Transformer substation equipment fault diagnosis method and system based on virtual mapping

Abstract

The invention provides a transformer substation equipment fault diagnosis method and system based on virtual mapping, and relates to the technical field of digital twinning, and the method comprises the steps: calling a component-level CAD model; calling physical field control parameters, and performing multi-physical field coupling parameter assignment; positioning P key component codes; p multi-mode heterogeneous sensing units are configured; mapping the operation states of the P equipment components to P virtual component sub-models in the multi-physics field coupling simulation model; and executing fault propagation directed graph simulation in the multi-physics field coupling simulation model, and outputting a real-time fault diagnosis result. According to the method and the device, the technical problem of poor fault diagnosis precision caused by the fact that most of transformer substation equipment fault diagnosis can only be accurate to specific equipment and cannot be accurately positioned to equipment parts in the prior art is solved, and the fault diagnosis precision is improved by constructing component-level digital twinning which is synchronously operated with physical equipment. Therefore, the crossing from equipment-level early warning to component-level fine diagnosis is realized, and the fault diagnosis precision is improved.

Description

Technical Field

[0001] This application relates to the field of digital twin technology, specifically to a method and system for diagnosing substation equipment faults based on virtual mapping. Background Technology

[0002] Current substation equipment fault diagnosis technologies generally suffer from limitations in granular diagnostic capabilities. Most online monitoring and intelligent diagnostic systems can only achieve overall equipment-level condition assessments, such as identifying overheating or discharge faults within transformers, but cannot precisely pinpoint the root cause of the fault in specific components. This necessitates secondary troubleshooting by maintenance personnel based on experience and additional testing, offering little improvement in actual fault-solving efficiency. Existing methods largely rely on threshold alarms and data-driven models, lacking in-depth modeling of the internal physical field state of equipment and the fault propagation mechanism between components. This not only increases diagnostic time and cost but may also lead to misdiagnosis or missed diagnosis, thereby affecting maintenance efficiency and grid stability. The lack of refined monitoring and analysis of individual components within the equipment fails to fully reflect subtle changes in component-level faults, resulting in low fault diagnosis accuracy. In complex fault scenarios, it is difficult to provide comprehensive and accurate diagnostic results, impacting the efficiency and accuracy of fault troubleshooting.

[0003] In summary, existing technologies suffer from the problem that fault diagnosis of substation equipment can only be accurate to the specific equipment, but cannot accurately locate the equipment components, resulting in poor fault diagnosis accuracy and further affecting the efficiency of fault troubleshooting. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for fault diagnosis of substation equipment based on virtual mapping, in order to solve the technical problem that in the prior art, fault diagnosis of substation equipment can only be accurate to the specific equipment and cannot accurately locate the equipment components, resulting in poor fault diagnosis accuracy and further affecting the efficiency of fault troubleshooting.

[0005] To achieve the above objectives, this application provides a method and system for diagnosing substation equipment faults based on virtual mapping.

[0006] Firstly, this application provides a substation equipment fault diagnosis method based on virtual mapping. This method is implemented through a substation equipment fault diagnosis system based on virtual mapping. The method includes: retrieving a component-level CAD model from a geometric model library based on the equipment ID of the target substation equipment; retrieving physical field control parameters from a multiphysics parameter library based on the equipment ID; performing multiphysics coupling parameter assignment on the component-level CAD model to obtain a multiphysics coupling simulation model; and retrieving historical fault causes to determine component fault frequency. Statistical analysis is performed to locate and encode P key components. Based on the P key component codes, P multimodal heterogeneous sensing units are configured on the P equipment components of the target substation equipment body. During the operation of the target substation equipment, the operating states of the P equipment components are mapped to P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units. A directed graph simulation of fault propagation is performed in the multiphysics coupling simulation model, and real-time fault diagnosis results are output. The real-time fault diagnosis results include the fault root cause component code, component fault probability estimate, and fault component propagation chain.

[0007] Optionally, the physical field control parameters are decomposed according to the physical domain to obtain multivariate correlation parameters; the component-level CAD model is segmented based on physical characteristics to obtain conductor geometry, insulation geometry, fluid geometry, and structural geometry; the multivariate physical control equations of the conductor geometry, insulation geometry, fluid geometry, and structural geometry are mapped and matched; according to parameter-geometry binding rules, the multivariate correlation parameters are split and bound to the conductor geometry, insulation geometry, fluid geometry, and structural geometry, and the coefficient configuration of the multivariate physical control equations is executed in the conductor geometry, insulation geometry, fluid geometry, and structural geometry; coupling terms are loaded at the geometric interfaces of the conductor geometry, insulation geometry, fluid geometry, and structural geometry to generate an initial simulation model; the initial simulation model is discretized by computational mesh to output the multiphysics coupling simulation model.

[0008] Optionally, the multivariate correlation parameters include electromagnetic correlation parameters, thermodynamic correlation parameters, structural mechanics correlation parameters, and coupling relationship correlation parameters.

[0009] Optionally, constrained by the operating scenario of the target substation equipment, multi-source operation and maintenance records are retrieved based on the equipment ID; the causes of time-series faults are retrieved from the multi-source operation and maintenance records to obtain the historical causes of faults; component fault frequency records are recorded based on the historical causes of faults to obtain multiple time-series fault records for multiple equipment components; multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs are calculated and output based on the multiple time-series fault records; component weights are dynamically configured according to the equipment attributes of the target substation equipment, and the multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs are weighted and quantified to output multiple component fault risk values; based on the descending order of the multiple component fault risk values, the P key component codes are located in the multiple equipment components.

[0010] Optionally, based on the P key component codes, P fault anomaly correlation parameter groups of the P device components are retrieved, wherein each fault anomaly correlation parameter group includes at least one fault-sensitive physical quantity; sensor type matching is performed based on the fault-sensitive physical quantity composition of the P fault anomaly correlation parameter groups to obtain the P multimodal heterogeneous sensing units; the P multimodal heterogeneous sensing units are deployed on the P device components, and a real-time data channel is established between the P multimodal heterogeneous sensing units and the P virtual component sub-models.

[0011] Optionally, based on the physical coupling relationship of the P equipment components in the target substation equipment, multiple coupled topologies are established; after the P virtual component sub-models receive the P multimodal operating status data returned by the P multimodal heterogeneous sensing units, they perform local anomaly identification and output P fault state confidence vectors; the P fault state confidence vectors are loaded into the multiple coupled topologies, and a directed graph simulation of fault propagation is performed to output multiple initial fault propagation chains; the maximum a posteriori probability estimation is performed on the multiple initial fault propagation chains to output the fault component propagation chain; the fault root cause component code is located in the fault component propagation chain, and the fault state confidence vector is matched according to the fault root cause component code as the component fault probability estimate.

[0012] Optionally, the system interactively obtains multiple operational status data records of the first device component under multiple fault scenarios; constructs multiple fault similarity analysis models based on the multiple operational status data records; constructs a first fault confidence network by connecting the multiple fault similarity analysis models in parallel; dynamically deploys the first fault confidence network to a first virtual component sub-model; the first virtual component sub-model receives and inputs the first multimodal operational status data into the first fault confidence network, and outputs multiple fault confidence scores through parallel analysis by the multiple fault similarity analysis models; and fuses the multiple fault confidence scores to output a first fault state confidence vector.

[0013] Optionally, using the first coupled-association topology as a dynamic update framework, after loading the P fault state confidence vectors into the multiphysics coupled simulation model, the directed edge weights of the first coupled-association topology are calculated based on the P real-time physical field gradient states output by the multiphysics coupled simulation model, and the first updated-association topology is output. The P fault state confidence vectors are traversed using a preset confidence threshold, and the output directed edges are activated in the first updated-association topology to locate N associated directed edges. The propagation intensity values ​​of the N fault state confidence vectors are calculated along the N associated directed edges to generate the first initial fault propagation chain.

[0014] Optionally, the spatial coordinates of the terminal node of the first initial fault propagation chain are extracted, and the transient simulation verification of the multiphysics coupling simulation model is performed to output the first transient simulation propagation chain. If the deviation between the first initial fault propagation chain and the first transient simulation propagation chain meets the preset tolerance threshold, the first initial fault propagation chain is retained. If the deviation between the first initial fault propagation chain and the first transient simulation propagation chain exceeds the preset tolerance threshold, the first transient simulation propagation chain is used to replace the first initial fault propagation chain.

[0015] Secondly, this application also provides a substation equipment fault diagnosis system based on virtual mapping, used to execute the substation equipment fault diagnosis method based on virtual mapping as described in the first aspect, wherein the substation equipment fault diagnosis system based on virtual mapping includes: a model retrieval module, used to retrieve a component-level CAD model from a geometric model library according to the equipment ID of the target substation equipment; a coupling parameter assignment module, used to retrieve physical field control parameters from a multiphysics parameter library according to the equipment ID, perform multiphysics coupling parameter assignment on the component-level CAD model, and obtain a multiphysics coupling simulation model; and a frequency statistics module, used to retrieve historical fault causes to perform component fault frequency statistics and locate faults. The system includes: a P key component encoding module, a component configuration module, and a virtual mapping module, which maps the operating states of the P equipment components to P virtual component sub-models in the multiphysics coupling simulation model during the operation of the target substation equipment using the P multiphysics coupling simulation models; and a fault diagnosis module, which performs directed graph simulation of fault propagation in the multiphysics coupling simulation model and outputs real-time fault diagnosis results, wherein the real-time fault diagnosis results include fault root cause component encoding, component fault probability estimation, and fault component propagation chain.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages: The component-level CAD model is retrieved from the geometric model library based on the equipment ID of the target substation equipment. Physical field control parameters are retrieved from the multiphysics parameter library based on the equipment ID, and multiphysics coupling parameters are assigned to the component-level CAD model to obtain a multiphysics coupling simulation model. Historical fault causes are retrieved to perform component fault frequency statistics, and P key component codes are located. Based on the P key component codes, P multimodal heterogeneous sensing units are configured on the P equipment components of the target substation equipment. During the operation of the target substation equipment, the operating states of the P equipment components are mapped to P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units. A directed graph simulation of fault propagation is performed in the multiphysics coupling simulation model, and real-time fault diagnosis results are output. These real-time fault diagnosis results include the fault root cause component code, component fault probability estimation, and fault component propagation chain. In other words, by constructing a component-level digital twin that runs synchronously with the physical device, the operating data of the physical device is loaded into the virtual device, and operational fault diagnosis at the component level of the physical device is performed. This achieves a leap from device-level early warning to component-level precision diagnosis, improves the accuracy of fault diagnosis, and thus enhances the efficiency of fault troubleshooting.

[0017] 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

[0018] 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.

[0019] Figure 1 This is a flowchart illustrating the substation equipment fault diagnosis method based on virtual mapping proposed in this application.

[0020] Figure 2 This is a schematic diagram of the substation equipment fault diagnosis system based on virtual mapping in this application.

[0021] Figure labeling: Model retrieval module 11, Coupling parameter assignment module 12, Frequency statistics module 13, Component configuration module 14, Virtual mapping module 15, Fault diagnosis module 16. Detailed Implementation

[0022] This application provides a method and system for fault diagnosis of substation equipment based on virtual mapping. It addresses the technical problem in existing technologies where fault diagnosis can only pinpoint specific equipment, failing to accurately locate equipment components, leading to poor fault diagnosis accuracy and further impacting troubleshooting efficiency. By constructing a component-level digital twin that operates synchronously with the physical equipment, the operating data of the physical equipment is loaded into the virtual equipment, enabling operational fault diagnosis at the component level. This achieves a leap from equipment-level early warning to component-level precision diagnosis, improving fault diagnosis accuracy and thus enhancing troubleshooting efficiency.

[0023] 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.

[0024] Example 1, please refer to the appendix. Figure 1 This application provides a substation equipment fault diagnosis method based on virtual mapping, wherein the substation equipment fault diagnosis method based on virtual mapping is applied to a substation equipment fault diagnosis system based on virtual mapping, and the substation equipment fault diagnosis method based on virtual mapping specifically includes the following steps: The component-level CAD model is retrieved from the geometric model library based on the equipment ID of the target substation equipment.

[0025] Specifically, in a substation, each piece of equipment typically has a unique identifier called an equipment ID. The equipment ID of the target substation equipment is obtained and retrieved from the geometric model library. The equipment ID is a unique identifier for the equipment in the digital management system, usually a structured code, such as SS-ZB-220-T001, where SS represents the substation, ZB represents the main transformer, 220 represents the voltage level, and T001 is the serial number. The geometric model library is a database of three-dimensional digital models storing all substation equipment and its components. It is not a simple image library, but a computable collection of geometric models containing precise dimensions, assembly relationships, and material information.

[0026] For complex equipment in substations, each component, such as the transformer's tank, windings, and cooler, has a corresponding CAD model. Component-level CAD models contain the geometric dimensions, shape, and design details of each component. By retrieving the component-level CAD model from the geometric model library using the device ID, the geometric structure and dimensions of each component of the substation equipment can be accurately obtained.

[0027] Based on the device ID, physical field control parameters are retrieved from the multiphysics parameter library, and multiphysics coupling parameters are assigned to the component-level CAD model to obtain a multiphysics coupling simulation model.

[0028] Furthermore, this application also includes the following steps: decomposing the physical field control parameters according to the physical domain to obtain multivariate correlation parameters; segmenting the component-level CAD model based on physical characteristics to obtain conductor geometry, insulation geometry, fluid geometry, and structural geometry; mapping and matching the multivariate physical control equations of the conductor geometry, insulation geometry, fluid geometry, and structural geometry; according to parameter-geometry binding rules, splitting and binding the multivariate correlation parameters to the conductor geometry, insulation geometry, fluid geometry, and structural geometry, and then executing the coefficient configuration of the multivariate physical control equations in the conductor geometry, insulation geometry, fluid geometry, and structural geometry; loading coupling terms at the geometric domain interfaces of the conductor geometry, insulation geometry, fluid geometry, and structural geometry to generate an initial simulation model; and discretizing the initial simulation model using a computational mesh to output the multiphysics coupling simulation model.

[0029] Furthermore, this application also includes the following steps: the multivariate correlation parameters include electromagnetic correlation parameters, thermodynamic correlation parameters, structural mechanics correlation parameters, and coupling relationship correlation parameters.

[0030] Specifically, the physical field control parameters of the target device are retrieved from the multiphysics parameter library using the device ID. The multiphysics parameter library is a database storing various physical field control parameters and the physical characteristics of different devices, used to analyze device behavior in different physical environments. The physical field control parameters are decomposed according to the physical domain to obtain multivariate correlation parameters, including electromagnetic correlation parameters, thermodynamic correlation parameters, structural mechanics correlation parameters, and coupling correlation parameters. Electromagnetic field control parameters include current, voltage, and electromagnetic field strength; thermodynamic field control parameters include temperature, heat flux, and thermal conductivity; structural mechanics field control parameters include stress, strain, and displacement; and coupling correlation parameters represent the coupling relationships between different physical fields, such as thermoelectric effects and electromagnetic coupling. For example, the electrical conductivity of the copper winding is 5.8e7 S / m; the relative permeability of the silicon steel core sheet has a non-linear BH curve, approximately 1500 at 1.5T; copper has a density of 8960 kg / m³ and a specific heat capacity of 385 J / (kg·K); the thermal conductivity of transformer oil is 0.12 W / (m·K); the Young's modulus of the tank steel is 2.0e11 Pa, and the Poisson's ratio is 0.29; the Joule thermal coupling coefficient of copper is 1.0 (indicating that all electromagnetic losses are converted into heat); and the coefficient of thermal expansion of steel is 1.2e -5 1 / K.

[0031] Based on the physical characteristics of the equipment, the component-level CAD model of the target equipment is decomposed according to its physical characteristics, forming multiple geometric domains, including conductor geometry, insulation geometry, fluid geometry, and structural geometry. The conductor geometry is the region composed of conductive materials (such as copper and aluminum), such as windings; the insulation geometry is the region composed of insulating materials (such as insulating paper, epoxy resin, and SF6 gas); the fluid geometry is the region occupied by fluids (such as transformer oil and air); and the structural geometry mainly consists of structural components that bear mechanical loads, such as transformer tanks, clamps, and cores.

[0032] This involves mapping and matching multivariate physical governing equations to the geometric domains of conductors, insulation, fluids, and structures. These multivariate physical governing equations are mathematical equations describing the intrinsic laws of physical fields, such as Maxwell's equations for electromagnetic fields, Fourier's law for heat conduction, and the equilibrium equations of solid mechanics in structural mechanics. For example, assuming a transformer in a substation is to be simulated, different governing equations are applied to different geometric domains of the transformer based on its physical characteristics. Electromagnetic fields play a dominant role in the conductor geometric domain. Taking transformer windings as an example, the distribution of current and voltage can be described by Maxwell's equations. Specifically, the current distribution can be calculated using Ampere's law, and the electric field distribution can be described by Gauss's law. The electric field distribution of the insulating material can be described by the potential equation. Assuming the insulating oil region of the transformer, the electric field distribution is calculated using Poisson's equation, and the relationship between the electric field and voltage can be expressed by the potential. The flow of cooling oil in the transformer involves thermodynamic governing equations, using the superheat conduction equation to describe the temperature distribution of the oil flow. The relationship between the temperature and flow rate of the cooling system can be calculated using heat flux density.

[0033] Based on the parameter-geometry domain binding rules, the retrieved parameters are precisely assigned to the corresponding domains as equation coefficients. Electromagnetic correlation parameters are bound to the conductor and insulation geometries to ensure accurate simulation of current, voltage, and electromagnetic field strength within the conductor region. Thermodynamic correlation parameters are bound to all geometries, including the conductor, insulation, fluid, and structural geometries. Heat is transferred from the current-flowing conductor to the insulating material, and then to the outer shell through the oil flow region. The temperature field in each geometries needs to be accurately calculated using thermal equations. Structural mechanics correlation parameters are bound to the structural geometries to ensure accurate simulation of the equipment's mechanical stress, strain, and displacement. Coupling correlation parameters are applied to the interfaces between geometries. The coefficient configuration of the physical control equations is executed to ensure that correlations and couplings between different physical domains are handled correctly.

[0034] Coupling terms are added to the interfaces of the conductor, insulation, fluid, and structural geometries to generate the initial simulation model. For example, at the interface between the transformer winding and coolant, the electromagnetic and thermal fields may interact, thus requiring the addition of electromagnetic-thermal coupling terms. Joule thermal coupling is established at the interface between the winding (conductor domain) and the insulating paper (insulation domain), transferring electromagnetic losses as a heat source to the thermal field analysis. Coupling terms are mathematical expressions added to the governing equations to describe how a variable in one physical field affects the source terms of another physical field.

[0035] The initial simulation model is discretized into a computational mesh, enabling numerical computation on a computer. Mesh discretization divides the entire model into multiple small computational units, typically using the finite element method, resulting in a multiphysics coupled simulation model capable of simulating the operation of equipment under the interaction of multiple physical fields such as electromagnetics, thermodynamics, and mechanics.

[0036] Retrieve historical fault causes to perform component fault frequency statistics and locate the codes of P key components.

[0037] Furthermore, this application also includes the following steps: using the operating scenario of the target substation equipment as a constraint, retrieving multi-source operation and maintenance records based on the equipment ID; retrieving the causes of time-series faults from the multi-source operation and maintenance records to obtain the historical causes of faults; recording component fault frequencies based on the historical causes of faults to obtain multiple time-series fault records for multiple equipment components; calculating and outputting multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs based on the multiple time-series fault records; dynamically configuring component weights according to the equipment attributes of the target substation equipment, performing weighted quantization of the multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs, and outputting multiple component fault risk values; and locating the P key component codes in the multiple equipment components based on the descending order of the multiple component fault risk values.

[0038] Specifically, the operating scenario of the target substation equipment is determined, which refers to the specific working environment and conditions in which the target substation equipment operates, including but not limited to voltage level, load level, environmental conditions, and its role in the power grid. Using the operating scenario of the target substation equipment as a constraint, relevant multi-source maintenance records are retrieved based on the equipment ID and its operating scenario. These multi-source maintenance records are data records from different maintenance sources, including equipment maintenance history, testing data, fault logs, and inspection records, reflecting the maintenance status and fault conditions of the target substation equipment.

[0039] From multi-source operation and maintenance records, time-series fault cause information is extracted to obtain historical fault causes. Time-series fault causes refer to fault events and their root cause analyses extracted from the operation and maintenance records and arranged in chronological order, such as a fault caused by overload or equipment damage due to high temperature. Based on historical fault causes, component fault frequency records are compiled, and the fault frequency of each equipment component is statistically analyzed, resulting in multiple time-series fault records for multiple equipment components. That is, for each specific equipment component, a record is compiled chronologically of all historical fault events. Based on multiple time-series fault records, multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs are calculated and output. Fault frequency is the number of times a component fails within a statistical time window; fault cycle average is the average time period for equipment or its components to fail within a certain time range; fault operation and maintenance costs are the expenses required for repair and maintenance after equipment failure, including repair costs, downtime costs, labor costs, and spare parts costs. For example, component A experienced 4 failures with an average failure interval (average cycle) of 2.5 years and an average maintenance cost of 150,000 per failure (including spare parts and power outage losses); component B experienced 2 failures with an average failure interval of 5 years, but the maintenance cost per failure was as high as 800,000 due to severe damage and prolonged power outage caused by bushing explosion; component C experienced 8 failures with an average failure interval of 0.75 years, but the maintenance cost per failure was only 5,000, which did not affect the operation of the main transformer and could be replaced online.

[0040] Based on equipment attributes and component importance, different weights are assigned to each component. For example, windings are highly critical in transformers and therefore receive a higher weight, while oil cooling systems may have a slightly lower weight; for instance, component A might have a weight of 0.7, component B a weight of 1.0, and component C a weight of 0.3. The fault risk value is calculated by weighting fault frequency, average fault cycle, and maintenance cost with the component weights to obtain the fault risk value for multiple components of the target substation equipment. Fault Risk Value = a Fault frequency + b (1 / mean of failure cycle) + c Fault maintenance cost, where a, b, and c are the weights of fault frequency, average fault cycle, and maintenance cost configuration. For example, if the weights of fault frequency, average fault cycle, and maintenance cost configuration are 0.4, 0.3, and 0.3, respectively, then the fault risk value of component A is 0.7. (0.4 4 + 0.3 (1 / 2.5) + 0.3 (150000 / 1000000))=1.43, the failure risk value of component B=1.24, and the failure risk value of component C=1.36.

[0041] Components are sorted in descending order based on their failure risk values, and the top P components are identified as the P critical component codes. These P critical component codes typically represent components with high failure frequency and significant impact, requiring priority monitoring and maintenance. P is a positive integer greater than or equal to 1. By dynamically configuring component weights and using weighted quantization, the critical components with the greatest impact on equipment operation are accurately identified, thereby optimizing resource allocation and maintenance priorities.

[0042] Based on the coding of the P key components, P multimodal heterogeneous sensing units are configured in the P equipment components of the target substation equipment body.

[0043] Furthermore, this application also includes the following steps: based on the P key component codes, retrieve the P fault anomaly association parameter groups of the P device components, wherein each fault anomaly association parameter group includes at least one fault-sensitive physical quantity; perform sensor type matching based on the fault-sensitive physical quantity composition of the P fault anomaly association parameter groups to obtain the P multimodal heterogeneous sensing units; deploy the P multimodal heterogeneous sensing units on the P device components, and establish a real-time data channel between the P multimodal heterogeneous sensing units and the P virtual component sub-models.

[0044] Specifically, based on the codes of P key components, P fault-anomaly correlation parameter groups are retrieved from the P equipment components of the target substation equipment body. Each fault-anomaly correlation parameter group includes at least one fault-sensitive physical quantity, listing all common fault modes of the component and their corresponding fault-sensitive physical quantities. Fault-sensitive physical quantities are physical quantities that can sensitively reflect equipment or component faults, such as temperature, vibration, pressure, and power. These physical quantities usually change significantly when a fault occurs.

[0045] Based on the fault-sensitive physical quantities in the fault anomaly correlation parameter group of each key component, appropriate sensors are selected to monitor these physical quantities. The sensor type matching process is crucial because different physical quantities require different types of sensors. To monitor fault-sensitive physical quantities, it is necessary to select appropriate sensor types. The sensor type matching process involves selecting suitable sensors to detect these physical quantities based on their properties. Sensors can be thermocouples, pressure sensors, accelerometers, etc. For example, component B deploys distributed fiber optic temperature sensors along the switching oil chamber wall, with a measurement range of 0-150℃ and an accuracy of ±0.5℃; an online oil chromatography monitoring unit is integrated in the oil pipeline, with a detection limit of 0.5μL / L; and a high-frequency acceleration vibration sensor is installed, with a frequency range of 1Hz-10kHz.

[0046] P multimodal heterogeneous sensing units are sensor units capable of simultaneously acquiring multiple types of information from P device components. They typically contain different types of sensors and can detect multiple physical quantities simultaneously. P corresponding multimodal heterogeneous sensing units are deployed at appropriate locations on the P device components to monitor the device's operating status in real time. A real-time data channel is established between the P multimodal heterogeneous sensing units and P virtual component sub-models for data transmission. The virtual component sub-models are virtual models created based on digital modeling and simulation technology of the device. They are used to simulate the behavior and performance of each component, serving as digital twins of the device for real-time monitoring of its status. Establishing a real-time data channel between the sensors and the virtual component sub-models ensures that data collected from the sensors is transmitted in real time to the virtual model, enabling the virtual component sub-models to update the device status based on real-time data, allowing for dynamic simulation and fault diagnosis.

[0047] During the operation of the target substation equipment, the operating status of the P equipment components is mapped to the P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units.

[0048] Specifically, during the operation of the target substation equipment, P multimodal heterogeneous sensing units continuously collect the operating status of P equipment components. This data is then transmitted to P virtual component sub-models within a multiphysics coupled simulation model via an established real-time data channel. The multiphysics coupled simulation model updates the virtual component sub-models in real-time based on this data, simulating the equipment's behavior under its current state. Each virtual component sub-model, upon receiving its specific real-time data, immediately uses it as input to drive local or global physics simulation calculations. This process is not simply data mapping; rather, it uses real data to correct and drive the simulation model's operating state, enabling the virtual model to dynamically and faithfully reflect the current state of the physical equipment, and even predict its internal, invisible states, such as internal temperature and stress field distributions. For example, a distributed fiber optic temperature sensor installed on the wall of the tap changer oil chamber detected that the temperature of a hot spot rose from a steady-state 85°C to 92°C; an online oil chromatography unit detected that the hydrogen concentration slowly increased from 85 μL / L to 110 μL / L, and acetylene appeared at 3 μL / L; a high-frequency vibration sensor detected that the effective vibration value increased from 0.8 m / s² to 1.5 m / s², and a new spectral peak appeared at 1200 Hz; these data were timestamped and component-coded, and transmitted to the digital twin within 100 milliseconds via industrial Ethernet.

[0049] By mapping component status to the virtual sub-model in real time, the operation of the equipment can be dynamically monitored and potential faults can be detected in a timely manner. Through the dynamic updating of the virtual model, the operating status and health status of the equipment are more comprehensively assessed, which helps maintenance personnel to take appropriate maintenance measures before equipment problems occur, extend the service life of the equipment and reduce the risk of sudden failures.

[0050] The multiphysics coupled simulation model performs a directed graph simulation of fault propagation and outputs real-time fault diagnosis results, which include fault root cause component encoding, component fault probability estimation, and fault component propagation chain.

[0051] Furthermore, this application also includes the following steps: establishing multiple coupled topologies based on the physical coupling relationships of the P equipment components in the target substation equipment; after the P virtual component sub-models receive the P multimodal operating status data returned by the P multimodal heterogeneous sensing units, they perform local anomaly identification and output P fault state confidence vectors; loading the P fault state confidence vectors into the multiple coupled topologies, performing directed graph simulation of fault propagation, and outputting multiple initial fault propagation chains; solving the maximum a posteriori probability estimation for the multiple initial fault propagation chains and outputting the fault component propagation chain; tracing and locating the fault root cause component code in the fault component propagation chain, and matching the fault state confidence vector based on the fault root cause component code as the component fault probability estimate.

[0052] Furthermore, this application also includes the following steps: interactively obtaining multiple operating status data records of the first device component under multiple fault scenarios; constructing multiple fault similarity analysis models based on the multiple operating status data records; constructing a first fault confidence network by connecting the multiple fault similarity analysis models in parallel; dynamically deploying the first fault confidence network to a first virtual component sub-model; the first virtual component sub-model receiving and inputting first multimodal operating status data into the first fault confidence network, and outputting multiple fault confidence scores through parallel analysis of the multiple fault similarity analysis models; fusing the multiple fault confidence scores to output a first fault state confidence vector.

[0053] Specifically, based on the physical coupling relationships of P equipment components within the target substation equipment, a corresponding coupling association topology is established to describe how they influence each other. Physical coupling relationships refer to the mutual influence relationships between different equipment components during operation. The coupling association topology, based on the physical coupling relationships of the equipment components, constructs the connection and influence paths between each component; that is, it represents the physical interactions between equipment components through a topological structure.

[0054] Randomly select one of the P device components as the first device component. Interact with the system to obtain multiple operational status data records of the first device component under various fault scenarios. Retrieve a large amount of operational status data under different fault scenarios from its historical database to obtain multiple operational status data records. The operational status data records are data collected by sensors about the device component under different operating states, including records of changes in physical quantities such as temperature, current, pressure, and vibration, reflecting the health status and operating environment of the device component.

[0055] Based on multiple operational status data, a dedicated fault similarity analysis model is trained for each fault scenario. Each model is trained using a large amount of historical operational status data records for the corresponding fault scenario, thereby learning to distinguish the unique data patterns of that fault from normal data. For example, for the bearing wear model, 4500 sets of historical vibration data records are used for training, including the changes in high-frequency vibration energy (e.g., 1500Hz-2000Hz) during the process of bearing wear from healthy to severely worn. The trained model can then determine the similarity between the current data and the bearing wear pattern. For the blade imbalance model, 3800 sets of historical data are used for training, focusing on the amplitude of the vibration signal at one rotational frequency. The trained logistic regression model can effectively identify imbalance characteristics.

[0056] Multiple trained fault similarity analysis models are integrated in parallel to form a unified first fault confidence network. This involves combining multiple independent fault similarity analysis models in parallel, sharing the same input interface, but each performing calculations and judgments independently, outputting analysis results specific to its assigned fault type. Each analysis model outputs a confidence level for a particular fault mode based on the current operating status of the equipment. The first fault confidence network is dynamically deployed to the first virtual component sub-model. The virtual component sub-model receives real-time multimodal operating status data from the equipment and inputs this data into the fault confidence network for analysis, enabling real-time monitoring of equipment status and fault prediction.

[0057] The first virtual component sub-model receives and inputs the first multimodal operating status data into the first fault confidence network. Multiple fault similarity analysis models analyze the data in parallel, outputting multiple fault confidence scores, thus obtaining fault confidence scores for different fault scenarios. These multiple fault confidence scores are then fused to output a first fault state confidence vector, reflecting the probability of occurrence of each fault mode of the first device component and assessing its health status. For P device components, each corresponding virtual component sub-model, after receiving and processing P multimodal operating status data, will perform local anomaly identification and output P fault state confidence vectors.

[0058] Furthermore, this application also includes the following steps: using the first coupled-association topology as a dynamic update framework, after loading the P fault state confidence vectors into the multiphysics coupled simulation model, and based on the P real-time physical field gradient states output by the multiphysics coupled simulation model, calculating the directed edge weights of the first coupled-association topology, and outputting the first updated-association topology; traversing the P fault state confidence vectors using a preset confidence threshold, activating the output directed edges in the first updated-association topology, and locating N associated directed edges; calculating the propagation intensity values ​​of the N fault state confidence vectors along the N associated directed edges, and generating the first initial fault propagation chain.

[0059] Furthermore, this application also includes the following steps: extracting the spatial coordinates of the terminal node of the first initial fault propagation chain, performing transient simulation verification of the multiphysics coupling simulation model, and outputting the first transient simulation propagation chain; if the deviation between the first initial fault propagation chain and the first transient simulation propagation chain meets a preset tolerance threshold, then the first initial fault propagation chain is retained; if the deviation between the first initial fault propagation chain and the first transient simulation propagation chain exceeds the preset tolerance threshold, then the first transient simulation propagation chain is used to replace the first initial fault propagation chain.

[0060] Specifically, P fault state confidence vectors are loaded into multiple coupled topologies to perform directed graph simulation of fault propagation, outputting multiple initial fault propagation chains. Taking the first coupled topology as an example, it serves as a dynamically updated framework. The P fault state confidence vectors are loaded into the multiphysics coupled simulation model. Each component's fault state confidence vector represents its fault probability or health status. After being input into the simulation model, the model outputs the real-time physical field gradient state of the component based on this information. The real-time physical field gradient state is field distribution information reflecting the drastic changes in physical quantities, calculated and output by the multiphysics coupled simulation model. For example, a temperature gradient can indicate the intensity and direction of heat flow; a stress gradient can indicate the transmission path of mechanical loads. These gradients are the physical driving forces of fault propagation.

[0061] The weight of each directed edge in the topology is dynamically calculated based on the physical field gradient, thus outputting a first updated associated topology that better reflects the current operating conditions. The weight of each directed edge represents the strength of the physical interaction between components, such as the strength of heat conduction, stress transfer, or current coupling. In the associated topology, the directed edge weight is used to quantify the ease or probability of a fault propagation path represented by a directed edge. The higher the weight, the greater the risk of a fault propagating along this path.

[0062] Using a pre-set confidence threshold, P fault state confidence vectors are traversed. Directed edges are activated in the first updated associated topology, locating N associated directed edges. In other words, based on the pre-set confidence threshold, all confidence vectors are scanned, all faulty components exceeding the pre-set confidence threshold are located, and directed edges originating from these components are activated in the updated first associated topology. The propagation strength value of the fault state confidence vectors is calculated along the activated N associated directed edges, generating the first initial fault propagation chain. The propagation strength value represents the degree of impact of fault information propagation in the device, i.e., how a fault in one component affects other components. This propagation process is progressively transmitted, thus helping to locate the fault propagation path. N is a positive integer greater than or equal to 1 and less than or equal to P. The propagation strength value is a quantitative indicator that combines the fault source confidence and the directed edge weights, used to measure the strength or risk of a fault propagating along an activated edge to the next component.

[0063] Repeat the above steps for multiple coupled topologies to ultimately output multiple initial fault propagation chains to cover different physical mechanisms of fault propagation. For example, component A has an overheating confidence of 0.88. The multiphysics model output shows a temperature gradient of up to 45°C / m from the tap changer to the adjacent insulating paper. Based on this temperature gradient, the weight of the tap changer → insulating paper edge in the topology is dynamically increased from the baseline value of 0.6 to 0.95. The first updated coupled topology is output. Setting the confidence threshold to 0.75, after traversal, it is found that the tap changer (contact overheating 0.88) exceeds the threshold, so all its edges are activated, especially the tap changer → insulating paper edge with updated weights. The propagation intensity along the tap changer → insulating paper edge = 0.88 (confidence) × 0.95 (edge ​​weight) = 0.836. The first initial propagation chain is generated: tap changer 0.836 → insulating paper (thermal aging risk). The same confidence vector is loaded into other coupled topologies to obtain multiple initial fault propagation chains.

[0064] By introducing real-time physical field gradients to dynamically update the propagation model, the inference process closely aligns with the actual physical state of the equipment, significantly improving prediction accuracy and physical reliability. By setting confidence thresholds and calculating propagation strength, the system focuses on high-risk fault sources and propagation paths, quantifying propagation risks and effectively preventing alarm overload.

[0065] The spatial coordinates of the terminal node of the first initial fault propagation chain are extracted, representing the spatial location coordinates of the last device component in the chain. The terminal node typically marks the end of the chain, indicating which devices or components the fault may ultimately affect. The spatial coordinates of the terminal node are input into a multiphysics coupled simulation model, and transient simulation is performed on that node to determine the dynamic changes of the device components during operation, including the time-varying values ​​of physical quantities such as temperature, current, and pressure. The simulation process verifies whether the fault propagation within the equipment conforms to actual conditions, particularly how the fault further expands or weakens over time. The first transient simulation propagation chain, obtained through transient simulation, is compared with the initial fault propagation chain to determine their consistency. The transient simulation propagation chain considers the impact of time variations on fault propagation and is based on the dynamic calculation results of the multiphysics simulation model.

[0066] Compare the differences between the first initial fault propagation chain and the first transient simulation propagation chain. Calculate the deviation between the two to determine their similarity. If the deviation is less than a preset tolerance threshold, the initial fault propagation chain is considered reasonable and can be retained. If the deviation is greater than the tolerance threshold, the first transient simulation propagation chain needs to replace the first initial fault propagation chain. The preset tolerance threshold is a predefined allowable error range used to determine whether two propagation chains are consistent. For example, a propagation path overlap of 90% or a propagation time error of less than 10% can be set as within the tolerance range. Repeat the above steps for multiple initial fault propagation chains to obtain the final multiple initial fault propagation chains.

[0067] The most probable fault propagation chain is determined by performing maximum a posteriori probability estimation (MAP) on multiple initial fault propagation chains. The probability of each propagation chain's occurrence is assessed based on historical data, operating environment, and current state. Ultimately, the most probable fault component propagation chain is output, representing the most likely fault propagation path. MAP is a Bayesian statistical method used to calculate the most likely hypothesis by considering prior knowledge (such as historical fault probabilities) and combining it with currently observed evidence (such as multiple initial fault propagation chains and their strengths). In this scenario, it involves identifying the most probable and credible fault propagation chain from multiple possible chains. The fault component propagation chain, determined by MAP, is the final and unique fault propagation path, representing the final and most reliable output of the entire diagnostic process, indicating the complete impact path of the fault from its source to its endpoint. For example, chain A (thermal propagation path): tap changer (contact overheating) → insulation paper (thermal aging). The prior probability (based on historical data) of the first chain is 0.3, and the current propagation intensity is 0.85. Chain B (electric-stress propagation path): tap changer (contact arc) → high-voltage winding (electromagnetic force impact) → clamp (loosening). The prior probability of this chain is 0.1, but the current propagation intensity is 0.90. Calculating the posterior probability, the posterior probability of chain A is 0.255, and the posterior probability of chain B is 0.090. The posterior probability of chain A (0.255) is much higher than that of chain B (0.090), therefore chain A is ultimately determined to be the more reliable fault component propagation chain. Tracing backward along the final chain A (tap changer → insulation paper), the root cause component of the fault is located. The fault state confidence vector generated before the tap changer component is retrieved is [contact overheating 0.88, mechanical jamming 0.15, seal failure 0.02]. The highest value of 0.88 is taken as the component fault probability estimate. The final diagnostic report output includes the root cause component code: tap changer; component failure probability estimate of 0.88, i.e. 88% certainty; failure component propagation chain: tap changer (contact overheating) → insulation paper (thermal aging risk).

[0068] Locating the root cause component code in fault propagation chain tracing involves tracing the starting point of the fault propagation chain to find the source of the fault. The source of the fault is usually the equipment or component where the fault initially occurred. According to the tracing process, the root cause component code is located, which indicates the component in the equipment that initially failed. The root cause component code indicates the location or component where the fault initially occurred. By tracing the starting point of the fault propagation chain, the equipment or component that initially caused the fault is located, and a unique code for that component is generated.

[0069] Based on the component encoding of the root cause of the fault, a fault state confidence vector is matched as the component fault probability estimate. The component fault probability estimate is an estimate of the probability that a device or component will fail at a certain moment.

[0070] By integrating the root cause component coding, component failure probability estimation, and failure propagation chain, real-time fault diagnosis results are obtained. The root cause component coding is the unique code for the initial location or component where the fault occurred. The tracing process identifies the component that initially caused the fault by analyzing the propagation chain. The component failure probability estimation calculates the probability of a component failing—the likelihood of that component failing—using the fault state confidence vector and maximum a posteriori probability estimation. The failure propagation chain represents the path of fault propagation. By coupling the associated topology and maximum a posteriori probability estimation, the source of the fault can be accurately located, ensuring the accuracy of the diagnostic results. The combination of fault propagation chain simulation and fault state confidence vector allows for a comprehensive assessment of equipment failure risk, timely detection of equipment problems, and targeted maintenance, reducing downtime and repair costs.

[0071] In summary, the substation equipment fault diagnosis method based on virtual mapping provided in this application has the following technical effects: A component-level CAD model is retrieved from a geometric model library based on the equipment ID of the target substation equipment; physical field control parameters are retrieved from a multiphysics parameter library based on the equipment ID, and multiphysics coupling parameters are assigned to the component-level CAD model to obtain a multiphysics coupling simulation model; historical fault causes are retrieved to perform component fault frequency statistics, and P key component codes are located; based on the P key component codes, P multimodal heterogeneous sensing units are configured on the P equipment components of the target substation equipment; during the operation of the target substation equipment, the operating states of the P equipment components are mapped to P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units; a directed graph simulation of fault propagation is performed in the multiphysics coupling simulation model, and real-time fault diagnosis results are output, wherein the real-time fault diagnosis results include fault root cause component codes, component fault probability estimates, and fault component propagation chains. In other words, by constructing a component-level digital twin that runs synchronously with the physical device, the operating data of the physical device is loaded into the virtual device, and operational fault diagnosis at the component level of the physical device is performed. This achieves a leap from device-level early warning to component-level precision diagnosis, improves the accuracy of fault diagnosis, and thus enhances the efficiency of fault troubleshooting.

[0072] Example 2: Based on the same inventive concept as the substation equipment fault diagnosis method based on virtual mapping in Example 1, this application also provides a substation equipment fault diagnosis system based on virtual mapping. Please refer to the appendix. Figure 2The substation equipment fault diagnosis system based on virtual mapping includes: The model retrieval module 1111 is used to retrieve the component-level CAD model from the geometric model library according to the equipment ID of the target substation equipment; the coupling parameter assignment module 1212 is used to retrieve physical field control parameters from the multiphysics parameter library according to the equipment ID, and perform multiphysics coupling parameter assignment on the component-level CAD model to obtain a multiphysics coupling simulation model; the frequency statistics module 1313 is used to retrieve historical fault causes to perform component fault frequency statistics and locate P key component codes; the component configuration module 1414 is used to configure the target substation equipment according to the P key component codes. The host device has P equipment components configured with P multimodal heterogeneous sensing units; a virtual mapping module 1515 is used to map the operating state of the P equipment components to P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units during the operation of the target substation equipment; a fault diagnosis module 1616 is used to perform directed graph simulation of fault propagation in the multiphysics coupling simulation model and output real-time fault diagnosis results, wherein the real-time fault diagnosis results include fault root cause component encoding, component fault probability estimation, and fault component propagation chain.

[0073] Furthermore, the coupling parameter assignment module 1212 in the substation equipment fault diagnosis system based on virtual mapping is also used for: decomposing the physical field control parameters according to the physical domain to obtain multivariate correlation parameters; segmenting the component-level CAD model based on physical characteristics to obtain conductor geometry, insulation geometry, fluid geometry, and structural geometry; mapping and matching the multivariate physical control equations of the conductor geometry, insulation geometry, fluid geometry, and structural geometry; splitting and binding the multivariate correlation parameters to the conductor geometry, insulation geometry, fluid geometry, and structural geometry according to parameter-geometry binding rules, and then executing the coefficient configuration of the multivariate physical control equations in the conductor geometry, insulation geometry, fluid geometry, and structural geometry; loading coupling terms at the geometric domain interfaces of the conductor geometry, insulation geometry, fluid geometry, and structural geometry to generate an initial simulation model; and performing computational mesh discretization processing on the initial simulation model to output the multiphysics coupling simulation model.

[0074] Furthermore, the coupling parameter assignment module 1212 in the substation equipment fault diagnosis system based on virtual mapping is also used for: the multivariate correlation parameters include electromagnetic correlation parameters, thermodynamic correlation parameters, structural mechanics correlation parameters and coupling relationship correlation parameters.

[0075] Furthermore, the frequency statistics module 1313 in the substation equipment fault diagnosis system based on virtual mapping is also used for: retrieving multi-source operation and maintenance records based on the equipment ID, constrained by the operating scenario of the target substation equipment; retrieving the causes of time-series faults from the multi-source operation and maintenance records to obtain the historical causes of faults; recording component fault frequencies based on the historical causes of faults to obtain multiple time-series fault records for multiple equipment components; calculating and outputting multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs based on the multiple time-series fault records; dynamically configuring component weights according to the equipment attributes of the target substation equipment, performing weighted quantization of the multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs, and outputting multiple component fault risk values; and locating the P key component codes in the multiple equipment components according to the descending order of the multiple component fault risk values.

[0076] Furthermore, the component configuration module 1414 in the substation equipment fault diagnosis system based on virtual mapping is also used to: retrieve P fault anomaly association parameter groups of the P equipment components according to the P key component codes, wherein each fault anomaly association parameter group includes at least one fault-sensitive physical quantity; perform sensor type matching based on the fault-sensitive physical quantity composition of the P fault anomaly association parameter groups to obtain the P multimodal heterogeneous sensing units; deploy the P multimodal heterogeneous sensing units on the P equipment components, and establish a real-time data channel between the P multimodal heterogeneous sensing units and the P virtual component sub-models.

[0077] Furthermore, the fault diagnosis module 1616 in the substation equipment fault diagnosis system based on virtual mapping is also used for: establishing multiple coupled topologies based on the physical coupling relationship of the P equipment components in the target substation equipment; after the P virtual component sub-models receive the P multimodal operating status data returned by the P multimodal heterogeneous sensing units, they perform local anomaly identification and output P fault state confidence vectors; loading the P fault state confidence vectors into the multiple coupled topologies, performing directed graph simulation of fault propagation, and outputting multiple initial fault propagation chains; solving the maximum a posteriori probability estimation for the multiple initial fault propagation chains and outputting the fault component propagation chain; tracing and locating the fault root cause component code in the fault component propagation chain, and matching the fault state confidence vector based on the fault root cause component code as the component fault probability estimate.

[0078] Furthermore, the fault diagnosis module 1616 in the substation equipment fault diagnosis system based on virtual mapping is also used for: interactively obtaining multiple operating status data records of the first equipment component under multiple fault scenarios; constructing multiple fault similarity analysis models based on the multiple operating status data records; constructing a first fault confidence network by connecting the multiple fault similarity analysis models in parallel; dynamically deploying the first fault confidence network to the first virtual component sub-model; the first virtual component sub-model receiving and inputting the first multimodal operating status data into the first fault confidence network, and outputting multiple fault confidence scores through parallel analysis of the multiple fault similarity analysis models; fusing the multiple fault confidence scores to output a first fault state confidence vector.

[0079] Furthermore, the fault diagnosis module 1616 in the substation equipment fault diagnosis system based on virtual mapping is also used for: using the first coupled association topology as a dynamic update framework, loading the P fault state confidence vectors into the multiphysics coupled simulation model, calculating the directed edge weights of the first coupled association topology based on the P real-time physical field gradient states output by the multiphysics coupled simulation model, and outputting the first updated association topology; traversing the P fault state confidence vectors using a preset confidence threshold, activating the output directed edges in the first updated association topology, and locating N associated directed edges; calculating the propagation intensity values ​​of the N fault state confidence vectors along the N associated directed edges, and generating the first initial fault propagation chain.

[0080] Furthermore, the fault diagnosis module 1616 in the substation equipment fault diagnosis system based on virtual mapping is also used to: extract the spatial coordinates of the terminal node of the first initial fault propagation chain, perform transient simulation verification of the multiphysics coupling simulation model, and output the first transient simulation propagation chain; if the deviation between the first initial fault propagation chain and the first transient simulation propagation chain meets a preset tolerance threshold, then the first initial fault propagation chain is retained; if the deviation between the first initial fault propagation chain and the first transient simulation propagation chain exceeds the preset tolerance threshold, then the first transient simulation propagation chain is used to replace the first initial fault propagation chain.

[0081] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The substation equipment fault diagnosis method and specific examples based on virtual mapping in Example 1 are also applicable to the substation equipment fault diagnosis system based on virtual mapping in this example. Through the foregoing detailed description of the substation equipment fault diagnosis method based on virtual mapping, those skilled in the art can clearly understand the substation equipment fault diagnosis system based on virtual mapping in this example. Therefore, for the sake of brevity, it will not be described in detail here.

[0082] 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.

[0083] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for fault diagnosis of substation equipment based on virtual mapping, characterized in that, include: Retrieve the component-level CAD model from the geometric model library based on the equipment ID of the target substation equipment; Based on the device ID, retrieve the physical field control parameters from the multiphysics parameter library, perform multiphysics coupling parameter assignment on the component-level CAD model, and obtain the multiphysics coupling simulation model. Retrieve historical fault causes to perform component fault frequency statistics and locate the codes of P key components; Based on the coding of the P key components, P multimodal heterogeneous sensing units are configured in the P equipment components of the target substation equipment body; During the operation of the target substation equipment, the operating status of the P equipment components is mapped to the P virtual component sub-models in the multi-physics field coupled simulation model through the P multimodal heterogeneous sensing units. The multiphysics coupled simulation model performs a directed graph simulation of fault propagation and outputs real-time fault diagnosis results, which include fault root cause component encoding, component fault probability estimation, and fault component propagation chain.

2. The substation equipment fault diagnosis method based on virtual mapping as described in claim 1, characterized in that, The method involves performing a directed graph simulation of fault propagation in the multiphysics coupled simulation model and outputting real-time fault diagnosis results. Based on the physical coupling relationships of the P equipment components in the target substation equipment, multiple coupling association topologies are established; After receiving the P multimodal operating status data from the P multimodal heterogeneous sensing units, the P virtual component sub-models perform local anomaly identification and output P fault status confidence vectors. The P fault state confidence vectors are loaded into the multiple coupled and associated topologies, and a directed graph simulation of fault propagation is performed to output multiple initial fault propagation chains. The maximum a posteriori probability estimation is performed on the multiple initial fault propagation chains to output the fault component propagation chain. The root cause component code of the fault is located by tracing the fault propagation chain, and the fault state confidence vector is matched based on the root cause component code as the fault probability estimate of the component.

3. The substation equipment fault diagnosis method based on virtual mapping as described in claim 2, characterized in that, After receiving P multimodal operating status data from the P multimodal heterogeneous sensing units, the P virtual component sub-models perform local anomaly identification and output P fault state confidence vectors. The method includes: The system interactively obtains multiple operational status data records of the first device component under various fault scenarios. Multiple fault similarity analysis models are constructed based on the aforementioned multiple operational status data records; A first fault confidence network is constructed by connecting the multiple fault similarity analysis models in parallel; The first fault belief network is dynamically deployed to the first virtual component sub-model; The first virtual component sub-model receives and inputs the first multimodal operating state data into the first fault confidence network, and outputs multiple fault confidence scores through parallel analysis by the multiple fault similarity analysis models; By fusing the multiple fault confidence scores, a first fault state confidence vector is output.

4. The substation equipment fault diagnosis method based on virtual mapping as described in claim 2, characterized in that, The method involves loading the P fault state confidence vectors into the multiple coupled and associated topologies, performing a directed graph simulation of fault propagation, and outputting multiple initial fault propagation chains. Using the first coupled topology as a dynamic update framework, after loading the P fault state confidence vectors into the multiphysics coupled simulation model, the directed edge weights of the first coupled topology are calculated based on the P real-time physical field gradient states output by the multiphysics coupled simulation model, and the first updated topology is output. The P fault state confidence vectors are traversed using a preset confidence threshold, and the directed edges are activated in the first updated associated topology to locate N associated directed edges. Calculate the propagation strength values ​​of the N fault state confidence vectors along the N associated directed edges to generate the first initial fault propagation chain.

5. The substation equipment fault diagnosis method based on virtual mapping as described in claim 4, characterized in that, Also includes: Extract the spatial coordinates of the terminal node of the first initial fault propagation chain, perform transient simulation verification of the multiphysics coupling simulation model, and output the first transient simulation propagation chain; If the deviation between the first initial fault propagation chain and the first transient simulation propagation chain meets the preset tolerance threshold, then the first initial fault propagation chain is retained. If the deviation between the first initial fault propagation chain and the first transient simulation propagation chain exceeds a preset tolerance threshold, then the first transient simulation propagation chain is used to replace the first initial fault propagation chain.

6. The substation equipment fault diagnosis method based on virtual mapping as described in claim 1, characterized in that, Based on the device ID, physical field control parameters are retrieved from the multiphysics parameter library, and multiphysics coupling parameters are assigned to the component-level CAD model to obtain a multiphysics coupling simulation model. The method includes: The physical field control parameters are decomposed according to the physical domain to obtain multivariate correlation parameters; The component-level CAD model is segmented based on physical characteristics to obtain conductor geometry, insulation geometry, fluid geometry, and structural geometry. The multivariate physical governing equations of the conductor geometry, insulation geometry, fluid geometry, and structural geometry are mapped and matched. Based on the parameter-geometric domain binding rules, the multivariate correlation parameters are split and bound to the conductor geometry domain, insulation geometry domain, fluid geometry domain, and structural geometry domain. Then, the coefficient configuration of the multivariate physical control equations is executed in the conductor geometry domain, insulation geometry domain, fluid geometry domain, and structural geometry domain. A coupling term is applied at the interface between the geometric domains of the conductor, insulation, fluid, and structural domains to generate an initial simulation model; The initial simulation model is discretized into a computational grid to output the multiphysics coupled simulation model.

7. The substation equipment fault diagnosis method based on virtual mapping as described in claim 6, characterized in that, The multivariate correlation parameters include electromagnetic correlation parameters, thermodynamic correlation parameters, structural mechanics correlation parameters, and coupling relationship correlation parameters.

8. The substation equipment fault diagnosis method based on virtual mapping as described in claim 1, characterized in that, The method involves retrieving historical fault causes to perform component fault frequency statistics and locating the codes of P key components. Based on the operating scenario of the target substation equipment, multi-source operation and maintenance records are retrieved according to the equipment ID; The causes of time-series faults are retrieved from the multi-source operation and maintenance records to obtain the causes of historical faults; Based on the historical causes of failure, the frequency of component failures is recorded to obtain multiple time-series failure records for multiple device components. Based on the multiple time-series fault records, calculate and output multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs; Based on the equipment attributes of the target substation equipment, the component weights are dynamically configured, and the multiple fault frequencies, multiple fault cycle averages, and multiple fault operation and maintenance costs are weighted and quantified to output multiple component fault risk values. Based on the descending order of the failure risk values ​​of the multiple components, the codes of the P key components are located in the multiple equipment components.

9. The substation equipment fault diagnosis method based on virtual mapping as described in claim 1, characterized in that, Based on the codes of the P key components, P multimodal heterogeneous sensing units are configured in the P equipment components of the target substation equipment body, the method comprising: Based on the P key component codes, retrieve the P fault and anomaly association parameter groups of the P equipment components, wherein each fault and anomaly association parameter group includes at least one fault-sensitive physical quantity. Based on the fault-sensitive physical quantities of the P fault-anomaly correlation parameter groups, sensor type matching is performed to obtain the P multimodal heterogeneous sensing units. The P multimodal heterogeneous sensing units are deployed in the P device components, and a real-time data channel is established between the P multimodal heterogeneous sensing units and the P virtual component sub-models.

10. A substation equipment fault diagnosis system based on virtual mapping, characterized in that, The steps for implementing the substation equipment fault diagnosis method based on virtual mapping according to any one of claims 1 to 9, wherein the substation equipment fault diagnosis system based on virtual mapping comprises: The model retrieval module is used to retrieve component-level CAD models from the geometric model library based on the device ID of the target substation equipment. The coupling parameter assignment module is used to retrieve physical field control parameters from the multiphysics parameter library based on the device ID, perform multiphysics coupling parameter assignment on the component-level CAD model, and obtain a multiphysics coupling simulation model. The frequency statistics module is used to retrieve historical fault causes to perform component fault frequency statistics and locate the codes of P key components. The component configuration module is used to configure P multimodal heterogeneous sensing units in the P equipment components of the target substation equipment body according to the P key component codes. The virtual mapping module is used to map the operating status of the P equipment components to the P virtual component sub-models in the multiphysics coupling simulation model through the P multimodal heterogeneous sensing units during the operation of the target substation equipment. The fault diagnosis module is used to perform directed graph simulation of fault propagation in the multiphysics coupled simulation model and output real-time fault diagnosis results, wherein the real-time fault diagnosis results include fault root cause component encoding, component fault probability estimation and fault component propagation chain.

Images