A substation defect scene generation and fusion simulation method and device

Abstract

The application discloses a substation defect scene generation and fusion simulation method and device, and relates to the technical field of substations.The method comprises the following steps: collecting on-site images of a substation, constructing a time-space correlation scene library, collecting defect images and defect characteristic parameters of historical faults of the substation, constructing a defect working condition tracing correlation library, obtaining multi-dimensional defect generation constraint conditions, generating virtual defect images and virtual defect parameters of the substation by means of artificial intelligence content generation technology, adaptively fusing the virtual defect images and the time-space correlation scene library, constructing a substation defect simulation scene, diagnosing the substation defect simulation scene by means of a defect scene diagnosis algorithm, and generating scene diagnosis data.The application can realize accurate generation and simulation of a substation defect scene, solve the problems of single traditional simulation scene and poor environmental adaptability, overcome the defect that virtual defects lack logical constraints, improve the fidelity of the simulation scene, and guarantee the quality of the scene.

Description

Technical Field

[0001] This invention relates to the field of substation technology, and in particular to a method and apparatus for generating and fusion simulation of substation defect scenarios. Background Technology

[0002] As the core hub of the power system, the accurate detection and early warning of equipment defects in substations are crucial for the safe operation of the power grid. Traditional substation defect detection relies on manual inspections or machine learning models based on real defect samples, but this has significant limitations: First, obtaining real defect samples is difficult and covers limited scenarios, especially samples of extreme environments or rare defects are scarce, resulting in insufficient generalization ability of the detection model; second, existing defect simulations mostly use simple image synthesis methods, lacking correlation modeling of the complex spatiotemporal environment of substations. The generated defect scenarios have low physical consistency and visual realism with the actual operating scenarios, making it difficult to meet the needs of high-precision model training or simulation training for operation and maintenance personnel. In addition, the occurrence of defects is closely related to the operating conditions of equipment. Existing technologies have not established a deep source-tracing correlation between defect features and operating parameters, resulting in the generated virtual defects lacking physical and logical constraints, further affecting the practicality of the simulation scenarios. Summary of the Invention

[0003] The purpose of this invention is to overcome at least one technical problem existing in the prior art and to provide a method and device for generating and fusion simulation of substation defect scenarios.

[0004] On one hand, this invention provides a method for generating and fusing simulation of substation defect scenarios, including: Step S1, collecting substation site images under normal conditions from different environments, angles, and time points and recording the collection parameters; performing multi-dimensional spatiotemporal scene association based on the collection parameters to construct a spatiotemporal associated scenario library; Step S2, collecting defect images and defect feature parameters of historical faults in the substation, collecting substation operating condition data and associating it with the defect images to construct a defect operating condition tracing association library; Step S3, parsing the association relationships in the defect operating condition tracing association library and deduce multi-dimensional defect generation constraints; based on the tracing association library and constraints, generating virtual defect images and virtual defect parameters of the substation using artificial intelligence content generation technology; Step S4, using a scene defect adaptive fusion algorithm to fuse the virtual defect images based on the virtual defect parameters. Step S5: Adaptively fuse the simulation scenario with the spatiotemporal related scene library to construct a substation defect simulation scenario; Step S6: Diagnose the substation defect simulation scenario using a defect scenario diagnosis algorithm to determine whether the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold; Step S7: In response to the fact that the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold, mark this substation defect simulation scenario as a successfully fused scenario and output the substation defect simulation scenario and its detailed parameter report; Step S8: In response to the fact that the authenticity quantification score of the substation defect simulation scenario does not meet the preset authenticity quantification score threshold, automatically locate the abnormal problem in the substation defect simulation scenario, and feed back the optimization instruction containing the abnormal problem location information to Step S4 to trigger the optimization and reconstruction of the substation defect simulation scenario.

[0005] Furthermore, the acquisition parameters include one or a combination of acquisition time, environmental data, acquisition angle, device type, device spatial coordinates, device operation data, acquisition device parameters, and historical defect records. The environmental data includes temperature, light intensity, and weather type. Step S1 includes: establishing an image-physical location mapping based on device spatial coordinates and generating a three-dimensional point cloud map; sorting images of devices at the same location by acquisition time to form a time series; marking the trend of device status changes in combination with acquisition parameters; generating a spatiotemporal index and dividing the image scene sub-library according to device type, environment, and angle; and integrating them to form a spatiotemporally related scene library.

[0006] Furthermore, step S2 includes: step S201, extracting defect images from historical fault data and automatically extracting their geometric, physical, and appearance feature parameters as defect feature parameters; step S202, acquiring historical operating condition data corresponding to the defect images, the historical operating condition data including electrical parameters, environmental parameters, and equipment status parameters; step S203, establishing a mapping relationship between defect types, defect feature parameters, and historical operating condition data, storing it in a structured form to form the defect operating condition traceability association library.

[0007] Furthermore, the multi-dimensional defect generation constraints in step S3 include physical constraints, operating condition constraints, and environmental constraints. The physical constraints limit the size and shape of the defect to not exceed the range of the equipment body. The operating condition constraints limit the defect to be generated only under the corresponding operating conditions. The environmental constraints limit the defect generation to match the corresponding environmental parameters.

[0008] Furthermore, step S3 includes: encoding the constraint conditions into a condition vector, inputting it into a pre-trained conditional generation model, and generating a virtual defect image and detailed virtual defect parameters that match the condition vector; the conditional generation model is a conditional generative adversarial network or a latent diffusion model; the virtual defect parameters include one or a combination of defective equipment, defect type, location coordinates, geometric information, physical parameters, and associated virtual working condition data.

[0009] Furthermore, step S4 includes: step S401, performing multimodal parameter matching and fusion based on the spatiotemporal correlated scene library and virtual defect parameters to generate a scene defect fusion parameter set to guide image fusion; step S402, based on the scene defect fusion parameter set, adaptively fusing the substation site images in the spatiotemporal correlated scene library with the substation virtual defect images through a scene defect adaptive fusion algorithm to construct a substation defect simulation scene, wherein the execution of the scene defect adaptive fusion algorithm is controlled by the scene defect fusion parameter set.

[0010] Furthermore, the multimodal parameter matching and fusion in step S401 includes: dividing the virtual defect parameters and parameters in the spatiotemporally related scene library into spatial attribute parameters, physical and environmental attribute parameters, and logical feature attribute parameters; performing dimensionless processing on the parameters; assigning a matching fusion factor, error adjustment coefficient, and fusion orthogonal weight to each individual index in each type of parameter; and generating the scene defect fusion parameter set through multi-dimensional collaborative calculation; wherein, the spatial attribute parameters include at least one of location coordinates, geometric information, defect influence range, image angle, and spatial index; the physical and environmental attribute parameters include at least one of defect operating conditions, appearance features, equipment operating conditions, environmental data, and image physical features; and the logical feature attribute parameters include at least one of equipment type, defect type, defect triggering conditions, development sequence, credibility score, acquisition time, time series, and historical defects of the equipment.

[0011] Furthermore, the execution of the scene defect adaptive fusion algorithm in step S402, controlled by the scene defect fusion parameter set, includes: performing geometric transformation and spatial positioning on the substation virtual defect image based on the spatial attribute parameters in the scene defect fusion parameter set; performing relighting, material fusion, and environmental effect rendering on the substation virtual defect image based on the physical and environmental attribute parameters in the scene defect fusion parameter set; and verifying the rationality of the fusion result and associating it with the defect parameter control mechanism based on the logical feature attribute parameters in the scene defect fusion parameter set.

[0012] Furthermore, step S5 includes: calculating the quantitative score of the realism of the substation defect simulation scene through a defect scene diagnosis and scoring algorithm, including collecting three core evaluation indicators: physical rationality, visual realism, and spatiotemporal consistency; setting evaluation limit thresholds for individual evaluation items under each evaluation indicator; and obtaining the quantitative score of the realism of the substation defect simulation scene through multi-indicator weighted collaborative calculation; wherein, the physical rationality evaluation indicator includes evaluating at least one of defect morphology rationality, location rationality, and operating condition matching; the visual realism evaluation indicator includes evaluating at least one of illumination consistency, texture fusion, color consistency, and noise matching; and the spatiotemporal consistency evaluation indicator includes evaluating at least one of dynamic time decay coefficient, adjacent frame time interval, and spatiotemporal overlap.

[0013] Secondly, embodiments of the present invention provide a substation defect scene generation and fusion simulation device. The device is implemented using the aforementioned substation defect scene generation and fusion simulation method. The device includes: a spatiotemporal correlation scene library construction module, suitable for collecting substation site images under normal conditions from different environments, angles, and time points, and recording the collection parameters; performing multi-dimensional spatiotemporal scene correlation based on the collection parameters to construct a spatiotemporal correlation scene library; a defect condition tracing correlation library construction module, suitable for collecting defect images and defect feature parameters of historical substation faults, collecting substation operating condition data, and correlating it with the defect images to construct a defect condition tracing correlation library; a substation virtual defect image generation module, suitable for parsing the correlation relationships in the defect condition tracing correlation library and deduce multi-dimensional defect generation constraints; and generating substation virtual defect images and virtual defect parameters based on the tracing correlation library and constraints using artificial intelligence content generation technology; and a substation defect simulation scene construction module, suitable for constructing virtual defect scenes based on virtual defect images. The system employs a scene defect adaptive fusion algorithm to adaptively fuse virtual defect images and a spatiotemporally related scene library to construct a substation defect simulation scene. A realism quantification scoring module is used to diagnose the substation defect simulation scene using a defect scene diagnosis algorithm, determining whether the realism quantification score of the substation defect simulation scene meets a preset realism quantification scoring threshold. A substation defect simulation scene output module is used to mark the substation defect simulation scene as a successfully fused scene and output a report of the substation defect simulation scene and its detailed parameters when the realism quantification score of the substation defect simulation scene meets the preset realism quantification scoring threshold. An optimization instruction feedback module is used to automatically locate abnormal problems in the substation defect simulation scene when the realism quantification score of the substation defect simulation scene does not meet the preset realism quantification scoring threshold, and feeds back optimization instructions containing the location information of the abnormal problems to the substation defect simulation scene construction module to trigger the optimization and reconstruction of the substation defect simulation scene.

[0014] Thirdly, embodiments of the present invention also provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the above-described method for generating and fusion simulation of substation defect scenarios.

[0015] Fourthly, embodiments of the present invention also provide a readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to execute the above-described method for generating and fusion simulation of substation defect scenarios.

[0016] The beneficial effects achieved by this invention are as follows: This invention can accurately generate and simulate substation defect scenarios, solve the problems of traditional simulation scenarios being single and having poor environmental adaptability, overcome the lack of logical constraints in virtual defects, ensure the visual consistency and physical rationality of defects in terms of lighting, angle, material reflection, etc., improve the realism of simulation scenarios, ensure scenario quality, provide high-quality simulation data support for immersive training of operation and maintenance personnel and power system risk early warning, and significantly improve the technical level of intelligent operation and maintenance of substations. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart of a substation defect scene generation and fusion simulation method provided in Embodiment 1 of this application.

[0019] Figure 2 This is a schematic diagram of the structure of a substation defect scene generation and fusion simulation device provided in Embodiment 2 of this application.

[0020] Figure 3 This is a partial block diagram of the electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1

[0023] like Figure 1 The diagram shown is a flowchart of a substation defect scene generation and fusion simulation method provided in Embodiment 1 of this application.

[0024] As an example, the method includes: Step S1, collecting substation site images under normal conditions from different environments, angles, and time points and recording the collection parameters; performing multi-dimensional spatiotemporal scene association based on the collection parameters to construct a spatiotemporal associated scene library; Step S2, collecting defect images and defect feature parameters of historical faults in the substation, collecting substation operating condition data and associating it with the defect images to construct a defect operating condition tracing association library; Step S3, parsing the association relationships in the defect operating condition tracing association library and deducing multi-dimensional defect generation constraints; based on the tracing association library and constraints, generating virtual defect images and virtual defect parameters of the substation using artificial intelligence content generation technology; Step S4, using a scene defect adaptive fusion algorithm to combine the virtual defect images and the spatiotemporal associated scene library based on the virtual defect parameters. Adaptive fusion is used to construct a substation defect simulation scenario; Step S5: The substation defect simulation scenario is diagnosed using a defect scenario diagnosis algorithm to determine whether the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold; Step S6: In response to the authenticity quantification score of the substation defect simulation scenario meeting the preset authenticity quantification score threshold, this substation defect simulation scenario is marked as a successfully fused scenario, and a substation defect simulation scenario and its detailed parameter report are output; Step S7: In response to the authenticity quantification score of the substation defect simulation scenario not meeting the preset authenticity quantification score threshold, abnormal problems in the substation defect simulation scenario are automatically located, and optimization instructions containing the abnormal problem location information are fed back to Step S4 to trigger the optimization and reconstruction of the substation defect simulation scenario.

[0025] In some feasible implementations, the acquisition parameters include one or a combination of acquisition time, environmental data, acquisition angle, device type, device spatial coordinates, device operation data, acquisition device parameters, and historical defect records. The environmental data includes temperature, light intensity, and weather type. Step S1 includes: establishing an image-physical location mapping based on device spatial coordinates and generating a three-dimensional point cloud map; sorting images of devices at the same location by acquisition time to form a time series; marking the trend of device status changes in combination with acquisition parameters; generating a spatiotemporal index and dividing the image scene sub-library according to device type, environment, and angle; and integrating them to form a spatiotemporally related scene library.

[0026] Specifically, appropriate image acquisition devices are selected based on the characteristics of the substation equipment. Based on preset acquisition control conditions, on-site images of each substation device under normal operating conditions and various environmental conditions and angles are acquired. Image acquisition devices include, but are not limited to, thermal imaging cameras, high-definition network cameras, multispectral imaging detectors, and inspection robots and drones equipped with image acquisition devices. Acquisition control conditions include, but are not limited to, environmental control conditions, angle control conditions, and time control conditions. During the acquisition process, acquisition parameters such as acquisition time, environmental data, acquisition angle, type of corresponding equipment, operating data, spatial coordinates, acquisition equipment parameters, and historical defect records are recorded in real time for each substation's on-site images. Environmental data includes, but is not limited to, temperature data, illumination data, and weather type data. Based on the spatial coordinates of the corresponding equipment in the acquisition parameters, an image-physical location mapping relationship is established, generating a 3D point cloud map of the substation to establish spatial association. Building upon this spatial association, the substation field images corresponding to the same coordinate equipment are sorted by acquisition time to construct a time series. Based on the operating data of the corresponding equipment in each substation field image, acquisition equipment parameters, acquisition environment data, and other acquisition parameters, the changing trends of the corresponding equipment in each substation field image are marked with a time series, establishing a spatial-temporal association. A spatiotemporal index is generated based on this association. Image scene sub-libraries are then divided based on the spatiotemporal index and the type, acquisition angle, and acquisition environment data of the corresponding equipment in the acquisition parameters. Each image scene sub-library is categorized by equipment type, including multi-environment image sets, multi-angle image sets, and other image sets categorized by different acquisition parameters. Each image scene sub-library is connected to the spatial-temporal association by equipment type to construct a spatiotemporally associated scene library. Finally, a spatiotemporally associated scene library with interconnected spatial location, time series, and scene parameters is constructed.

[0027] In other words, step S1 aims to create a "normal state" image database covering all equipment, angles, and environmental conditions in the substation, with each image containing rich spatiotemporal and environmental metadata. The implementation involves: 1. Multi-source data acquisition: Equipment: Deploying high-definition visible light cameras, infrared thermal imagers, and ultraviolet imagers, and utilizing inspection robots and drones for mobile, multi-angle data acquisition. Strategy: Taking multiple images of the same equipment at different times (morning, noon, dusk, night), in different seasons, and under different weather conditions (sunny, rainy, snowy, fog) to cover various lighting and environmental conditions. 2. Data annotation and association: Automatically or manually recording the metadata of each acquired image, including: GPS / 3D coordinates, shooting time, equipment ID, ambient temperature and humidity, light intensity, weather conditions, and real-time operating parameters of the equipment (current, voltage, etc.). 3. Spatiotemporal modeling: Using 3D reconstruction technology (such as photogrammetry), fusing all 2D images with their spatial coordinates to generate a 3D point cloud model of the entire substation. In this way, each image can find its corresponding location in the 3D model. 4. Sort all images of the same location by time to form a time series, which is used to observe normal changes in the equipment over time (such as seasonal vegetation growth and normal equipment thermal cycling). 5. Build an index library: This ultimately forms a structured database that can be quickly retrieved by "equipment type-spatial location-time-environmental conditions". For example, it can instantly retrieve all normal state images of the main transformer under winter snow, nighttime, and high load conditions.

[0028] In some feasible implementations, step S2 includes: step S201, extracting defect images from historical fault data and automatically extracting their geometric, physical, and appearance feature parameters as defect feature parameters; step S202, acquiring historical operating condition data corresponding to the defect images, the historical operating condition data including electrical parameters, environmental parameters, and equipment status parameters; step S203, establishing a mapping relationship between defect types, defect feature parameters, and historical operating condition data, storing it in a structured form to form the defect operating condition traceability association library.

[0029] Specifically, defect images of various defect types from equipment in substations that have experienced historical faults are collected. While collecting these images, defect characteristic parameters such as geometric, physical, and apparent parameters are also collected. Real-time operating condition data of the substation related to the defects is acquired, including electrical parameters, environmental parameters, and equipment status parameters. A defect condition mapping table is established between the defect characteristic parameters and the real-time operating conditions of the substation. Based on the defect characteristic parameters, defect images are associated with the defect condition mapping table to construct a defect condition traceability and association database, stored in the format of "defect type - defect characteristic parameter - operating condition parameter - defect image". This forms a traceable and strongly correlated defect condition traceability and association database.

[0030] In other words, step S2 aims to deeply correlate historical defects with the equipment's operating status and environmental conditions at the time of their occurrence, forming a "defect knowledge graph." This is achieved through the following steps: 1. Collecting historical defect data: Image evidence of historical defects (such as photos / videos of discharge, overheating, and damage) is collected from maintenance records. Simultaneously, characteristic parameters of these defects are extracted, such as: defect geometry, specific temperature of the overheating point, number of ultraviolet photons in the discharge, color and area of ​​corrosion, etc. 2. Correlating operating condition data: Comprehensive operating condition data for a period before and after the defect occurs is retrieved from the SCADA (Supervisory Control and Data Acquisition) system and environmental monitoring system, including: load current, voltage, power, ambient temperature and humidity, and whether any abnormal operations occurred. 3. Building a correlation database: A mapping table of "defect characteristics - operating conditions" is established using database technology. For example, a record might be: Defect type: Wire overheating - Characteristic: Temperature 128°C - Operating conditions: Load current 1800A, ambient temperature 38°C, wind speed 0.5m / s.

[0031] In some feasible implementations, the multi-dimensional defect generation constraints in step S3 include physical constraints, operating condition constraints, and environmental constraints. The physical constraints limit the size and shape of the defect to within the range of the equipment body. The operating condition constraints limit the generation of the defect only under corresponding operating conditions. The environmental constraints limit the generation of the defect to match the corresponding environmental parameters. Step S3 includes: encoding the constraints into a condition vector, inputting it into a pre-trained conditional generation model, and generating a virtual defect image and detailed virtual defect parameters that match the condition vector. The conditional generation model is a conditional generative adversarial network or a latent diffusion model. The virtual defect parameters include one or a combination of defective equipment, defect type, location coordinates, geometric information, physical parameters, and associated virtual operating condition data.

[0032] Specifically, the correlation strength between defect feature parameters and operating condition data in the defect condition tracing and association database is analyzed. Key constraint rules are extracted, and constraint conditions are deduced from multiple dimensions such as physical, operating condition, and environmental constraints based on these key constraint rules. This generates multi-dimensional defect generation constraints, including physical constraints, operating condition constraints, and environmental constraints. These multi-dimensional defect generation constraints are encoded into constraint feature vectors. A defect generation constraint mechanism is constructed based on these constraint feature vectors. The defect condition tracing and association database is used as sample learning data, and fine-tuned to a preset number of iterations using transfer learning techniques to generate a virtual defect dataset. Based on this virtual defect dataset, virtual images and virtual defect parameters of substations in various scenarios, such as single defects, composite defects, and sparse defects, are generated using the defect generation constraint mechanism and artificial intelligence content generation technology. Virtual defect parameters include parameters such as defective equipment, defect type, defect triggering conditions, defect impact range, development sequence, location coordinates, geometric information, appearance features, virtual operating conditions, virtual environment, and credibility score.

[0033] More specifically, the relationship between the defect condition tracing database and the defect condition data is analyzed. Data analysis tools are used to mine the correlation strength between defect feature parameters and condition data, extracting key constraint rules such as "specific conditions trigger specific defects" and "defect parameters are limited by environmental conditions." Based on these rules, multi-dimensional defect generation constraints are deduced from three dimensions: physical, operating condition, and environmental. Physical constraints limit the size and shape of defects to the equipment body and conform to the equipment's physical characteristics. Operating condition constraints limit defects to be generated only under corresponding operating conditions, such as triggering wire overheating defects under overload conditions. Environmental constraints limit defect generation to match corresponding environmental parameters, such as preventing high-temperature defects from being generated in low-temperature environments. These multi-dimensional constraints are encoded into feature vectors to construct a defect generation constraint mechanism, limiting the generation boundaries of virtual defects. Using defect images, feature parameters, and operating condition data from a defect condition tracing and correlation database as sample learning data, LoRA (Low-Rank Adaptive) transfer learning technology is employed to fine-tune AI-generated content (AIGC) models such as Stable Diffusion or Midjourney, iterating 50-100 times to enable the model to accurately learn the correlation logic between substation equipment morphology, defect features, and operating conditions. Through the fine-tuned model and defect generation constraint mechanism, virtual defect images of various scenarios are generated, including single defects, compound defects (such as conductor overheating + insulator damage), and rare defects (such as equipment short circuits under extreme rainstorms). Simultaneously, corresponding virtual defect parameters are output, including defect equipment type, defect triggering conditions, defect impact range, development sequence, location coordinates, geometric information, appearance features, virtual operating conditions, virtual environmental parameters, and credibility score.

[0034] In other words, step S3 aims to achieve the following: using artificial intelligence to generate large-scale, automated defect samples that conform to physical laws and are difficult to collect in reality. The implementation is as follows: 1. Deducing Constraints: Data mining and machine learning are performed on the "defect knowledge graph" constructed in S2 to automatically summarize the physical, working condition, and environmental logic rules for defect generation. For example, the system learns that "the severity of overheating defects at connection points is proportional to the square of the load current." These rules are transformed into "drawing instructions" for the AI, i.e., multi-dimensional defect generation constraints. 2. Constrained AIGC Generation: AIGC techniques such as Generative Adversarial Networks (GANs) or Diffusion Models are employed. During training, not only are real defect images from S2 used, but more importantly, the aforementioned constraints are input as control signals to the model. Generation Result: Virtual Defect Image: A realistic, unprecedented defect image (e.g., a rare composite defect under specific lighting angles). Virtual defect parameters: A complete "identity card" describing the defect, including defect type, location, size, severity, confidence level, and the virtual operating conditions that trigger the defect (e.g., this defect will appear when the current exceeds X amperes).

[0035] In some feasible implementations, step S4 includes: step S401, performing multimodal parameter matching and fusion based on a spatiotemporal correlated scene library and virtual defect parameters to generate a scene defect fusion parameter set to guide image fusion; step S402, based on the scene defect fusion parameter set, adaptively fusing substation field images and substation virtual defect images in the spatiotemporal correlated scene library with a scene defect adaptive fusion algorithm to construct a substation defect simulation scene, wherein the execution of the scene defect adaptive fusion algorithm is controlled by the scene defect fusion parameter set.

[0036] Preferably, the multimodal parameter matching and fusion in step S401 includes: dividing the virtual defect parameters and parameters in the spatiotemporally related scene library into spatial attribute parameters, physical and environmental attribute parameters, and logical feature attribute parameters; performing dimensionless processing on the parameters; assigning a matching fusion factor, error adjustment coefficient, and fusion orthogonal weight to a single indicator in each type of parameter; and generating the scene defect fusion parameter set through multi-dimensional collaborative calculation; wherein, the spatial attribute parameters include at least one of location coordinates, geometric information, defect influence range, image angle, and spatial index; the physical and environmental attribute parameters include at least one of defect operating conditions, appearance features, equipment operating conditions, environmental data, and image physical features; and the logical feature attribute parameters include at least one of equipment type, defect type, defect triggering conditions, development sequence, credibility score, acquisition time, time series, and historical defects of the equipment.

[0037] Preferably, the execution of the scene defect adaptive fusion algorithm in step S402, controlled by the scene defect fusion parameter set, includes: performing geometric transformation and spatial positioning on the substation virtual defect image based on the spatial attribute parameters in the scene defect fusion parameter set; performing relighting, material fusion, and environmental effect rendering on the substation virtual defect image based on the physical and environmental attribute parameters in the scene defect fusion parameter set; and verifying the rationality of the fusion result and associating it with the defect parameter control mechanism based on the logical feature attribute parameters in the scene defect fusion parameter set.

[0038] Preferably, step S402 further includes: constructing a defect parameter adjustment mechanism, associating the scenario defect fusion parameter set with the substation defect simulation scenario, so that when the defect parameters are adjusted through the mechanism, steps S401 and S402 can be re-executed to dynamically update the substation defect simulation scenario.

[0039] In other words, the goal of step S4 is to seamlessly and realistically integrate the virtual defect generated in S3 into the real normal scene in S1, making it interactive and adjustable. The implementation is as follows: 1. Multimodal parameter matching (S401): When the system receives a virtual defect (e.g., "a corona discharge on a bushing"), it automatically searches the spatiotemporal scene library in S1 for the most matching real background image (e.g., "a normal image of the same bushing taken at dusk"). Next, the system performs a detailed comparison between the parameters of the virtual defect and the parameters of the real scene, categorizing them into three main attributes: spatial, physical and environmental, and logical. Through weighted calculations, a "scene defect fusion parameter set" is generated. This parameter set is the detailed "fusion recipe," precisely specifying how the defect should be deformed, how the lighting should be applied, and how the colors should be adjusted for perfect fusion. 2. Adaptive fusion of scene defects (S402): The algorithm strictly follows the "fusion recipe": Geometric transformation: Based on spatial attributes, the virtual defect image is distorted and deformed to fit the surface and perspective of the real device. Relighting and Rendering: Based on physical and environmental properties, a lighting model of the real scene is calculated, and then the virtual defects are relit to generate highlights and shadows that conform to the direction of the scene's light source. Material Blending: Using advanced image fusion techniques (such as Poisson blending), the edges of the defects are seamlessly blended with the textures and noise of the background, eliminating the sense of patchwork. 3. Constructing a Defect Parameter Adjustment Mechanism: The system dynamically associates the fused simulation scene with all generated parameters, creating an adjustment interface. Users can adjust defect parameters through the interface (such as a slider) (e.g., adjusting the "overheating temperature" from 100°C to 200°C). The system will immediately re-execute S401 and S402 according to the new parameters, generating a new, more severe but still realistically blended simulation image in real time.

[0040] In some feasible implementations, the method further includes step S5: diagnosing the substation defect simulation scenario using a defect scenario diagnosis algorithm to determine whether the authenticity quantification score of the substation defect simulation scenario meets a preset authenticity quantification score threshold; step S6: in response to the authenticity quantification score of the substation defect simulation scenario meeting the preset authenticity quantification score threshold, marking this substation defect simulation scenario as a successfully fused scenario, and outputting a report of the substation defect simulation scenario and its detailed parameters; step S7: in response to the authenticity quantification score of the substation defect simulation scenario not meeting the preset authenticity quantification score threshold, automatically locating abnormal problems in the substation defect simulation scenario, and feeding back optimization instructions containing the abnormal problem location information to step S4 to trigger the optimization and reconstruction of the substation defect simulation scenario.

[0041] Preferably, step S5 includes:

[0042] The simulation of substation defects is quantitatively scored using a defect scene diagnosis and scoring algorithm. This algorithm includes collecting three core evaluation indicators: physical rationality, visual realism, and spatiotemporal consistency. Evaluation thresholds are set for individual evaluation items under each indicator category. The quantitative score is obtained through weighted collaborative calculation of multiple indicators. Specifically, the physical rationality evaluation indicators include evaluating at least one of defect morphology rationality, location rationality, and operating condition matching; the visual realism evaluation indicators include evaluating at least one of illumination consistency, texture blending, color consistency, and noise matching; and the spatiotemporal consistency evaluation indicators include evaluating at least one of dynamic time decay coefficient, adjacent frame time interval, and spatiotemporal overlap.

[0043] Specifically, it is determined whether the authenticity quantification score of the substation defect simulation scenario meets the authenticity quantification score threshold. When it meets the authenticity quantification score threshold, the substation defect simulation scenario is marked as a successfully fused scenario, and the substation defect simulation scenario and its detailed parameter report are output. When it does not meet the authenticity quantification score threshold, the abnormal problems of the substation defect simulation scenario are automatically located, scenario diagnostic data is generated, and it is fed back to step S4 for re-optimization and fusion.

[0044] Specifically, after calculating the authenticity quantification score through the defect scene diagnosis scoring formula, the preset scoring threshold is 80 points: if the score meets the standard, the scene is marked as a "successfully fused scene", and the simulation scene image and detailed parameter report (including defect parameters, fusion parameters, working condition data, and scoring results) are output; if the score does not meet the standard, abnormal problems (such as defect position offset, visual fusion stiffness, physical logic conflict) are automatically located, scene diagnosis data is generated and fed back to step S4, and the fusion parameters or fusion algorithm parameters are readjusted for secondary fusion optimization until the scene score meets the standard.

[0045] In other words, steps S5 to S7 aim to: automatically assess the quality of the simulation scenes generated in S4, automatically rework unqualified scenes, and form a closed loop for producing high-quality scenes. The implementation is as follows: 1. Realism Quantification Scoring: Design a diagnostic algorithm to score the simulation scene from three core dimensions: Physical Reasonableness: Does the location and shape of the defects conform to physical laws? (e.g., does the crack appear at a structural stress point?); Visual Realism: Are the lighting, shadows, textures, and colors of the defects consistent with the background? (Using image quality evaluation indicators such as SSIM and perceptual loss); Spatiotemporal Consistency: If it is a dynamic scene, is the evolution of the defects natural and smooth? An objective score of 0-100 is given through weighted calculation. 2. Diagnosis and Feedback Optimization: Qualification Judgment: If the score is higher than a preset threshold (e.g., 80 points), the scene is marked as qualified and stored in the "High-Quality Simulation Scene Library". Unqualified Handling: If the score is lower than the threshold, the system will automatically analyze the deduction items and locate the problem (e.g., "Diagnosis Result: Shadow direction deviation 35 degrees"). Closed-loop feedback: This specific diagnostic report is automatically fed back to the fusion system in step S4. The S4 system adjusts the fusion strategy accordingly, regenerates the scene, and sends it back to S5 for diagnosis. This cycle continues until the scene score meets the standard.

[0046] To facilitate understanding of the above embodiments, a specific example is given here: Project objective: To generate a simulation scenario of "surface contamination leading to partial discharge" for the "A-phase high-voltage bushing of the main transformer" of a 500kV substation, in order to train operation and maintenance personnel to identify such defects.

[0047] Step S1: Constructing a Spatiotemporal Correlation Scene Library (Establishing a Normal State Digital Archive) Implementation Process: 1. Data Acquisition: Using drones equipped with high-definition cameras and infrared thermal imagers, the main transformer is photographed from all angles at dawn, noon, and dusk in spring, summer, autumn, and winter. Simultaneously, a fixed high-altitude camera continuously records the transformer's state under different weather conditions (sunny, rainy, foggy). Each shot records: precise timestamps, GPS coordinates, ambient temperature and humidity, light intensity, transformer load current (e.g., 650A), and other parameters. 2. Spatiotemporal Correlation: All images are used to generate a refined 3D model of the main transformer using 3D reconstruction technology. Within the model, any location on the A-phase bushing can be precisely clicked, instantly retrieving all normal state images of that location under different seasons, time periods, and weather conditions. For example, the system can quickly retrieve 12 visible light and 12 infrared images of the A-phase bushing under "winter dawn, light fog, and medium load conditions." Output: A three-dimensional, dynamic "normal state digital archive" containing tens of thousands of precisely searchable images.

[0048] Step S2: Constructing a Defect Condition Tracing and Association Database (Establishing a Defect Knowledge Graph) Implementation Process: 1. Collecting Historical Defects: Retrieve 5 similar cases of "partial discharge caused by bushing surface contamination" from the provincial power grid defect database. Collect defect images (including ultraviolet discharge spot images), report texts, and maintenance records for these cases. 2. Feature Extraction and Association: Extract key information from the reports: the contamination component is "cement plant dust mixed with salt," the contaminated area is located in the "middle of the bushing skirt," the "ambient humidity > 85%" during discharge, and the "equipment voltage is 525kV." This information is structured to form a knowledge record: Defect type: surface contamination partial discharge; Key characteristics: the contaminant is a mixture of industrial dust and salt, located in the middle of the umbrella skirt; Triggering conditions: ambient humidity > 85%, operating voltage > 500kV; Associated image: the ultraviolet discharge spot is clustered, and the visible light shows a gray-black contaminant band; Output: a "defect knowledge graph" containing dozens of defect types and hundreds of records, which clarifies "what defects will occur under what conditions and what the defects look like".

[0049] Step S3: AIGC Generates Virtual Defects (Creating New Defects that Conform to Physical Laws) Implementation Process: 1. Deducing Constraints: After analyzing the knowledge graph, the system automatically summarizes the rules for generating this type of defect: Physical Constraints: The contaminated area must be distributed along the bushing skirt and cannot appear on the metal flange. Environmental Constraints: High humidity is a necessary condition. Logical Constraints: The composition of the contaminant affects the discharge intensity; industrial salt will lead to a stronger discharge. 2. Constraint-Based Generation: The user inputs the command: "Generate a moderately severe surface contamination discharge defect in phase A bushing, with an ambient humidity of 90%." Under the constraints of the rules, the AIGC model generates an unprecedented but perfectly reasonable virtual defect image: On the bushing skirt in the middle of the bushing, there is a gray-black contaminant band that conforms to the characteristics of industrial dust. Simultaneously, a set of virtual defect parameters is generated: Defect location: 3rd-5th skirt from top to bottom of the bushing; Contaminant composition: Cement dust mixed with sea salt; Discharge intensity: Medium (approximately 8000 pps ultraviolet photons); Triggering conditions: Humidity > 88%, Voltage > 515kV; Confidence score: 92 / 100. Output: A pair of "virtual defect images" and their detailed "identity cards" (parameter set).

[0050] Step S4: Adaptive Fusion (Seamlessly Integrating Defects into the Real Scene) Implementation Process: 1. S401: Formulating the Fusion Recipe: After receiving the virtual defect, the system retrieves the most matching background image from the "Digital Archive" in S1: a high-resolution visible light image of the main transformer taken at dusk after a spring rain, with a humidity of 92% and a load of 600A. The system performs precise calculations, comparing the virtual defect with the real background: Spatial attributes: The defect needs to precisely fit the curved surfaces of the 3rd-5th umbrella skirts. Physical attributes: Diffuse light at dusk, reflective characteristics of wet surfaces after rain. Logical attributes: Defect type matches the equipment. Generating the "Fusion Recipe": {Lighting Model: Soft diffuse light, light source direction: west; Surface reflectivity: Increase by 15% to simulate a feeling of wetness; Color adjustment: Shift 5% towards the cool tone of the background...}. 2. S402: Executing the Fusion and Imparting Interactivity: The algorithm works strictly according to the "recipe": The virtual filth strip is perspective-distorted to perfectly fit the curved surfaces of the real sleeve's umbrella skirts. The contaminated area is relit to generate shadows and highlights that match the soft light of dusk. The color and texture of the contaminated area are adjusted to make it appear as "wet" as the background. Advanced image fusion technology is used to seamlessly blend the edges of the contaminated area with the background, leaving no visible stitching. A control mechanism is built: after generating the final simulation image, the system assigns a "contamination severity" slider to it. When the trainee drags the slider from "medium" to "severe," the system immediately re-executes S401 based on the new severity, generating a new "fusion recipe" (e.g., increasing the contaminated area by 30% and deepening the color). Then, S402 is re-executed, rendering a new simulation image almost in real-time, showing a dirtier, higher-risk casing, but whose lighting, wetness, and other physical characteristics remain perfectly consistent with the background after a twilight rain. Output: A highly realistic simulation image and an interactive simulation scene where the severity of the defect can be dynamically adjusted.

[0051] Steps S5 to S7: Scene Diagnosis and Optimization (Automated Quality Inspection) Implementation Process: 1. Quantitative Scoring: The system performs a "physical examination" on the simulation scene generated in S4, scoring it from three dimensions: Physical Reasonableness (95 points): The contamination location is on the umbrella skirt, conforming to the creepage distance logic, excellent. Visual Realism (88 points): The lighting fusion is excellent, but the edge fusion is slightly unnatural, deducting points. Spatiotemporal Consistency (Not applicable, this is a static scene). Overall Score: 90 points. 2. Diagnosis and Feedback: The system judges that 90 points is greater than the preset 80-point passing line, and judges it as qualified. The scene is automatically stored in the "High-Quality Training Scene Library" and marked as "Bus Contamination Discharge - Moderate Level". Hypothetical Scene: If the score is only 70 points, the system will diagnose the specific problem: "The direction of the shadow of the contamination is 25 degrees away from the direction of the background light source". Then it will automatically feed back this problem to step S4, and S4 will readjust the "fusion formula" accordingly, fuse again, and resubmit for inspection until the score meets the standard.

[0052] Final Application: Maintenance personnel log into the training system and retrieve the "Main Transformer A-phase Bushing Pollution Discharge" scenario from the scenario library. They can observe the defect characteristics from different angles in a highly realistic simulation environment. When using the "Severity Slider," they can intuitively see the development process of the defect from minor to severe, gaining a deep understanding of the risk differences under different pollution levels.

[0053] In the above implementation, the constructed spatiotemporal related scenario library covers multiple environments, multiple angles, and multiple time nodes, solving the problem of the single nature of traditional simulation scenarios; the defect condition tracing and correlation library establishes a deep correlation between defects and operating conditions, and combined with multi-dimensional constraints, it enables virtual defects to have strong logical rationality, overcoming the drawback of virtual defects lacking constraints; the use of AIGC technology and adaptive fusion algorithms improves the visual realism and physical consistency of simulation scenarios, and can cover special scenarios such as rare defects and composite defects; the introduction of a realism quantification scoring and feedback optimization mechanism ensures the quality of simulation scenarios, and can provide high-quality data support for immersive training of operation and maintenance personnel, defect detection model training, and power grid risk early warning, significantly improving the level of intelligent operation and maintenance of substations.

[0054] Example 2

[0055] like Figure 2 As shown in the figure, Embodiment 2 of this application provides a structural diagram of a substation defect scene generation and fusion simulation device.

[0056] As an example, the device is implemented using the substation defect scenario generation and fusion simulation method described in Example 1, and the device includes:

[0057] The spatiotemporal correlation scene library construction module 210 is suitable for collecting images of substation sites under normal conditions, from different environments, angles, and time points, and recording the collection parameters. Based on the collection parameters, it performs multi-dimensional spatiotemporal scene correlation to construct a spatiotemporal correlation scene library.

[0058] The defect condition tracing and association library construction module 220 is suitable for collecting defect images and defect feature parameters of historical faults in substations, collecting substation operating condition data and associating them with the defect images to construct a defect condition tracing and association library.

[0059] The substation virtual defect image generation module 230 is suitable for parsing the association relationship of the defect condition traceability association library and deduce multi-dimensional defect generation constraints. Based on the traceability association library and constraints, it generates substation virtual defect images and virtual defect parameters through artificial intelligence content generation technology.

[0060] The substation defect simulation scenario construction module 240 is suitable for constructing substation defect simulation scenarios by adaptively fusing virtual defect images and spatiotemporally related scene libraries based on virtual defect parameters and using a scene defect adaptive fusion algorithm.

[0061] The authenticity quantification scoring module 250 is suitable for diagnosing substation defect simulation scenarios through defect scenario diagnosis algorithms, and determining whether the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification scoring threshold.

[0062] The substation defect simulation scenario output module 260 is suitable for marking a substation defect simulation scenario as a successfully fused scenario in response to the fact that the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold, and outputting the substation defect simulation scenario and its detailed parameter report.

[0063] The optimization instruction feedback module 270 is suitable for responding to situations where the authenticity quantification score of a substation defect simulation scenario does not meet the preset authenticity quantification score threshold. It automatically locates abnormal problems in the substation defect simulation scenario and feeds back optimization instructions containing the location information of the abnormal problems to the substation defect simulation scenario construction module to trigger the optimization and reconstruction of the substation defect simulation scenario.

[0064] It is not difficult to see that this embodiment is a system implementation corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.

[0065] It is worth mentioning that all modules involved in this embodiment are logical units. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0066] Example 3

[0067] Please see Figure 3 The present invention also provides an electronic device, including: a memory and a processor; the memory stores at least one program instruction; the processor loads and executes the at least one program instruction to implement the substation defect scenario generation and fusion simulation method provided in Embodiment 1.

[0068] The memory 302 and processor 301 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 301 and memory 302 together. The bus may also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 301 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 301.

[0069] Processor 301 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 302 can be used to store data used by processor 301 during operation.

[0070] Example 4

[0071] This invention also proposes a storage medium storing a method for generating and fusing simulations of substation defect scenarios. When the substation defect scenario generation and fusing simulation program is executed by a processor, it implements the steps of the substation defect scenario generation and fusing simulation method described above. Since this storage medium adopts all the technical solutions of all the above embodiments, it has at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated here.

[0072] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, based on the guidance provided in this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for generating and fusing simulations of substation defect scenarios, characterized in that, include: Step S1: Collect images of the substation site under normal conditions, from different environments, angles, and time points, and record the collection parameters. Based on the collection parameters, perform multi-dimensional spatiotemporal scene association to construct a spatiotemporal associated scene library. Step S2: Collect defect images and defect feature parameters of historical faults in the substation, collect substation operating condition data and associate them with the defect images to build a defect operating condition traceability association library. Step S3: Analyze the association relationship of the defect condition tracing association library and deduce the multi-dimensional defect generation constraints. Based on the tracing association library and constraints, generate virtual defect images and virtual defect parameters of the substation through artificial intelligence content generation technology. Step S4: Based on the virtual defect parameters, the virtual defect image and the spatiotemporal related scene library are adaptively fused using a scene defect adaptive fusion algorithm to construct a substation defect simulation scene. Step S5: Diagnose the substation defect simulation scenario using the defect scenario diagnosis algorithm, and determine whether the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold. Step S6: In response to the fact that the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification score threshold, mark this substation defect simulation scenario as a successfully fused scenario and output the substation defect simulation scenario and its detailed parameter report. Step S7: In response to the fact that the quantitative score of the substation defect simulation scenario does not meet the preset quantitative score threshold, based on the non-compliance of the three core evaluation indicators of physical rationality, visual realism, and spatiotemporal consistency, the specific abnormal problem of the substation defect simulation scenario is automatically located, and the optimization instruction containing the abnormal problem location information is fed back to step S4 to trigger the optimization and reconstruction of the substation defect simulation scenario. The location information of the abnormal problem includes the non-compliance evaluation indicator category, the specific non-compliance evaluation item, and the corresponding quantitative deviation data.

2. The substation defect scenario generation and fusion simulation method according to claim 1, characterized in that, The acquisition parameters include one or a combination of acquisition time, environmental data, acquisition angle, equipment type, equipment spatial coordinates, equipment operation data, acquisition equipment parameters, and historical defect records. The environmental data includes temperature, light intensity, and weather type. Step S1 includes: Image-physical location mapping is established based on device spatial coordinates to generate a 3D point cloud map. Images of devices at the same location are sorted by acquisition time to form a time series. The trend of device status change is marked by combining acquisition parameters. After generating a spatiotemporal index, image scene sub-libraries are divided according to device type, environment, and angle, and integrated to form a spatiotemporally related scene library.

3. The substation defect scenario generation and fusion simulation method according to claim 1, characterized in that, Step S2 includes: Step S201: Extract defect images from historical fault data and automatically extract their geometric, physical and appearance feature parameters as defect feature parameters; Step S202: Obtain historical operating condition data corresponding to the defect image, wherein the historical operating condition data includes electrical parameters, environmental parameters and equipment status parameters; Step S203: Establish the mapping relationship between defect types, defect characteristic parameters and historical operating condition data, store them in a structured form, and form the defect operating condition traceability association library.

4. The substation defect scenario generation and fusion simulation method according to claim 1, characterized in that, The multi-dimensional defect generation constraints in step S3 include physical constraints, operating condition constraints, and environmental constraints. The physical constraints limit the size and shape of the defect to not exceed the range of the equipment body. The operating condition constraints limit the generation of defects only under the corresponding operating conditions. The environmental constraints limit the generation of defects to match the corresponding environmental parameters.

5. The substation defect scenario generation and fusion simulation method according to claim 4, characterized in that, Step S3 includes: The constraints are encoded into condition vectors and input into a pre-trained condition generation model to generate a virtual defect image and detailed virtual defect parameters that match the condition vectors. The conditional generation model is a conditional generative adversarial network or a potential diffusion model; the virtual defect parameters include one or a combination of defective equipment, defect type, location coordinates, geometric information, physical parameters, and associated virtual working condition data.

6. The substation defect scenario generation and fusion simulation method according to claim 4, characterized in that, Step S4 includes: Step S401: Perform multimodal parameter matching and fusion based on the spatiotemporal related scene library and virtual defect parameters to generate a scene defect fusion parameter set to guide image fusion; Step S402: Based on the scene defect fusion parameter set, the substation site images and virtual substation defect images in the spatiotemporal related scene library are adaptively fused using the scene defect adaptive fusion algorithm to construct a substation defect simulation scene. The execution of the scene defect adaptive fusion algorithm is controlled by the scene defect fusion parameter set.

7. The substation defect scenario generation and fusion simulation method according to claim 6, characterized in that, The multimodal parameter matching and fusion in step S401 includes: The virtual defect parameters and parameters in the spatiotemporal related scenario library are divided into spatial attribute parameters, physical and environmental attribute parameters, and logical characteristic attribute parameters. The parameters are dimensionless. Assign a matching fusion factor, error adjustment coefficient, and fusion orthogonal weight to each individual index in each parameter class; The scene defect fusion parameter set is generated through multi-dimensional collaborative computing; The spatial attribute parameters include at least one of location coordinates, geometric information, defect impact range, image angle, and spatial index; the physical and environmental attribute parameters include at least one of defect operating conditions, appearance features, equipment operating conditions, environmental data, and image physical features; the logical feature attribute parameters include at least one of equipment type, defect type, defect triggering conditions, development sequence, credibility score, acquisition time, time series, and historical defects of the equipment.

8. The substation defect scenario generation and fusion simulation method according to claim 7, characterized in that, The execution of the scene defect adaptive fusion algorithm in step S402 is controlled by the scene defect fusion parameter set, including: Based on the spatial attribute parameters in the scene defect fusion parameter set, the substation virtual defect image is subjected to geometric transformation and spatial localization. Based on the physical and environmental attribute parameters in the scene defect fusion parameter set, the virtual defect image of the substation is subjected to relighting, material fusion and environmental effect rendering. Based on the logical feature attribute parameters in the defect fusion parameter set of the scenario, the rationality of the fusion result is verified and the defect parameter control mechanism is associated.

9. The substation defect scenario generation and fusion simulation method according to claim 1, characterized in that, Step S5 includes: The simulation of substation defects is quantitatively scored using a defect scene diagnosis and scoring algorithm. This algorithm includes collecting three core evaluation indicators: physical rationality, visual realism, and spatiotemporal consistency. Evaluation thresholds are set for individual evaluation items under each indicator category. The quantitative score is obtained through weighted collaborative calculation of multiple indicators. Specifically, the physical rationality evaluation indicators include evaluating at least one of defect morphology rationality, location rationality, and operating condition matching; the visual realism evaluation indicators include evaluating at least one of illumination consistency, texture blending, color consistency, and noise matching; and the spatiotemporal consistency evaluation indicators include evaluating at least one of dynamic time decay coefficient, adjacent frame time interval, and spatiotemporal overlap.

10. A substation defect scene generation and fusion simulation device, wherein the device is implemented using the substation defect scene generation and fusion simulation method according to any one of claims 1-9, characterized in that, The device includes: The spatiotemporal correlation scene library construction module is suitable for collecting images of substation sites under normal conditions from different environments, angles, and time points, recording the collection parameters, and performing multi-dimensional spatiotemporal scene correlation based on the collection parameters to construct a spatiotemporal correlation scene library. The defect condition tracing and association library construction module is suitable for collecting defect images and defect feature parameters of historical faults in substations, collecting substation operating condition data and associating them with the defect images to construct a defect condition tracing and association library. The substation virtual defect image generation module is suitable for parsing the correlation relationship of the defect condition traceability association library and deduce multi-dimensional defect generation constraints. Based on the traceability association library and constraints, it generates substation virtual defect images and virtual defect parameters through artificial intelligence content generation technology. The substation defect simulation scenario construction module is suitable for constructing substation defect simulation scenarios by adaptively fusing virtual defect images and spatiotemporally related scene libraries based on virtual defect parameters and using a scene defect adaptive fusion algorithm. The authenticity quantification scoring module is suitable for diagnosing substation defect simulation scenarios using defect scenario diagnosis algorithms, and determining whether the authenticity quantification score of the substation defect simulation scenario meets the preset authenticity quantification scoring threshold. The substation defect simulation scenario output module is suitable for responding to the substation defect simulation scenario's authenticity quantification score meeting the preset authenticity quantification score threshold, marking this substation defect simulation scenario as a successfully fused scenario, and outputting the substation defect simulation scenario and its detailed parameter report. The optimized instruction feedback module is suitable for responding to situations where the authenticity quantification score of a substation defect simulation scenario does not meet the preset authenticity quantification score threshold. It automatically locates abnormal problems in the substation defect simulation scenario and feeds back optimization instructions containing the location information of the abnormal problems to the substation defect simulation scenario construction module to trigger the optimization and reconstruction of the substation defect simulation scenario.

Images