Power dispatching model lightweight method and system under meta universe, and storage medium

Abstract

The invention relates to a lightweight method and system for a power dispatching model under meta-universe and a storage medium, and the method comprises the following steps: carrying out the intelligent data simplification of a three-dimensional model based on a deep learning adaptive reduction compression technology in combination with an advanced texture mapping and curve thinning algorithm, so as to optimize the details of the model and reduce the data size, thereby achieving the intelligent data simplification of the three-dimensional model; meanwhile, key features are reserved; through power grid cascade loading and visual elimination, display details of a power grid model are dynamically adjusted according to the visual angle of an observer, and model resource management and dynamic scheduling are realized through asynchronous loading and rendering; in combination with an intelligent preloading and prediction mechanism, a multi-level-of-detail technology and a progressive loading technology are applied to realize rapid loading and rendering of the model, and resource allocation and detail management are refined. According to the invention, the lightweight processing of the three-dimensional model of the power dispatching element universe can be realized, and the goals of real-time efficient rendering and rapid interaction of meteorological data, power grid operation information and a plant station equipment model are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid control and regulation, and relates to a method for lightweighting a power dispatching model, and in particular to a method, system and storage medium for lightweighting a power dispatching model under a metaverse. Background Art

[0002] The power system is undergoing a digital and intelligent transformation, facing the challenges of energy structure transformation and grid modernization. The coupling relationship between the operation decision-making of the new power system and the spatiotemporal of the meteorological environment is becoming increasingly close and complex, and the existing power grid model is difficult to achieve effective integration and visualization of multi-source information such as geography, meteorology, power grid operation data, and plant equipment status. The main reason is that the loading and rendering efficiency of the model is low. Large-scale data processing puts higher requirements on hardware equipment and increases the construction and maintenance costs of power control. Therefore, it is necessary to lightweight the three-dimensional model of the power dispatch metaverse.

[0003] After searching, no public documents of the prior art that are identical or similar to the present invention were found. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the above-mentioned prior art, and to propose a lightweight method, system and storage medium for a power dispatching model under a metaverse, so as to perform lightweight processing on the three-dimensional model of the power dispatching metaverse.

[0005] The present invention solves the practical problem by adopting the following technical solutions:

[0006] A lightweight method for power dispatching model under the metaverse includes the following steps:

[0007] Adaptive surface reduction compression technology based on deep learning, combined with texture mapping and curve thinning algorithms, intelligently simplifies the data of the power dispatch model to obtain a power dispatch model that retains key features;

[0008] For the power dispatching model retaining key features, a power dispatching model with optimized display details is obtained through grid cascade loading, visual culling, asynchronous loading and rendering;

[0009] The power dispatching model with optimized display details is combined with intelligent preloading and prediction mechanism, and multi-level-of-detail technology and progressive loading technology are applied to realize the lightweight of the power dispatching model.

[0010] Moreover, the adaptive surface reduction compression technology based on deep learning, combined with texture mapping and curve thinning algorithm, performs intelligent data simplification on the power dispatching model, and the specific steps of obtaining the power dispatching model retaining key features include:

[0011] Dynamically determine the degree of simplification based on the geometric characteristics and visual importance of the model, perform geometric and topological analysis on the power dispatch model, identify the texture details, edge contours and shapes of the model, evaluate the contribution of each polygon to the visual effect of the model, retain polygons and texture details that have a significant impact on the overall appearance, and remove or simplify other polygons;

[0012] Convert high-resolution texture maps into a CPU-based compression format, and after reasonable texture cutting and optimization, use texture mapping technology to map the two-dimensional texture to the surface of the three-dimensional model;

[0013] The curve thinning algorithm and the shortest path algorithm in the power grid scenario are used to sample the nodes and lines in the power grid data, and representative data points are taken, including the inflection points, intersection points or contact points of the curve with other objects, to maintain the shape and outline of the line and delete redundant data points; the curvature distribution of the line is calculated, and thinning operations are performed in areas with lower curvature.

[0014] Moreover, the specific steps of obtaining the power dispatching model with optimized display details through grid cascade loading, visual culling, asynchronous loading and rendering for the power dispatching model retaining key features include:

[0015] First, the visual optimization of the power grid is carried out according to the voltage level, and highly distinguishable colors or effects are used for power grid lines of different voltage levels; secondly, a layered loading method is adopted when displaying the power grid model, and high-voltage lines are displayed first, and as the viewing angle decreases, low-voltage lines are gradually displayed;

[0016] Determine which objects are blocked or out of view based on the camera's position and viewing angle, and only render the model for the area the user is currently viewing;

[0017] Delay loading of data for undisplayed areas; start rendering other ready content while loading models in the background; use multi-core processors and multi-threading technology to process multiple rendering tasks in parallel.

[0018] Moreover, the specific steps of combining the intelligent preloading and prediction mechanism with the power dispatching model optimized for display details, applying the multi-level of detail technology and the progressive loading technology to realize the lightweight power dispatching model include:

[0019] Use machine learning algorithms to predict user behavior and field of view changes, and load models and texture resources that are about to enter the field of view in advance;

[0020] Treat models, textures, lighting and particles as different rendering layers, divide elements such as background, terrain and buildings into different hierarchical structures, and use different levels of detail models as the distance and field of view change;

[0021] The map data in the 3D engine is divided into tiles of different levels or resolutions. Each tile represents an area on the earth's surface. The tile level that needs to be displayed is calculated based on the user's perspective and distance, and the level of detail displayed is dynamically adjusted. Tiles at long distances are loaded at low resolution.

[0022] As you get closer to the surface or zoom in, higher resolution tile data is requested to replace the currently displayed low resolution tiles.

[0023] Moreover, the degree of simplification is dynamically determined according to the geometric characteristics and visual importance of the model. The geometric and topological analysis of the power dispatching model is performed to identify the texture details, edge contours and shapes of the model, and the contribution of each polygon to the visual effect of the model is evaluated. The polygons and texture details that have a significant impact on the overall appearance are retained first, and the specific steps for removing or simplifying other polygons are as follows:

[0024] First, the geometric distribution of the model is analyzed to identify key features such as edges, corners, textures, etc., and then an algorithm is used to determine which polygons are removed or merged.

[0025] A lightweight system for power dispatching model under the metaverse, comprising:

[0026] Power dispatch model intelligent data simplification module, power dispatch model resource management and dynamic dispatch module, power dispatch model refined resource allocation and detail management module.

[0027] Furthermore, the electric power dispatch model intelligent data reduction module is used to intelligently reduce the electric power dispatch model data;

[0028] The power dispatch model resource management and dynamic dispatch module is used to dynamically adjust and optimize the display details of the power dispatch model;

[0029] The power dispatching model refined resource allocation and detail management module is used to realize fast loading and rendering of the power dispatching model.

[0030] A computer storage medium storing one or more instructions, wherein when the instructions are executed by one or more computers, the one or more computers implement the method according to any one of claims 1 to 5.

[0031] Advantages and beneficial effects of the present invention:

[0032] The present invention proposes a lightweight method, system and storage medium for power dispatching model under the metaverse, performs lightweight processing on the three-dimensional model of the power dispatching metaverse, and realizes efficient loading and rendering of the three-dimensional model of the power grid through advanced data processing, resource scheduling technology and rendering optimization strategy, so as to achieve the goal of real-time and efficient rendering and integrated display of meteorological data, geographic information, power grid operation data and three-dimensional model of equipment status under ordinary office hardware conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the process of implementing the lightweight method of three-dimensional model in the power dispatching metaverse according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings:

[0035] The embodiment of the present invention is based on the early simulation training application scenario of the power dispatching metaverse. The power dispatching metaverse is a virtual digital power grid simulation space built based on digital technologies such as big data, artificial intelligence, the Internet of Things, and cloud computing. It can perform real-time analysis, control, and abnormal monitoring of the current power system, and simulate and evaluate the future power grid operation, thereby realizing the monitoring and full life cycle management of the power grid operation, thereby reducing the power grid operation and maintenance costs and providing highly reliable and high-quality power guarantee.

[0036] A lightweight method for power dispatching model in the metaverse, such as Figure 1 As shown, the following steps are included:

[0037] Step 1: Based on deep learning, adaptive surface reduction compression technology is combined with texture mapping and curve thinning algorithm to perform intelligent data simplification on the power dispatching model to obtain a power dispatching model that retains key features;

[0038] The specific steps of step 1 include:

[0039] (1) Adaptive face reduction and compression based on deep learning: The degree of simplification is dynamically determined based on the geometric characteristics and visual importance of the model. The model is subjected to geometric and topological analysis to identify key features and structures such as texture details, edge contours, and shapes, and to evaluate the contribution of each polygon to the visual effect of the model.

[0040] Use convolutional neural network (CNN) to extract the features of the 3D model. Feature extraction can be expressed as:

[0041] F=φ(M;θ)

[0042] Among them, M is the input 3D model, θ is the network parameter, φ is the forward propagation function of CNN, and F is the extracted feature.

[0043] Based on the extracted features, an importance weight is assigned to each part of the model and a fully connected layer is constructed:

[0044] w=σ(F·W+b)

[0045] Where W and b are the weights and biases of the fully connected layer, and σ is the activation function.

[0046] According to the importance weight and the predetermined compression rate c, the retention status of each vertex or polygon is adaptively determined:

[0047]

[0048] Among them, d i is the retained state of the i-th element, w i is its importance weight.

[0049] (2) Texture optimization: First, use the built-in texture compression tool in Cesium Lab to convert high-resolution textures into more efficient formats, such as JPEG or WebP, to reduce storage requirements and increase loading speed; reduce the size of texture maps through reasonable map cutting and optimization; use texture mapping technology to map two-dimensional textures to the surface of three-dimensional models, optimize the allocation of texture coordinates, make the texture coverage on the model surface more uniform, reduce visual distortion, and reduce the reuse of texture data; use a three-bilinear filtering algorithm to smooth texture transitions and reduce jagged effects.

[0050] (3) Line curve thinning: The curve thinning algorithm and the shortest path algorithm in the power grid scenario are used to sample the nodes and lines in the power grid data, and representative data points such as significant turning points are taken.

[0051] Use a corner detection algorithm to identify significant turning points in the line:

[0052]

[0053] If θ i If it exceeds a certain threshold, then p i is a corner point and needs to be retained. Redundant data points are deleted and only enough key points are retained to maintain the shape and contour of the line.

[0054] (4) Design a loss function to measure the difference between the simplified model and the original model and the visual quality loss, taking into account multiple factors to achieve the best simplification effect and performance. By adjusting the weight parameters, the model simplification algorithm is optimized for specific application requirements. For geometric error loss, the mean square error is used to compare the vertex position differences between the original model and the simplified model; for visual quality loss, the structural similarity (SSIM) index or peak signal-to-noise ratio (PSNR) is used to compare the appearance differences between the original and simplified models under different observation conditions; for topological consistency loss, the boundaries or connectivity of the original and simplified models are compared to ensure that the simplified model maintains the topological structure of the original model; for detail preservation loss, the frequency spectrum of the model is analyzed to measure the amount of high-frequency details retained in the simplified model.

[0055] Step 2: Implement model resource management and dynamic scheduling through cascade loading, visual culling, asynchronous loading and rendering;

[0056] (1) Grid cascade loading: First, based on the voltage level hierarchy, a voltage level function V(k) is defined, where k represents a node or line in the grid and V(k) represents its corresponding voltage level. i represents the voltage value of node or line i, w i is a weight factor used to reflect its importance in the power grid.

[0057]

[0058] V(k) is used to guide the visual optimization algorithm to ensure that nodes or lines with higher voltage levels are given priority during the rendering of the power grid model.

[0059] Next, considering the relationship between voltage level and ground height, the viewing angle height function H(h) is introduced to dynamically adjust the display details of the power grid model according to the observer's viewing angle height. Each tile represents an area on the earth's surface and is divided into tiles of different levels (or resolutions). The tile level that needs to be displayed is calculated based on the user's viewing angle and distance. Tiles at close range are loaded with high resolution, and tiles at a distance are loaded with low resolution. As the user's viewing angle changes, the system dynamically loads and replaces tiles to gradually improve the display effect. When the user approaches the surface or zooms in on the view, the system requests higher resolution tile data to replace the currently displayed low-resolution tiles.

[0060]

[0061] Among them, γ is a parameter that controls the attenuation rate of the function, and h0 is the reference height used to determine the starting point of the viewing angle height.

[0062] (2) Visual culling: Determine which objects are blocked or out of view based on the camera’s position and viewing angle, and only render the model of the area the user is currently viewing. First, define the camera’s position C and direction, as well as viewing angle parameters, including the horizontal field of view angle θ. x and vertical field angle θ y Next, in order to quickly determine whether an object is within the viewing frustum, the bounding volume B is used to approximate the spatial range of the object. For each object's bounding volume B i , it is necessary to test whether it intersects with each face of the viewing cone. If the bounding volume is completely on one side of a plane, the object can be culled. Finally, the projection matrix P and the view matrix V are used to transform the object in the world coordinate system to the camera coordinate system, and further projected onto the two-dimensional viewing plane. For objects within the viewing cone, the object depth value and the camera near-surface depth value are further used to cull those parts that are blocked by other objects.

[0063] (3) Asynchronous loading and rendering: First, the scene is divided into multiple areas or blocks. Each area only loads the necessary data, and the scene graph is used to manage these areas and their visibility status. The data of the undisplayed area is loaded lazily to reduce the amount of data during initial loading. Next, the task queue and worker threads are used to process the loading of resources in parallel. While the model is loading in the background, other ready content is started to be rendered. Multi-core processors and multi-threading technology are used to process multiple rendering tasks in parallel.

[0064] Through grid cascade loading and visual culling, the display details of the grid model are dynamically adjusted according to the observer's perspective, and asynchronous loading and rendering are used to achieve model resource management and dynamic scheduling. Combined with intelligent preloading and prediction mechanisms, multi-level of detail technology and progressive loading technology are used to achieve fast loading and rendering of the model, and refine resource allocation and detail management.

[0065] Step 3: Combine intelligent preloading and prediction mechanisms, apply multi-level-of-detail technology and progressive loading technology to achieve fast model loading and rendering, and refine resource allocation and detail management;

[0066] The specific steps of step 3 include:

[0067] (1) Use machine learning algorithms to predict user behavior and field of view changes, preload models and texture resources that are about to enter the field of view, and reduce rendering delays. Define a user behavior feature vector, which contains relevant information about user behavior, such as historical field of view position, movement speed, acceleration, etc., where is the th feature. Use a machine learning model to predict the future field of view position, and determine the set of resources that need to be preloaded based on the predicted field of view area. Dynamically adjust the resolution of texture resources based on the predicted field of view and distance.

[0068] (2) Treat models, textures, lighting, and particles as different rendering levels, divide elements such as background, terrain, and buildings into different hierarchical structures, use different levels of detail models as the distance and field of view change, and reduce the number of polygons and texture resolution;

[0069] (3) The map data in the 3D engine is divided into tiles of different levels (or resolutions). Each tile represents an area on the earth's surface. The tile level that needs to be displayed is calculated based on the user's viewing angle and distance, and the level of detail displayed is dynamically adjusted. Tiles at long distances are loaded at low resolution.

[0070] (4) When getting close to the surface or zooming in on the view, higher-resolution tile data is requested to replace the currently displayed low-resolution tiles, gradually improving the display effect.

[0071] A lightweight system for power dispatching model under the metaverse, comprising:

[0072] Power dispatch model intelligent data simplification module, power dispatch model resource management and dynamic dispatch module, power dispatch model refined resource allocation and detail management module.

[0073] The electric power dispatch model intelligent data reduction module is used to intelligently reduce the electric power dispatch model data;

[0074] The power dispatch model resource management and dynamic dispatch module is used to dynamically adjust and optimize the display details of the power dispatch model;

[0075] The power dispatching model refined resource allocation and detail management module is used to realize fast loading and rendering of the power dispatching model.

[0076] An embodiment of the present invention further provides a computer storage medium, wherein the computer storage medium stores a computer program, and the computer program implements the method when executed by the processor.

[0077] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0078] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0079] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0080] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

Claims

1. A lightweight method for power dispatching model in the metaverse, characterized by: The following steps are involved: Adaptive surface reduction compression technology based on deep learning, combined with texture mapping and curve thinning algorithms, intelligently simplifies the data of the power dispatch model to obtain a power dispatch model that retains key features; For the power dispatching model retaining key features, a power dispatching model with optimized display details is obtained through grid cascade loading, visual culling, asynchronous loading and rendering; The power dispatching model with optimized display details is combined with intelligent preloading and prediction mechanism, and multi-level-of-detail technology and progressive loading technology are applied to realize the lightweight of the power dispatching model.

2. The method for lightweighting the power dispatching model under the metaverse according to claim 1 is characterized in that: The adaptive surface reduction compression technology based on deep learning, combined with texture mapping and curve thinning algorithm, performs intelligent data simplification on the power dispatching model, and the specific steps of obtaining the power dispatching model that retains key features include: Dynamically determine the degree of simplification based on the geometric characteristics and visual importance of the model, perform geometric and topological analysis on the power dispatch model, identify the texture details, edge contours and shapes of the model, evaluate the contribution of each polygon to the visual effect of the model, retain polygons and texture details that have a significant impact on the overall appearance, and remove or simplify other polygons; Convert high-resolution texture maps into a CPU-based compression format, and after reasonable texture cutting and optimization, use texture mapping technology to map the two-dimensional texture to the surface of the three-dimensional model; The curve thinning algorithm and the shortest path algorithm in the power grid scenario are used to sample the nodes and lines in the power grid data, and representative data points are taken, including the inflection points, intersection points or contact points of the curve with other objects, to maintain the shape and outline of the line and delete redundant data points; the curvature distribution of the line is calculated, and thinning operations are performed in areas with lower curvature.

3. The method for lightweighting the power dispatching model under the metaverse according to claim 1 is characterized in that: The specific steps of obtaining the power dispatching model with optimized display details through grid cascade loading, visual culling, asynchronous loading and rendering for the power dispatching model retaining key features include: First, the visual optimization of the power grid is carried out according to the voltage level, and highly distinguishable colors or effects are used for power grid lines of different voltage levels; secondly, a layered loading method is adopted when displaying the power grid model, and high-voltage lines are displayed first, and as the viewing angle decreases, low-voltage lines are gradually displayed; Determine which objects are blocked or out of view based on the camera's position and viewing angle, and only render the model for the area the user is currently viewing; Delay loading of data for undisplayed areas; start rendering other ready content while loading models in the background; use multi-core processors and multi-threading technology to process multiple rendering tasks in parallel.

4. The method for lightweighting the power dispatching model under the metaverse according to claim 1 is characterized in that: The specific steps of combining the intelligent preloading and prediction mechanism with the power dispatching model optimized for display details, applying the multi-level of detail technology and the progressive loading technology to realize the lightweight power dispatching model include: Use machine learning algorithms to predict user behavior and field of view changes, and load models and texture resources that are about to enter the field of view in advance; Treat models, textures, lighting and particles as different rendering layers, divide elements such as background, terrain and buildings into different hierarchical structures, and use different levels of detail models as the distance and field of view change; The map data in the 3D engine is divided into tiles of different levels or resolutions. Each tile represents an area on the earth's surface. The tile level that needs to be displayed is calculated based on the user's perspective and distance, and the level of detail displayed is dynamically adjusted. Tiles at long distances are loaded at low resolution. When you get close to the surface or zoom in, request higher resolution tile data to replace the currently displayed low resolution tiles.

5. The method for lightweighting the power dispatching model under the metaverse according to claim 2 is characterized in that: The degree of simplification is dynamically determined based on the geometric characteristics and visual importance of the model. The geometric and topological analysis of the power dispatching model is performed to identify the texture details, edge contours and shapes of the model, and the contribution of each polygon to the visual effect of the model is evaluated. The polygons and texture details that have a significant impact on the overall appearance are retained first, and the specific steps for removing or simplifying other polygons are as follows: First, the geometric distribution of the model is analyzed to identify key features such as edges, corners, textures, etc., and then an algorithm is used to determine which polygons are removed or merged.

6. A lightweight system for power dispatching model under the metaverse, characterized by: include: Intelligent data reduction module for power dispatch model, resource management and dynamic dispatch module for power dispatch model, and refined resource allocation and detail management module for power dispatch model; The electric power dispatch model intelligent data reduction module is used to intelligently reduce the electric power dispatch model data; The power dispatch model resource management and dynamic dispatch module is used to dynamically adjust and optimize the display details of the power dispatch model; The power dispatching model refined resource allocation and detail management module is used to realize fast loading and rendering of the power dispatching model.

7. A computer storage medium, characterized in that: The computer storage medium stores one or more instructions, which, when executed by one or more computers, enable the one or more computers to implement the method of any one of claims 1 to 5.

Images