AI-driven Gaussian 3D substation modeling optimization method and system

Abstract

The present invention discloses an AI-driven Gaussian three-dimensional substation modeling optimization method and system, which relates to the technical field related to data processing. The method includes: obtaining a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment by performing multi-angle data collection and feature extraction on a set of substation equipment. An AI-driven algorithm channel is established to perform rendering modeling and optimization training on the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment, and to build a Gaussian three-dimensional model of the target substation. A multi-task learning framework is constructed to perform predictive analysis on the real-time operation data stream of the substation to obtain a set of substation operation analysis parameters. A reinforcement learning mechanism is introduced to adaptively optimize and update the Gaussian three-dimensional model of the target substation to generate a Gaussian three-dimensional optimized model of the substation. This solves the technical problem that the three-dimensional modeling method of the substation in the prior art has low three-dimensional model accuracy and fidelity, and cannot meet the needs of intelligent operation and maintenance.

Description

Technical Field

[0001] The present application relates to the technical field related to data processing, and specifically to an AI-driven Gaussian three-dimensional substation modeling optimization method and system. Background Art

[0002] With the development of smart grids and digital operations and maintenance (O&M), 3D modeling of substations is playing an increasingly important role in equipment management, fault diagnosis, and maintenance planning. Traditional 3D modeling methods for substations rely primarily on manual surveying and modeling, which is not only time-consuming and labor-intensive but also prone to inaccuracies and missing details. In recent years, the rapid development of computer vision and deep learning technologies has made it possible to automate data collection using image recognition and laser scanning equipment. High-resolution industrial cameras, drone-mounted cameras, and 3D lidar (lidar) devices can capture images and point cloud data from multiple angles of substation equipment. However, these multi-angle image and point cloud datasets often contain significant noise and redundant information, making the resulting 3D models inadequate for intelligent O&M.

[0003] Therefore, in the existing technology, the three-dimensional modeling method of the substation has the technical problem that the three-dimensional model accuracy and realism are not high and cannot meet the needs of intelligent operation and maintenance. Summary of the Invention

[0004] This application provides an AI-driven Gaussian 3D substation modeling optimization method and system, addressing the technical issues of existing substation 3D modeling methods, which suffer from low 3D model accuracy and fidelity, failing to meet the requirements of intelligent operation and maintenance. This method achieves the construction of high-precision, high-fidelity 3D substation models. Furthermore, through the introduction of a multi-task learning framework and transfer learning strategies, it further improves the accuracy of operational analysis and the adaptability of the model, thereby achieving intelligent and efficient 3D substation modeling and operation and maintenance management.

[0005] The present application provides an AI-driven Gaussian three-dimensional substation modeling optimization method, which includes: performing multi-angle data acquisition and feature extraction on a substation equipment set through an image recognition device and a laser scanning device to obtain a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment; building an AI-driven algorithm channel, the AI-driven algorithm channel including a 3DGS algorithm channel and a GAN algorithm channel, using the AI-driven algorithm channel to perform rendering modeling and optimization training on the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a Gaussian three-dimensional model of the target substation; constructing a multi-task learning framework, obtaining a multi-task substation operation analyzer based on the multi-task substation operation analyzer, performing predictive analysis on the real-time operation data stream of the substation based on the multi-task substation operation analyzer, and obtaining a substation operation analysis parameter set; introducing a reinforcement learning mechanism to adaptively optimize and update the target substation Gaussian three-dimensional model based on the substation operation analysis parameter set to generate a substation Gaussian three-dimensional optimization model.

[0006] The present application also provides an AI-driven Gaussian three-dimensional substation modeling optimization system, including: a feature extraction module, which is used to perform multi-angle data acquisition and feature extraction on a substation equipment set through an image recognition device and a laser scanning device to obtain a multi-dimensional image feature set of the substation equipment and a multi-dimensional point cloud feature set of the substation equipment; a three-dimensional model construction module, which is used to build an AI-driven algorithm channel, and the AI-driven algorithm channel includes a 3DGS algorithm channel and a GAN algorithm channel, and uses the AI-driven algorithm channel to render, model and optimize the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a target substation Gaussian three-dimensional model; an operation analysis module, which is used to build a multi-task learning framework, obtain a multi-task substation operation analyzer based on the multi-task substation operation analyzer, and perform predictive analysis on the real-time operation data stream of the substation based on the multi-task substation operation analyzer to obtain a substation operation analysis parameter set; a model optimization module, which is used to introduce a reinforcement learning mechanism to adaptively optimize and update the target substation Gaussian three-dimensional model based on the substation operation analysis parameter set to generate a substation Gaussian three-dimensional optimization model.

[0007] The AI-driven Gaussian 3D substation modeling optimization method and system proposed in this application uses image recognition equipment and laser scanning equipment to perform multi-angle data acquisition and feature extraction on a substation equipment set, obtaining a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment; constructing an AI-driven algorithm channel, which includes a 3DGS algorithm channel and a GAN algorithm channel, and using the AI-driven algorithm channel to perform rendering modeling and optimization training on the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a Gaussian 3D model of the target substation; constructing a multi-task learning framework, and obtaining a multi-task substation operation analyzer based on the transfer training of the multi-task learning framework. Based on the multi-task substation operation analyzer, predictive analysis is performed on the real-time substation operation data stream to obtain a substation operation analysis parameter set; and introducing a reinforcement learning mechanism to adaptively optimize and update the Gaussian 3D model of the target substation based on the substation operation analysis parameter set to generate a Gaussian 3D optimized substation model. This solves the technical problem that the existing substation 3D modeling method has low 3D model accuracy and fidelity, which cannot meet the needs of intelligent operation and maintenance. The construction of a high-precision and high-fidelity three-dimensional substation model has been achieved. At the same time, through the introduction of a multi-task learning framework and transfer learning strategy, the accuracy of operation analysis and the adaptability of the model have been further improved, thereby achieving the technical effect of intelligent and efficient three-dimensional modeling and operation and maintenance management of substations. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] To more clearly illustrate the technical solutions of the embodiments of the present disclosure, the accompanying drawings of the embodiments of the present disclosure are briefly introduced below. Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in precise order. Instead, various steps may be processed in reverse order or simultaneously as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0009] Figure 1 A schematic diagram of the process flow of the AI-driven Gaussian 3D substation modeling optimization method provided in an embodiment of the present application;

[0010] Figure 2 Schematic diagram of the structure of the AI-driven Gaussian three-dimensional substation modeling and optimization system provided in an embodiment of the present application.

[0011] Explanation of the reference numerals: feature extraction module 11 , three-dimensional model construction module 12 , operation analysis module 13 , model optimization module 14 . DETAILED DESCRIPTION

[0012] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0013] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0014] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict, and the terms “first\second” involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. The terms “including” and “having” and any variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or that are inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. The terms used herein are for the purpose of describing the embodiments of this application only.

[0015] The present application embodiment provides an AI-driven Gaussian three-dimensional substation modeling optimization method and system, such as Figure 1 As shown, the method includes:

[0016] Multi-angle data collection and feature extraction are performed on the substation equipment set through image recognition equipment and laser scanning equipment to obtain a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment; an AI-driven algorithm channel is built, and the AI-driven algorithm channel includes a 3DGS algorithm channel and a GAN algorithm channel. The AI-driven algorithm channel is used to render, model and optimize the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a Gaussian three-dimensional model of the target substation.

[0017] Image recognition equipment and laser scanning equipment are used to collect multi-angle data from a set of substation equipment. The set is a pre-defined set of all equipment required for data collection, resulting in a multi-angle image set and a multi-angle point cloud dataset of the substation equipment. The image recognition device is hardware capable of capturing visible light or infrared images and performing preliminary recognition processing, such as a high-resolution industrial camera or a camera mounted on a drone. The laser scanning device is a 3D lidar, which acquires 3D coordinate information of the target scene by emitting a laser beam and receiving the reflected signal. Feature extraction is performed on the multi-angle image set and the multi-angle point cloud dataset of the substation equipment to obtain a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment. Furthermore, an AI-driven algorithm pipeline is established, comprising a 3DGS algorithm channel and a GAN algorithm channel. 3DGS is a three-dimensional Gaussian surface construction method. It generates a smooth 3D surface model by applying Gaussian weighting to feature points and performing surface fitting. The GAN (Generative Adversarial Network) generator is responsible for producing more realistic 3D models, while the discriminator scores and judges the generated results and provides feedback to the generator for further improvement. The AI-driven algorithm pipeline is used to render, model, and optimize the multi-dimensional image feature set and multi-dimensional point cloud feature set of the substation equipment to build a Gaussian 3D model of the target substation.

[0018] The method provided in the embodiment of the present application also includes: performing edge recognition on the multi-angle image set and the multi-angle point cloud data set of the substation equipment to obtain a substation equipment image edge information set and a substation equipment point cloud edge information set; using a Gaussian filter based on the substation equipment image edge information set to filter and enhance the multi-angle image set of the substation equipment to obtain a usable multi-angle image set of substation equipment; determining a rectangular fitting rule according to the substation equipment point cloud edge information set, and performing filtering and denoising processing on the multi-angle point cloud data set of the substation equipment based on the rectangular fitting rule to obtain a usable multi-angle point cloud data set of substation equipment; using a convolutional neural network to extract key features from the usable multi-angle image set of the substation equipment and the usable multi-angle point cloud data set of the substation equipment to obtain a multi-dimensional image feature set of the substation equipment and a multi-dimensional point cloud feature set of the substation equipment.

[0019] Obtaining a multi-dimensional image feature set and a multi-dimensional point cloud feature set of substation equipment includes: performing edge recognition on the multi-angle image set of substation equipment, performing grayscale conversion on the images in the multi-angle image set to obtain converted image data; further extracting edge information using operators such as Canny and Sobel or a deep learning model to obtain edge contour information for each device image in the multi-angle image set, thereby obtaining a substation equipment image edge information set containing information such as the coordinates of primary edge pixels and device contours in each device image; performing edge recognition on the multi-angle point cloud dataset of substation equipment to obtain edges for each image in the multi-angle point cloud dataset of substation equipment, thereby obtaining a substation equipment point cloud edge information set; and then, filtering and enhancing the multi-angle image set of substation equipment using a Gaussian filter based on the edge information set of the substation equipment images. A smaller filter kernel may be used near edges to maximize contour preservation, while a larger filter kernel may be used in non-edge areas to reduce noise interference. After filtering, a usable multi-angle image set of substation equipment is obtained.

[0020] Furthermore, according to the substation equipment point cloud edge information set, professional and technical personnel set rectangular fitting parameters, including: distance threshold, angle threshold, minimum point number threshold, etc., to control the accuracy of searching for rectangular edges in the point cloud, and then determine the rectangular fitting rule. The rectangular fitting rule is used to find a point set or boundary that meets the rectangular shape characteristics in a point cloud containing a large number of three-dimensional point coordinates. The rectangular or approximately rectangular part of the equipment (such as the transformer base, the facade of the switch cabinet, etc.) is identified. The point cloud in the multi-angle point cloud dataset of the substation equipment is searched and matched using the rectangular fitting rule, and outliers (noise points) that do not conform to the rectangular outline or deviate from the main body of the equipment are filtered out. The filtering and denoising process is completed to obtain a usable multi-angle point cloud dataset of the substation equipment. In the usable multi-angle point cloud dataset of the substation equipment, the key points on the surface or edge of the equipment are retained, and redundant data that interferes with subsequent modeling is removed. Finally, the available multi-angle image set of substation equipment is fed into a pre-trained or custom-trained CNN (e.g., a classic network like ResNet or VGG, or a specialized network designed specifically for power equipment features). Multiple convolutional and pooling layers extract deep features from the images, revealing information such as the surface texture of the equipment, the shape descriptions of different components, and color distribution, thereby obtaining a multi-dimensional image feature set for the substation equipment. A convolutional neural network is then used to extract key features from the multi-angle point cloud dataset of available substation equipment. This dataset is then fed into a point cloud-specific neural network, such as PointNet, PointNet++, or another 3D CNN, to extract point cloud distribution features. This information reflects the equipment's geometric structure, surface information, and the spatial locations of key components in 3D space, ultimately resulting in a multi-dimensional point cloud feature set for the substation equipment.

[0021] The method provided in an embodiment of the present application also includes: determining a rectangle fitting rule according to the substation equipment point cloud edge information set, and setting rectangle fitting parameters based on the rectangle fitting rule; randomly selecting four points in the substation equipment multi-angle point cloud dataset, connecting them to obtain a first fitting rectangle as a rectangle optimal solution, and iteratively fitting and comparing the rectangle optimal solution using the rectangle fitting parameters to obtain a rectangle fitting result; performing point cloud data denoising processing on the substation equipment multi-angle point cloud dataset based on the rectangle fitting result to obtain the usable substation equipment multi-angle point cloud dataset within a preset point cloud threshold.

[0022] Obtaining a usable multi-angle point cloud dataset of substation equipment includes determining a rectangle fitting rule based on the substation equipment point cloud edge information set. The rectangle fitting rule is a set of criteria and constraints for identifying and fitting rectangular or approximately rectangular shapes in a point cloud. The rectangle fitting rule can be configured by a professional technician based on the identification object, such as setting a maximum side length, a minimum side length, an angle range, a side length ratio, etc. Based on the rectangle fitting rule, rectangle fitting parameters are set. These parameters include a maximum number of iterations, a distance threshold (used to determine the distance between a point and the edge of the fitted rectangle; for example, if the distance threshold is set to 5 mm, points within 5 mm of the edge of the fitted rectangle are considered eligible points), an angle threshold (used to determine the angular relationship between points; for example, if the angle threshold is set to ±5 degrees, the angle of the fitted rectangle should be between 85 and 95 degrees), and a minimum point count threshold (to ensure that the number of points in the fitting area is sufficient to eliminate noise or false edges). Furthermore, four points are randomly selected from the multi-angle point cloud dataset of substation equipment as candidate points for the initial fitted rectangle. The purpose of random selection is to increase diversity in the fitting process and avoid bias caused by fixed selection. The first fitted rectangle is connected as the optimal rectangular solution, and the optimal rectangular solution is iteratively fitted and compared using the rectangle fitting parameters to obtain a rectangular fitting result. For example, four points A, B, C, and D are randomly selected from the switchgear point cloud dataset and attempted to be connected to form a rectangle. The length ratio and angle of each side are calculated. If they meet the set fitting rules, they are recorded as candidate rectangles. Multiple random selections and evaluations are performed within a maximum number of iterations, and the rectangle that best matches the switchgear facade features is ultimately selected as the optimal solution. The iterative process involves comparing the number of point cloud data points whose iterative rectangles are less than a distance threshold. If the number of second fitted rectangles exceeds the number of first fitted rectangles, the second fitted rectangle is replaced as the optimal solution. For each point in the point cloud dataset, the distance to the edge of the rectangular fitting result is calculated. If the point is within a preset point cloud threshold, the point is retained; otherwise, it is considered a noise point and removed, thereby obtaining the multi-angle point cloud dataset of available substation equipment. The preset point cloud threshold is a pre-set distance judgment threshold. When the distance is greater than the preset point cloud threshold, the corresponding point is outside the fitting edge; otherwise, it is within the fitting edge.

[0023] The method provided in an embodiment of the present application also includes: spatially mapping and registering the multidimensional image feature set of the substation equipment to the multidimensional point cloud feature set of the substation equipment based on a 3DGS algorithm channel to obtain a three-dimensional Gaussian feature point set; using a three-dimensional Gaussian function to calculate the influence range and perform Gaussian weighting processing on each feature point in the three-dimensional Gaussian feature point set to obtain a Gaussian weighted point cloud data set; performing surface fitting and smoothing processing based on the Gaussian weighted point cloud data set to obtain a preliminary three-dimensional model, iteratively rendering and optimizing the preliminary three-dimensional model to generate a substation Gaussian three-dimensional model; iteratively optimizing and training the substation Gaussian three-dimensional model based on the GAN algorithm channel to build a target substation Gaussian three-dimensional model.

[0024] Based on the 3DGS algorithm, the multidimensional image feature set of the substation equipment and the multidimensional point cloud feature set of the equipment are matched and converted into the same spatial coordinate system, so that the image features and point cloud features correspond to each other, completing the spatial mapping for subsequent fusion and processing. An iterative closest point algorithm is then used to align the image feature set with the point cloud feature set, achieving precise registration by minimizing the error between the two. After registration, the image features and point cloud features are fused to generate a set of three-dimensional Gaussian feature points that integrate the visual and geometric information of the equipment.

[0025] Furthermore, a three-dimensional Gaussian function is used to calculate the influence range of each feature point in the three-dimensional Gaussian feature point set, that is, to determine the contribution of the point to the surrounding area in space. At the same time, Gaussian weighting processing is performed, and a three-dimensional Gaussian function is applied to weight each feature point. The weight decreases with increasing distance to ensure that the feature point has a greater impact on its nearby area and a smaller impact on distant areas, thereby obtaining a Gaussian weighted point cloud data set. For a transformer base point cloud, the standard deviation of the Gaussian function is set to 5 mm. A smaller standard deviation of the Gaussian function means that the influence range is concentrated, while a larger standard deviation of the Gaussian function expands the influence range. For each feature point on the base, its Gaussian weight in three-dimensional space is calculated, and the weight is applied to the point cloud data to generate a smooth and continuous Gaussian weighted point cloud data set, eliminating small-scale noise points caused by scanning errors.

[0026] Subsequently, a 3D surface model is fitted to the point cloud data using a surface fitting algorithm (Poisson surface reconstruction or marching cubes) based on the Gaussian-weighted point cloud dataset. The 3D surface model is smoothed, and a filtering algorithm is used to reduce surface irregularities and improve the visual quality of the model, thereby generating a preliminary 3D model. This preliminary 3D model is then input into a rendering engine (such as Blender or Maya) for preliminary visual rendering. The model's appearance and structure are evaluated. Based on the rendering results, model deficiencies, such as missing details and rough surfaces, are identified and optimization requirements are recorded. Based on the feedback, surface fitting and smoothing parameters are adjusted (e.g., a smaller σ value or higher smoothing strength). Surface fitting and model generation are then performed again, and the rendering and optimization steps are repeated. Through multiple iterations, the accuracy and detail of the 3D model are continuously improved until the desired high quality is achieved, generating a Gaussian 3D model of the substation. Finally, the Gaussian 3D substation model is iteratively optimized and trained using the GAN algorithm pipeline to construct the target substation Gaussian 3D model.

[0027] The method provided in an embodiment of the present application also includes: building a three-dimensional model generator and a three-dimensional model discriminator according to the GAN algorithm channel, optimizing and generating the Gaussian three-dimensional model of the substation as input to the three-dimensional model generator to obtain a Gaussian optimized three-dimensional model; scoring and discriminating the Gaussian optimized three-dimensional model based on the three-dimensional model discriminator to obtain three-dimensional model performance evaluation parameters; and feeding back the three-dimensional model performance evaluation parameters to the three-dimensional model generator for iterative optimization to obtain a Gaussian three-dimensional model of the target substation.

[0028] The process of constructing a Gaussian 3D model of the target substation includes: constructing a 3D model generator and a 3D model discriminator based on the GAN algorithm pipeline. The 3D model of the substation is optimized and generated using the Gaussian 3D model as input to the 3D model generator, resulting in a Gaussian optimized 3D model. The Gaussian optimized 3D model generated by the generator is input into the discriminator, which scores the input model to determine its authenticity and obtains 3D model performance evaluation parameters. The score typically ranges from 0 to 1, with 1 representing a true model and 0 representing a generated model. The 3D model performance evaluation parameters are fed back to the 3D model generator for iterative tuning. In the first round of iterations, the transformer model generated by the generator received a score of 0.6. Based on the feedback from the discriminator, the generator adjusts its network parameters and optimizes details. In the second round of iterations, the score of the model generated by the generator improves to 0.75. As the number of iterations increases, the generator gradually generates high-quality models with scores approaching 1. When the discriminator's score for the generated model stabilizes at a high level (e.g., above 0.95) and further iterations do not significantly improve the score, the Gaussian 3D model of the target substation is obtained.

[0029] The method provided in an embodiment of the present application also includes: building an initial generator and an initial discriminator according to the GAN algorithm channel, and determining a generator-discriminator alternating training strategy; performing alternating training verification on the initial generator and the initial discriminator based on the generator-discriminator alternating training strategy to obtain generator loss data and discriminator loss data; and performing gradient balance update on the initial generator and the initial discriminator based on the generator loss data and the discriminator loss data to obtain the three-dimensional model generator and the three-dimensional model discriminator.

[0030] The construction of the three-dimensional model generator and the three-dimensional model discriminator includes: building an initial generator and an initial discriminator according to the GAN algorithm channel. The GAN algorithm is a deep learning framework consisting of two parts: a generator and a discriminator. The task of the generator is to generate realistic data samples, while the task of the discriminator is to distinguish between generated samples and real data samples. Through adversarial training between the two, the generator continuously improves the quality of the generated data until the discriminator can no longer effectively distinguish between true and false samples. Determine the generator-discriminator alternating training strategy, that is, during the training process, the generator and the discriminator are trained in turn. Specifically, first fix the generator to train the discriminator, and then fix the discriminator to train the generator to ensure that the two make progress together in the adversarial process.

[0031] Set the number of iterations for each training. Usually, a higher number of rounds can be set in the initial stage to ensure that the model has enough time to learn. Based on the generator-discriminator alternating training strategy, the initial generator and initial discriminator are alternately trained and verified. In each training round, the loss data of the generator and discriminator are recorded as a reference for model optimization. Train the discriminator: Extract a batch of real Gaussian three-dimensional models from the real data set as real samples. The generator generates a batch of Gaussian optimized three-dimensional models as fake samples. The real samples and fake samples are input into the discriminator respectively, and the discriminator's discrimination probability for the real samples and the discrimination probability for the fake samples are calculated. Based on the discrimination results of the real samples and fake samples, the loss function of the discriminator (such as binary cross entropy loss) is calculated. The weight parameters of the discriminator are updated through the backpropagation algorithm to improve its ability to distinguish between real and fake samples. Train the generator: The generator generates a batch of Gaussian optimized three-dimensional models as fake samples. The fake samples are input into the discriminator to obtain the discriminator's scores for these samples. The generator's loss function is calculated with the goal of maximizing the discriminator's score on fake samples, meaning that the discriminator considers the fake samples to be as realistic as possible. Using the backpropagation algorithm, the generator's weight parameters are updated to improve its ability to generate more realistic samples.

[0032] Finally, based on the generator loss data and the discriminator loss data, the initial generator and initial discriminator are updated with gradient balance, and the learning rate or weight update step size of the generator and discriminator are adjusted to prevent one from improving too quickly and causing the other to be unable to train effectively, ensuring balanced development of the two during the training process. When the loss values ​​of the generator and discriminator change tend to be stable over several consecutive rounds and the quality of the 3D model generated by the generator meets the expected standards, the training is terminated, and the 3D model generator and 3D model discriminator are obtained.

[0033] The method provided in the embodiment of the present application also includes: constructing a multi-task learning framework, obtaining a multi-task substation operation analyzer based on the multi-task learning framework transfer training, performing predictive analysis on the substation real-time operation data stream based on the multi-task substation operation analyzer, and obtaining a substation operation analysis parameter set; introducing a reinforcement learning mechanism to adaptively optimize and update the target substation Gaussian three-dimensional model based on the substation operation analysis parameter set to generate a substation Gaussian three-dimensional optimization model.

[0034] Multiple tasks that require simultaneous learning are identified. In substation operation analysis, these tasks include equipment status monitoring, fault prediction, and load forecasting. A multi-task learning framework is constructed based on these tasks. Transfer training is performed using this multi-task learning framework to obtain a multi-task substation operation analyzer. The multi-task substation operation analyzer performs predictive analysis on real-time substation operation data streams (e.g., real-time sensor data streams such as current, voltage, and temperature), outputs corresponding prediction indicators or parameters, and obtains a substation operation analysis parameter set. The substation operation analysis parameter set includes load change trends, equipment health scores, and fault risk levels. Furthermore, a reinforcement learning mechanism is introduced to use the substation operation analysis parameter set as feedback information for the current 3D model. The substation operation analysis parameter set is mapped to a target substation Gaussian 3D model and adaptively optimized and updated. This ensures that the constructed target substation Gaussian 3D model better reflects the operation analysis parameters of each device, thereby generating a Gaussian 3D optimized substation model. This solves the technical problem that existing substation 3D modeling methods lack high 3D model accuracy and fidelity, failing to meet the requirements of intelligent operation and maintenance. The construction of a high-precision and high-fidelity three-dimensional substation model has been achieved. At the same time, through the introduction of a multi-task learning framework and transfer learning strategy, the accuracy of operation analysis and the adaptability of the model have been further improved, thereby achieving the technical effect of intelligent and efficient three-dimensional modeling and operation and maintenance management of substations.

[0035] The method provided in an embodiment of the present application also includes: selecting a source domain equipment operation analysis dataset according to the multi-task learning framework, the source domain equipment operation analysis dataset including a multi-task operation analysis dataset of different types of equipment; analyzing and training the multi-task operation analysis datasets of the different types of equipment respectively to obtain a source domain equipment operation analyzer set; determining a substation operation analysis target domain according to the multi-task learning framework, performing correlation analysis based on the source domain equipment operation analyzer set and the substation operation analysis target domain to determine a transfer learning strategy; performing migration training and optimization on the substation historical operation dataset and the source domain equipment operation analyzer set based on the transfer learning strategy to obtain the multi-task substation operation analyzer.

[0036] The transfer training based on the multi-task learning framework to obtain a multi-task substation operation analyzer includes: selecting a source-domain device operation analysis dataset based on the multi-task learning framework. The multi-task learning framework includes an analysis framework composed of various substation device operation analyzers. The source-domain device operation analysis dataset includes multi-task operation analysis datasets for different types of equipment, including various substation core equipment such as transformers, circuit breakers, busbars, and disconnectors. Each device dataset contains data corresponding to multiple related operation tasks, such as temperature monitoring, current load, and fault detection. When selecting the source-domain device operation analysis dataset, the data volume of the source-domain device operation analysis data is determined to meet a preset data volume requirement. If it does not meet the requirement, the corresponding source-domain device is not selected. Generic operation analyzers are constructed for different device types. Each analyzer can handle multiple operation tasks for that device. During analysis training, a model is trained based on the multi-task operation analysis data of different types of equipment using a recursive neural network algorithm to obtain a set of source-domain device operation analyzers, with each device operation analyzer corresponding to a specific device. For example, for the transformer operation analyzer, according to the multi-task learning framework, the first few layers are shared convolutional layers, and the subsequent layers are responsible for temperature prediction, current load prediction, and fault detection, respectively.

[0037] Furthermore, based on the multi-task learning framework, a target domain for substation operation analysis is determined. The target domain for substation operation analysis is comprised of all target devices for which operation analysis needs to be performed. Since some devices in the target domain may not have an operation analyzer built due to a small amount of historical operation data, an association analysis is performed on the source domain device operation analyzer set for the device categories for which an operation analyzer cannot be built. The corresponding trained operation analyzers are then obtained, and a transfer learning strategy is determined based on the trained operation analyzers. The transfer learning strategy is based on the source domain device operation analyzer set and then makes minor parameter adjustments to the target domain data to adapt to the new task. The shared layer of the source domain model is used as a feature extractor, and only the task-specific layers of the target domain are trained. Finally, transfer training and optimization are performed on the substation historical operation dataset and the source domain device operation analyzer set based on the transfer learning strategy, thereby obtaining operation analyzers for all devices in the substation operation analysis target domain and obtaining the multi-task substation operation analyzer. The obtained multi-task substation operation analyzer not only has high-precision multi-task prediction capabilities, but can also flexibly adapt to the specific needs of different substations, providing strong technical support for the intelligent upgrade and digital operation and maintenance of modern power grids.

[0038] In the above, refer to Figure 1 The AI-driven Gaussian three-dimensional substation modeling optimization method according to an embodiment of the present invention is described in detail. Figure 2 The following describes an AI-driven Gaussian three-dimensional substation modeling and optimization system according to an embodiment of the present invention.

[0039] The AI-driven Gaussian three-dimensional substation modeling optimization system according to the embodiment of the present invention solves the technical problem that the existing substation three-dimensional modeling method has low three-dimensional model accuracy and fidelity, and cannot meet the needs of intelligent operation and maintenance. It realizes the construction of high-precision and high-fidelity substation three-dimensional models. At the same time, through the introduction of multi-task learning framework and transfer learning strategy, it further improves the accuracy of operation analysis and the adaptability of the model, thereby achieving the technical effect of intelligent and efficient substation three-dimensional modeling and operation and maintenance management. The AI-driven Gaussian three-dimensional substation modeling optimization system includes: a feature extraction module 11, a three-dimensional model construction module 12, an operation analysis module 13, and a model optimization module 14.

[0040] The feature extraction module 11 is used to perform multi-angle data acquisition and feature extraction on the substation equipment set through image recognition equipment and laser scanning equipment to obtain a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment.

[0041] The three-dimensional model construction module 12 is used to build an AI-driven algorithm channel, which includes a 3DGS algorithm channel and a GAN algorithm channel. The AI-driven algorithm channel is used to render, model and optimize the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a Gaussian three-dimensional model of the target substation.

[0042] The operation analysis module 13 is used to build a multi-task learning framework, obtain a multi-task substation operation analyzer based on the migration training of the multi-task learning framework, perform predictive analysis on the substation real-time operation data stream based on the multi-task substation operation analyzer, and obtain a substation operation analysis parameter set.

[0043] The model optimization module 14 is used to introduce a reinforcement learning mechanism to adaptively optimize and update the Gaussian three-dimensional model of the target substation based on the substation operation analysis parameter set to generate a Gaussian three-dimensional optimized model of the substation.

[0044] The specific configuration of the feature extraction module 11 will be described in detail below. The feature extraction module 11 may further include: synchronously acquiring images based on the elongated ceramic piece, the method comprising: arranging a multi-view camera array, the multi-view camera array being used to acquire images from various surfaces of the elongated ceramic piece; triggering the multi-view camera array to acquire images, and using an image registration algorithm to stitch and correct the multi-view images; applying an edge detection algorithm to identify surface contour features of the ceramic piece, and using texture analysis technology to extract surface texture features.

[0045] The specific configuration of the feature extraction module 11 will be described in detail below. The feature extraction module 11 may further include: obtaining the substation equipment multi-dimensional image feature set and the substation equipment multi-dimensional point cloud feature set, including: performing multi-angle data acquisition on the substation equipment set through image recognition equipment and laser scanning equipment to obtain the substation equipment multi-angle image set and the substation equipment multi-angle point cloud data set; performing edge recognition on the substation equipment multi-angle image set and the substation equipment multi-angle point cloud data set to obtain the substation equipment image edge information set and the substation equipment point cloud edge information set; using a Gaussian filter based on the substation equipment image edge information set, The multi-angle image set of substation equipment is filtered and enhanced to obtain a multi-angle image set of available substation equipment; a rectangular fitting rule is determined according to the edge information set of the substation equipment point cloud, and filtering and denoising processing is performed on the multi-angle point cloud data set of substation equipment based on the rectangular fitting rule to obtain a multi-angle point cloud data set of available substation equipment; a convolutional neural network is used to extract key features of the multi-angle image set of available substation equipment and the multi-angle point cloud data set of available substation equipment, respectively, to obtain a multi-dimensional image feature set of substation equipment and a multi-dimensional point cloud feature set of substation equipment.

[0046] The specific configuration of the three-dimensional model construction module 12 will be described in detail below. The three-dimensional model construction module 12 further includes: obtaining the available substation equipment multi-angle point cloud dataset, including: determining a rectangle fitting rule based on the substation equipment point cloud edge information set, and setting rectangle fitting parameters based on the rectangle fitting rule; randomly selecting four points in the substation equipment multi-angle point cloud dataset, connecting them to obtain a first fitting rectangle as the rectangle optimal solution, and iteratively fitting and comparing the rectangle optimal solution using the rectangle fitting parameters to obtain a rectangle fitting result; performing point cloud data denoising on the substation equipment multi-angle point cloud dataset based on the rectangle fitting result to obtain the available substation equipment multi-angle point cloud dataset within a preset point cloud threshold.

[0047] The specific configuration of the three-dimensional model construction module 12 will be described in detail below. The three-dimensional model construction module 12 may further include: the construction of the target substation Gaussian three-dimensional model, including: spatially mapping and registering the multi-dimensional image feature set of the substation equipment to the multi-dimensional point cloud feature set of the substation equipment based on the 3DGS algorithm channel to obtain a three-dimensional Gaussian feature point set; using a three-dimensional Gaussian function to calculate the influence range and Gaussian weighting of each feature point in the three-dimensional Gaussian feature point set to obtain a Gaussian weighted point cloud data set; performing surface fitting and smoothing based on the Gaussian weighted point cloud data set to obtain a preliminary three-dimensional model, iteratively rendering and optimizing the preliminary three-dimensional model to generate a substation Gaussian three-dimensional model; iteratively optimizing and training the substation Gaussian three-dimensional model based on the GAN algorithm channel to build a target substation Gaussian three-dimensional model.

[0048] The specific configuration of the 3D model construction module 12 will be described in detail below. The 3D model construction module 12 further includes: constructing a Gaussian 3D model of the target substation, including: constructing a 3D model generator and a 3D model discriminator based on the GAN algorithm channel, optimizing the substation Gaussian 3D model as input to the 3D model generator to obtain a Gaussian optimized 3D model; scoring and discriminating the Gaussian optimized 3D model based on the 3D model discriminator to obtain 3D model performance evaluation parameters; and feeding the 3D model performance evaluation parameters back to the 3D model generator for iterative optimization to obtain the Gaussian 3D model of the target substation.

[0049] The specific configuration of the operation analysis module 13 will be described in detail below. The operation analysis module 13 further includes: the construction of the three-dimensional model generator and the three-dimensional model discriminator, including: according to the GAN algorithm channel, building an initial generator and an initial discriminator, and determining a generator-discriminator alternating training strategy; based on the generator-discriminator alternating training strategy, performing alternating training verification on the initial generator and the initial discriminator to obtain generator loss data and discriminator loss data; based on the generator loss data and the discriminator loss data, performing gradient balance update on the initial generator and the initial discriminator to obtain the three-dimensional model generator and the three-dimensional model discriminator.

[0050] The specific configuration of the model optimization module 14 will be described in detail below. The model optimization module 14 further includes: the transfer training according to the multi-task learning framework to obtain a multi-task substation operation analyzer, including: selecting a source domain device operation analysis dataset according to the multi-task learning framework, the source domain device operation analysis dataset including a multi-task operation analysis dataset of different types of equipment; analyzing and training the multi-task operation analysis datasets of different types of equipment respectively to obtain a source domain device operation analyzer set; determining a substation operation analysis target domain according to the multi-task learning framework, performing correlation analysis based on the source domain device operation analyzer set and the substation operation analysis target domain, and determining a transfer learning strategy; performing transfer training and optimization on the substation historical operation dataset and the source domain device operation analyzer set based on the transfer learning strategy to obtain the multi-task substation operation analyzer.

[0051] The AI-driven Gaussian three-dimensional substation modeling optimization system provided in an embodiment of the present invention can execute the AI-driven Gaussian three-dimensional substation modeling optimization method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0052] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0053] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. AI-driven Gaussian 3D substation modeling optimization method, characterized by: The method comprises: Use image recognition equipment and laser scanning equipment to collect multi-angle data and extract features from the substation equipment set, and obtain a multi-dimensional image feature set and a multi-dimensional point cloud feature set of the substation equipment; Building an AI-driven algorithm channel, which includes a 3DGS algorithm channel and a GAN algorithm channel. Using the AI-driven algorithm channel, the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment are rendered, modeled, and optimized for training to build a Gaussian 3D model of the target substation. Constructing a multi-task learning framework, performing transfer training based on the multi-task learning framework to obtain a multi-task substation operation analyzer, and performing predictive analysis on the substation real-time operation data stream based on the multi-task substation operation analyzer to obtain a substation operation analysis parameter set; Introducing a reinforcement learning mechanism to adaptively optimize and update the Gaussian three-dimensional model of the target substation based on the substation operation analysis parameter set to generate a Gaussian three-dimensional optimized model of the substation; The multi-task substation operation analyzer is obtained by transfer training according to the multi-task learning framework, including: selecting a source domain device operation analysis dataset according to the multi-task learning framework, wherein the source domain device operation analysis dataset includes a multi-task operation analysis dataset of different types of devices; Analyze and train the multi-task operation analysis data sets of the different types of devices respectively to obtain a source domain device operation analyzer set; Determine the substation operation analysis target domain according to the multi-task learning framework, perform correlation analysis based on the source domain equipment operation analyzer set and the substation operation analysis target domain, and determine the transfer learning strategy; Based on the transfer learning strategy, migration training and optimization are performed on the substation historical operation data set and the source domain equipment operation analyzer set to obtain the multi-task substation operation analyzer.

2. The AI-driven Gaussian three-dimensional substation modeling optimization method according to claim 1 is characterized in that: The step of obtaining a multi-dimensional image feature set and a multi-dimensional point cloud feature set of substation equipment includes: Use image recognition equipment and laser scanning equipment to collect multi-angle data of substation equipment, and obtain multi-angle image sets and multi-angle point cloud data sets of substation equipment; Performing edge recognition on the multi-angle image set and the multi-angle point cloud data set of the substation equipment to obtain an edge information set of the substation equipment image and an edge information set of the substation equipment point cloud; Based on the substation equipment image edge information set, a Gaussian filter is used to filter and enhance the substation equipment multi-angle image set to obtain a usable substation equipment multi-angle image set; Determining a rectangular fitting rule based on the substation equipment point cloud edge information set, and performing filtering and denoising processing on the substation equipment multi-angle point cloud dataset based on the rectangular fitting rule to obtain a usable substation equipment multi-angle point cloud dataset; A convolutional neural network is used to extract key features of the available substation equipment multi-angle image set and the available substation equipment multi-angle point cloud data set, respectively, to obtain the substation equipment multi-dimensional image feature set and the substation equipment multi-dimensional point cloud feature set.

3. The AI-driven Gaussian three-dimensional substation modeling optimization method according to claim 2, characterized in that: The obtained multi-angle point cloud dataset of available substation equipment includes: Determining a rectangle fitting rule according to the substation equipment point cloud edge information set, and setting rectangle fitting parameters based on the rectangle fitting rule; Randomly selecting four points in the multi-angle point cloud data set of the substation equipment, connecting them to obtain a first fitting rectangle as a rectangular optimal solution, and iteratively fitting and comparing the rectangular optimal solution using the rectangular fitting parameters to obtain a rectangular fitting result; Point cloud data denoising processing is performed on the multi-angle point cloud dataset of the substation equipment based on the rectangle fitting result to obtain the usable multi-angle point cloud dataset of the substation equipment within a preset point cloud threshold.

4. The AI-driven Gaussian three-dimensional substation modeling optimization method according to claim 1, characterized in that: The construction of the target substation Gaussian three-dimensional model includes: Based on the 3DGS algorithm channel, the multidimensional image feature set of the substation equipment is spatially mapped and registered to the multidimensional point cloud feature set of the substation equipment to obtain a three-dimensional Gaussian feature point set; Using a three-dimensional Gaussian function to perform influence range calculation and Gaussian weighting processing on each feature point in the three-dimensional Gaussian feature point set to obtain a Gaussian weighted point cloud data set; Performing surface fitting and smoothing processing on the Gaussian weighted point cloud data set to obtain a preliminary three-dimensional model, and performing iterative rendering optimization on the preliminary three-dimensional model to generate a Gaussian three-dimensional model of the substation; The substation Gaussian three-dimensional model is iteratively optimized and trained based on the GAN algorithm channel to build a target substation Gaussian three-dimensional model.

5. The AI-driven Gaussian three-dimensional substation modeling optimization method according to claim 4 is characterized in that: The construction of the target substation Gaussian three-dimensional model includes: According to the GAN algorithm channel, a 3D model generator and a 3D model discriminator are built, and the Gaussian 3D model of the substation is used as input of the 3D model generator for optimization and generation to obtain a Gaussian optimized 3D model; Scoring and discriminating the Gaussian optimized three-dimensional model based on the three-dimensional model discriminator to obtain three-dimensional model performance evaluation parameters; The three-dimensional model performance evaluation parameters are fed back to the three-dimensional model generator for iterative optimization to obtain a Gaussian three-dimensional model of the target substation.

6. The AI-driven Gaussian three-dimensional substation modeling optimization method according to claim 5, characterized in that: The construction of the three-dimensional model generator and the three-dimensional model discriminator includes: According to the GAN algorithm pipeline, build the initial generator and initial discriminator, and determine the generator-discriminator alternating training strategy; Performing alternating training verification on the initial generator and the initial discriminator based on the generator-discriminator alternating training strategy to obtain generator loss data and discriminator loss data; Based on the generator loss data and the discriminator loss data, the initial generator and the initial discriminator are updated with gradient balance to obtain the three-dimensional model generator and the three-dimensional model discriminator.

7. AI-driven Gaussian 3D substation modeling and optimization system, characterized by: The system is used to perform the method according to any one of claims 1 to 6, and the system includes: A feature extraction module is used to collect multi-angle data and extract features from a set of substation equipment using image recognition equipment and laser scanning equipment, thereby obtaining a multi-dimensional image feature set and a multi-dimensional point cloud feature set of substation equipment. A three-dimensional model construction module is used to build an AI-driven algorithm channel, which includes a 3DGS algorithm channel and a GAN algorithm channel. The AI-driven algorithm channel is used to render, model, and optimize the multi-dimensional image feature set and the multi-dimensional point cloud feature set of the substation equipment to build a Gaussian three-dimensional model of the target substation; An operation analysis module is used to build a multi-task learning framework, obtain a multi-task substation operation analyzer based on the transfer training of the multi-task learning framework, and perform predictive analysis on the substation real-time operation data stream based on the multi-task substation operation analyzer to obtain a substation operation analysis parameter set; The model optimization module is used to introduce a reinforcement learning mechanism to adaptively optimize and update the Gaussian three-dimensional model of the target substation based on the substation operation analysis parameter set to generate a Gaussian three-dimensional optimized model of the substation.

Images