output.

[0038] B2. Set up multiple coupled physical fields and control equations on the computational grid model to simulate the actual operating conditions of power transmission and transformation equipment, and obtain a simulation model suitable for multi-physics coupling. In this embodiment, an electromagnetic field is first configured. Then, the electromagnetic losses, such as eddy current losses in transformer windings and hysteresis losses in the core, are obtained by calculating the magnetic field distribution of the power transmission and transformation equipment under operating current using Maxwell's equations. Subsequently, the calculated electromagnetic losses are input into the temperature field as a heat source. The temperature distribution of the equipment is obtained by solving the heat conduction and convection equations, with a focus on analyzing the distribution of hot spots in the transformer windings and the temperature rise in the transformer core. Simultaneously, the temperature distribution results are fed back to the material parameter database to update the electrical conductivity and thermal conductivity of the materials. Finally, the updated electromagnetic field parameters are analyzed for magnetic field distribution. The calculation of the cloth is recalculated to form a two-way coupling of electromagnetic and thermal fields. Simultaneously, a fluid field is configured to simulate the flow process of the insulating medium based on the Navier-Stokes equations, such as the circulating flow of transformer oil and the gas flow during circuit breaker opening. The velocity and pressure distribution of the fluid field are then fed back to the temperature field to update the convective heat transfer boundary conditions. After completing the multi-physics coupling setup, the preprocessed conjugate gradient method is used to iteratively solve the coupled equations. In each iteration, the parameters of the electromagnetic field, temperature field, and fluid field are updated sequentially, and the residual convergence criterion for each physics field is set to be less than 10. -6 When the calculation results of all physical fields reach a stable state, the construction of the multiphysics coupling simulation model is completed.

[0039] In this embodiment of the application, the step S200, which involves pulling standard data from a cloud platform to drive the initial rendering of the simulation model, includes C1~C2: C1. Obtain standard data from the cloud platform in real time and input the standard data into the physical field control equations corresponding to the simulation model to drive the simulation model to perform real-time calculations and obtain the multi-physics distribution results of the power transmission and transformation equipment under the current operating state. First, standard data processed in step S100 is retrieved from the cloud platform. Then, the standard data is used as real-time updated parameters and input into the control equations of each physical field corresponding to the simulation model constructed in step B2. In this embodiment, electrical standard data is used as the excitation source of the electromagnetic field, temperature standard data is used as the initial condition of the temperature field, and flow velocity standard data is used as the boundary condition of the fluid field. Based on these real-time parameters, the simulation model is driven to solve the problem and directly obtain the multi-physics distribution results of the power transmission and transformation equipment under the current operating state, including but not limited to magnetic flux density distribution, temperature distribution, and flow velocity distribution.

[0040] C2. Update the visualization status of the simulation model in real time based on the multiphysics field distribution results to complete the initial rendering of the simulation model; In this embodiment, the temperature distribution is converted into a visual image using a color mapping algorithm, for example, using a continuous color spectrum from blue to red to represent the temperature change from low to high; the magnetic flux density distribution is visualized using vector arrows, where the arrow direction indicates the magnetic field direction, and the arrow length and color depth indicate the magnitude of the field strength; the flow velocity distribution is generated into a streamline diagram using streamline tracing technology, where streamline density and color represent the flow rate. All visualization elements are rendered based on the computational grid model generated in step B1, and real-time 3D rendering on the browser side is achieved using WebGL technology, with a rendering frame rate of no less than 1Hz, ensuring that the visualization state is synchronized with the physical device in real time, thereby completing the initial rendering of the simulation model.

[0041] In an optional implementation, the initial rendering of the simulation model by pulling standard data from the cloud platform in step S200 can also be achieved as follows: After obtaining standard data from the cloud platform and inputting it into the simulation model to complete real-time solving, temperature distribution, magnetic flux density distribution, and velocity distribution data are extracted respectively. Subsequently, the extracted data are visualized and geometrically generated. For example, multiple temperature thresholds are set for the temperature field data, and an isosurface extraction algorithm is used to generate three-dimensional isosurface models for different temperature ranges; magnetic flux density field is subjected to magnetic field line tracing, and the three-dimensional magnetic field line distribution is reconstructed through numerical integration; and a particle system is deployed for visualization of the velocity field. The process involves injecting virtual particles into the flow field and numerically simulating their motion trajectories. Then, a WebGPU rendering architecture is used for visualization. The isosurface model is constructed in real-time using computational shaders, magnetic field lines utilize instantiation rendering technology to optimize rendering performance, and particle trajectories are dynamically generated using geometry shaders. Finally, a unified visual mapping system is established. Specifically, temperature isosurfaces use the HSL color space to define color gradients, magnetic field lines incorporate Fresnel effects to enhance visual depth, and the particle system uses changes in transparency to represent flow velocity differences. Ultimately, the comprehensive distribution of multiple physics fields is simultaneously presented in a 3D visualization interface, completing the initial rendering.

[0042] In this embodiment of the application, the step of calibrating the simulation model using a pre-trained machine learning module in step S300 includes D1~D2: D1. Compare the multiphysics distribution results with the measured data collected by the sensor at the corresponding time to generate multiple sets of error data, and aggregate the multiple sets of error data to obtain the overall error scalar. In this embodiment, firstly, the temperature values ​​at key points in the temperature field distribution are extracted and compared with the measured values ​​from the temperature sensor to calculate the absolute error; then, the field strength values ​​at specific locations in the magnetic flux density distribution are extracted and compared with the measured values ​​from the magnetic field sensor to calculate the relative error; finally, the flow velocity values ​​in key areas of the flow velocity distribution are extracted and compared with the measured values ​​from the flow velocity sensor to calculate the root mean square error; the above multiple sets of error data are aggregated into a total error scalar using a weighted average method, where the weight coefficients are determined according to the importance of each physical quantity in fault prediction. For example, the weight of temperature error is set to 0.5, the weight of electromagnetic error is set to 0.3, and the weight of flow velocity error is set to 0.2, finally obtaining the total error scalar that characterizes the accuracy of the current simulation model.

[0043] D2. The parameters inside the simulation model are iteratively adjusted using the optimization algorithm in the machine learning module until the total error scalar is less than the preset error threshold, thus completing the calibration of the simulation model. The Levenberg-Marquardt optimization algorithm is employed, with the objective function of minimizing the overall error scalar. This algorithm adjusts material parameters (e.g., transformer winding conductivity, transformer core permeability, insulating oil viscosity) and boundary condition coefficients in the simulation model. During each iteration, the algorithm calculates the partial derivatives of the overall error scalar with respect to each model parameter, thereby determining the parameter update direction that minimizes the error. The step size is calculated by the algorithm based on the curvature characteristics of the current error surface. In this embodiment, an error threshold of 5% is set. When the overall error scalar decreases below 5%, the iteration process terminates, and after parameter adjustment, the optimized parameters are updated into the simulation model to generate a calibrated simulation model.

[0044] In an optional implementation, the calibration of the simulation model using a pre-trained machine learning module in step S300 can also be achieved by constructing an initial simplex in the parameter space, followed by iterative calculations and geometric operations to move the initial simplex toward the optimal solution. Specifically, in each iteration, the total error scalar of each vertex of the simplex is calculated. New candidate points are generated through geometric operations such as reflection, expansion, and contraction, and the shape and position of the simplex are updated, thereby gradually narrowing the search range. An error threshold is set to 5%. The search terminates when the smallest total error scalar among the searched parameter combinations is less than 5% or the simplex size is less than the set tolerance, and the current parameter combination is updated in the simulation model to complete the calibration.

[0045] In this embodiment of the application, step S400, which involves extracting fault feature data from the prediction results and mapping it onto the calibration rendering model for visualization rendering, includes steps E1 to E3: E1. Extract fault feature data representing the health status of power transmission and transformation equipment from the prediction results, and spatially correlate the fault feature data with the corresponding geometric components in the calibration rendering model; In this embodiment, fault feature data characterizing the health status of power transmission and transformation equipment is first extracted from the prediction results. Specifically, the coordinates and temperature values ​​of high-temperature areas exceeding the threshold are extracted from the temperature field prediction results, the coordinates and velocity values ​​of abnormal flow velocity areas are extracted from the fluid field prediction results, and the coordinates and field strength values ​​of abnormal field strength areas are extracted from the electromagnetic field prediction results. Then, the spatial coordinates of the fault feature data are mapped to the corresponding grid cells in the calibration rendering model generated in step C2 through a coordinate matching algorithm, establishing a one-to-one correspondence between the fault feature data and the geometric components of the calibration rendering model.

[0046] E2. Based on the preset visual encoding rules, the spatially correlated fault feature data is converted into the visual attributes of the corresponding geometric components in the calibration rendering model. First, the corresponding visual coding rule is selected based on the type of abnormal feature data. In this embodiment, for temperature anomalies, a red, yellow, and blue color gradient rule is used to map temperature values ​​to corresponding colors, where red represents high-temperature risk areas, yellow represents temperature warning areas, and blue represents normal temperature areas. For flow velocity anomalies, a streamline and particle combination rule is used to convert flow velocity values ​​into streamline colors and particle movement speeds, where a gradient from blue to red in the streamline color represents a change in flow velocity from normal to abnormal. For electromagnetic field anomalies, a vector arrow and isosurface combination rule is used, where the vector arrow represents the magnitude of the field strength value through its size and color, and the direction of the magnetic field through its direction. The isosurface represents the spatial distribution range of different field strength values ​​through changes in transparency. It should be noted that the threshold range for each feature is set according to the specific operating standards of the power transmission and transformation equipment.

[0047] E3. Update the display status of the calibration rendering model based on the transformed visualization attributes to complete the visualization rendering of fault feature data; In this embodiment, the color attributes corresponding to the temperature anomaly features are first applied to the surface of the corresponding geometric components in the calibration rendering model through the material shader to visualize the thermal map of the temperature distribution. Then, the streamline and particle attributes of the flow velocity anomaly features are added to the 3D scene through the geometry shader and particle system rendering technology to generate dynamic streamline maps and particle animations. Next, the vector arrows and isosurface attributes of the electromagnetic field anomaly features are added to the 3D scene through instantiation rendering technology. The 3D arrow graphics represent the direction of the magnetic field, and the isosurface displays the field strength distribution with different transparency levels. Finally, a visualized rendering model containing complete fault information is output.

[0048] In an optional implementation, step S400, in which fault feature data is extracted from the prediction results and mapped onto the calibration rendering model for visualization rendering, can also employ a visualization method combining a dynamic particle system and holographic projection. First, fault feature data is extracted from the prediction results to establish a unified feature space mapping relationship. The fault feature data includes, but is not limited to, abnormal temperature areas, abnormal flow velocity areas, and electromagnetic field hotspots. Subsequently, during the 3D scene construction phase, two visualization subsystems are deployed simultaneously: on the one hand, a dynamic particle system is constructed, which characterizes the severity of faults in different areas by controlling the spatial distribution density of particles, reproduces the distribution of electromagnetic field lines by utilizing the motion trajectory of charged particles in a simulated magnetic field, and maps the temperature gradient into a continuous change in particle color based on the blackbody radiation law; on the basis of the dynamic particle system, a holographic projection layer is superimposed on the abnormal flow velocity areas, and a 3D streamline cloud map is generated by calculating the flow field vector characteristics, and the flow velocity gradient is converted into a visible change in interference fringe density using the principle of optical interference. The two visualization subsystems achieve visual fusion through deep buffer sharing, and use a ray-tracing algorithm to uniformly process the optical effects of particle scattering and holographic diffraction, ultimately forming a composite visualization effect that combines dynamic flow characteristics with holographic 3D projection. This multimodal fusion rendering method achieves a three-dimensional presentation of fault characteristics.

[0049] In this embodiment of the application, step S400, which involves performing risk rating and outputting warning information based on the visualization rendering results, includes steps F1 to F2: F1. Based on the risk characteristics shown in the visualization rendering results, determine the comprehensive risk level of the power transmission and transformation equipment through preset risk mapping rules, and generate corresponding early warning information according to the comprehensive risk level; First, a multi-dimensional risk assessment system is established. In this embodiment, the multi-dimensional risk assessment system includes, but is not limited to, temperature anomaly, flow velocity anomaly, and electromagnetic field anomaly dimensions. Each dimension is quantitatively scored using a 1-10 point system. The temperature anomaly dimension is scored based on the proportion of the high-temperature area and the degree to which the highest temperature deviates from the rated value. The flow velocity anomaly dimension is scored based on the range of the flow velocity anomaly area and the amplitude of flow velocity fluctuations. The electromagnetic field anomaly dimension is scored based on the range of the field strength anomaly area and the degree to which the peak field strength exceeds the safety threshold. Next, a weighted fusion calculation of the comprehensive risk value is performed, with the weight of the temperature dimension set to 0.5, the weight of the flow velocity dimension set to 0.2, and the weight of the electromagnetic field dimension set to 0.3. The calculation formula is: Comprehensive Risk Value = Temperature Score × 0.5 + Flow Velocity Score × 0.2 + Electromagnetic Field Score × 0.3. Finally, the comprehensive risk value is converted into an equipment status level according to a preset risk mapping rule. The specific mapping rule is: low risk (0-3 points), medium risk (4-7 points), and high risk (8-10 points), and an early warning message containing the risk level, risk location, risk type, and handling suggestions is generated.

[0050] F2. Based on the warning information, trigger visual warning markers at the corresponding positions of the calibration rendering model, and push the warning report containing risk decision elements to the operation and maintenance terminal through the interface. Based on the risk level and location in the warning information, tiered visual warning markers are triggered at the corresponding locations in the calibration rendering model. For example, low-risk areas are displayed as slowly flashing yellow dots, medium-risk areas as continuously rotating orange triangles, and high-risk areas as rapidly flashing red octagonal warning boxes. Simultaneously, structured warning reports are pushed to the operations and maintenance terminal via a RESTful API interface. The reports are encapsulated in JSON format, and the core data segments include the device's unique identifier, risk assessment timestamp, comprehensive risk level, risk scores for each dimension (temperature score, flow rate score, electromagnetic field score), risk hotspot distribution map, and maintenance priority recommendations based on the device's operating history.

[0051] In an optional implementation, step S400, which involves performing risk rating and outputting early warning information based on the visualization rendering results, can also employ a risk assessment method combining fuzzy comprehensive evaluation and evidence theory. First, membership functions for three risk factors—temperature, flow rate, and electromagnetic field—are established. The membership function for temperature anomalies uses a ridge distribution, the membership function for flow rate anomalies uses a triangular distribution, and the membership function for electromagnetic field anomalies uses a trapezoidal distribution. Then, through a fuzzy inference mechanism, the membership degrees of these three risk factors are transformed into basic probability assignments. The assessment results of the three risk factors are used as independent sources of evidence, with a confidence level of 0.45 assigned to temperature anomalies, 0.25 to flow rate anomalies, and 0 to electromagnetic field anomalies. The reliability score was set at 0.3. Then, the Dempster-Shafer evidence synthesis rule was used to fuse three independent evidence bodies. Evidence conflicts were eliminated through orthogonal summation formulas, resulting in a comprehensive reliability distribution for high-risk, medium-risk, and low-risk states. Subsequently, a warning level was determined according to preset decision rules: a red warning was triggered when the reliability score for a high-risk state exceeded 0.6; an orange warning was triggered when the reliability score for a medium-risk state exceeded 0.3 and the reliability score for a high-risk state was less than 0.6; and a yellow warning was triggered when the reliability score for a low-risk state exceeded 0.5 and the reliability scores for both medium- and high-risk states did not reach the threshold. Finally, the warning information was pushed to the operation and maintenance terminal via the MQTT protocol. The message body was serialized using Protocol Buffers format and included a timestamp, device ID, risk level, reliability distribution of each factor, and suggested handling measures. Simultaneously, warning layers with different transparency were used in the 3D interface to distinguish risk levels, forming a risk warning scheme based on uncertainty reasoning.

[0052] In summary, this invention constructs a simulation model by setting up multiple mutually coupled physical fields and control equations, establishing the interaction relationship between electromagnetic field, thermal field and fluid field, overcoming the problem of ignoring the multi-physical field coupling effect in single-physical field simulation, making the simulation results closer to the actual operating state of power transmission and transformation equipment, and solving the problem of insufficient prediction accuracy caused by ignoring the multi-physical field coupling effect; A pre-trained machine learning module is used to calibrate the simulation model. Error data is generated by comparing the multiphysics distribution results with the sensor measured data. An optimization algorithm is used to iteratively adjust the parameters inside the simulation model, and an optimization mechanism for the model parameters is established. This enables the simulation model to correct its parameters based on the actual operating data of the power transmission and transformation equipment, thus solving the problem that the simulation model lacks online calibration capabilities and cannot adapt to equipment aging and changes in operating conditions.

[0053] By pulling standard data from the cloud platform to drive the simulation model for real-time calculation, and combining data preprocessing and incremental transmission mechanisms from the edge gateway, data transmission latency is reduced. Real-time rendering technology is used to update the model's visualization status, achieving synchronization between the virtual model and physical devices, thus solving the problem of balancing computational efficiency and real-time performance in traditional simulation methods. Furthermore, by spatially associating fault feature data with geometric components in the calibration rendering model, and then converting the feature data into visual attributes according to visual encoding rules, fault features are accurately located and presented in the 3D model. Finally, based on the visualization rendering results, a comprehensive risk level is determined through risk mapping rules, generating early warning information and triggering visual warning markers. An early warning report containing risk decision-making elements is pushed to the operation and maintenance terminal via an interface, forming a complete risk early warning closed loop. The synergistic effect of these technical features achieves an effective combination of multiphysics coupled simulation and real-time data-driven approaches, improving the accuracy of fault prediction while ensuring computational efficiency.

[0054] Example 3 illustrates a schematic scheme for a multi-physics field coupled power transmission and transformation equipment fault prediction method. It should be noted that the technical solution of this multi-physics field coupled power transmission and transformation equipment fault prediction system belongs to the same concept as the technical solution of the multi-physics field coupled power transmission and transformation equipment fault prediction method described above. Details not described in detail in the technical solution of the multi-physics field coupled power transmission and transformation equipment fault prediction system in this embodiment can be found in the description of the technical solution of the multi-physics field coupled power transmission and transformation equipment fault prediction method described above.

[0055] This embodiment also provides a multi-physics coupled power transmission and transformation equipment fault prediction system, including: Data acquisition and preprocessing module: used to collect operating data of power transmission and transformation equipment through sensors, clean and normalize the operating data to obtain standard data, and upload the standard data to the cloud platform through the edge gateway; Simulation model building module: used to build corresponding simulation models for power transmission and transformation equipment; Real-time rendering module: used to pull standard data from the cloud platform to drive the simulation model for initial rendering, and to perform secondary rendering on the calibrated simulation model to generate a calibration rendering model. Model calibration module: integrates a pre-trained machine learning module, used to calibrate the simulation model to obtain a calibrated simulation model; Fault prediction and early warning module: It is used to predict faults using the calibrated simulation model, extract fault feature data from the prediction results, map them onto the calibrated rendering model for visualization rendering, and perform risk rating and output early warning information based on the visualization rendering results.

[0056] This embodiment also provides an electronic device suitable for fault prediction of power transmission and transformation equipment with multi-physics coupling, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the fault prediction method for power transmission and transformation equipment with multi-physics coupling as proposed in the above embodiment.

[0057] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the fault prediction method for power transmission and transformation equipment that achieves multi-physics coupling as proposed in the above embodiments.

[0058] The storage medium proposed in this embodiment belongs to the same inventive concept as the method for predicting faults in power transmission and transformation equipment by realizing multi-physics coupling proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0059] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0060] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for multi-physical field coupled power transmission and transformation equipment fault prediction, characterized in that, The method comprises the steps of: collecting operation data of power transmission and transformation equipment through sensors, cleaning and normalizing the operation data to obtain standard data, and uploading the standard data to a cloud platform through an edge gateway; constructing a corresponding simulation model for the power transmission and transformation equipment, and driving the simulation model to perform initial rendering by pulling the standard data from the cloud platform; calibrating the simulation model using a pre-trained machine learning module, and performing secondary rendering on the calibrated simulation model to generate a calibrated rendering model; using the calibrated simulation model to predict faults of the power transmission and transformation equipment, and extracting fault feature data from the prediction results and mapping it to the calibrated rendering model for visual rendering, and performing risk rating and outputting early warning information based on the visual rendering results.

2. The multi-physics coupled power transmission equipment failure prediction method of claim 1, wherein, The step of cleaning and normalizing the operation data to obtain standard data comprises: identifying and removing abnormal data in the operation data through the 3σ criterion, and filling in missing values in the operation data to obtain cleaned data; using the maximum and minimum normalization algorithm to normalize the cleaned data to the [0, 1] interval to obtain the standard data.

3. The multi-physics coupled power transmission equipment failure prediction method of claim 2, wherein, The step of constructing a corresponding simulation model for the power transmission and transformation equipment comprises: obtaining a three-dimensional structure diagram of the power transmission and transformation equipment, and generating a corresponding calculation grid model according to the three-dimensional structure diagram; setting a plurality of physical fields and control equations that are coupled with each other on the calculation grid model to simulate the actual working conditions of the power transmission and transformation equipment, and obtaining the simulation model suitable for multi-physical field coupling.

4. The multi-physics coupled power transmission equipment failure prediction method of claim 3, wherein, The step of driving the simulation model to perform initial rendering by pulling the standard data from the cloud platform comprises: real-time obtaining the standard data from the cloud platform, and inputting the standard data into the physical field control equation corresponding to the simulation model to drive the simulation model to perform real-time solving calculation, and obtaining the multi-physical field distribution results of the power transmission and transformation equipment under the current operating state; updating the visualization state of the simulation model in real time according to the multi-physical field distribution results, and completing the initial rendering of the simulation model.

5. The multi-physics coupled power transmission equipment failure prediction method of claim 4, wherein, The step of calibrating the simulation model using a pre-trained machine learning module comprises: comparing the multi-physical field distribution results with the measured data collected by the sensors at the corresponding time to generate a plurality of error data, and aggregating the plurality of error data to obtain a total error scalar; iteratively adjusting the parameters inside the simulation model through the optimization algorithm in the machine learning module until the total error scalar is less than a preset error threshold, and completing the calibration of the simulation model.

6. The multi-physics coupled power transmission device failure prediction method of claim 5, wherein, The step of extracting fault feature data from the prediction results and mapping it to the calibrated rendering model for visual rendering comprises: extracting fault feature data representing the health status of the power transmission and transformation equipment from the prediction results, and spatially correlating the fault feature data with corresponding geometric components in the calibrated rendering model; according to a preset visual coding rule, converting the spatially correlated fault feature data into visual attributes of the corresponding geometric components in the calibrated rendering model; updating the display state of the calibrated rendering model based on the converted visual attributes, and completing the visual rendering of the fault feature data.

7. The multi-physics coupled power transmission device failure prediction method of claim 6, wherein, The steps for risk assessment and early warning information output based on visualization rendering results include: Based on the risk characteristics shown in the visualization rendering results, the comprehensive risk level of the power transmission and transformation equipment is determined by a preset risk mapping rule, and corresponding early warning information is generated according to the comprehensive risk level. Based on the warning information, a visual warning marker is triggered at the corresponding position of the calibration rendering model, and a warning report containing risk decision-making elements is pushed to the operation and maintenance terminal through the interface.

8. A multi-physics coupled power transmission equipment failure prediction system applying the method of any one of claims 1-7, characterized in that, include: Data acquisition and preprocessing module: used to collect operating data of power transmission and transformation equipment through sensors, clean and normalize the operating data to obtain standard data, and upload the standard data to the cloud platform through the edge gateway; Simulation model building module: used to build corresponding simulation models for power transmission and transformation equipment; Real-time rendering module: used to pull the standard data from the cloud platform to drive the simulation model to perform the initial rendering, and to perform a secondary rendering of the calibrated simulation model to generate a calibration rendering model; Model calibration module: integrates a pre-trained machine learning module, used to calibrate the simulation model to obtain a calibrated simulation model; Fault prediction and early warning module: It is used to predict faults using the calibrated simulation model, extract fault feature data from the prediction results, map them onto the calibrated rendering model for visualization rendering, and perform risk rating and output early warning information based on the visualization rendering results.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the multi-physics field coupled power transmission and transformation equipment fault prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the multi-physics coupled power transmission and transformation equipment fault prediction method according to any one of claims 1 to 7.

Images