Detail level dynamic scheduling method and system for real-time rendering of three-dimensional scene of network distribution

Abstract

The application relates to the technical field of three-dimensional visualization rendering, in particular to a distribution network three-dimensional scene real-time rendering detail level dynamic scheduling method and system; the method comprises the following steps: based on device electrical topology relations and spatial distribution, spatial partition is carried out according to functional units, and spatial boundaries and in-zone device indexes are established; a main device set and an auxiliary device set are divided according to node levels in the partition, and a partition-level rendering priority sequence is formed; rendering load is estimated according to a viewpoint position, a view cone range and a total model face sheet amount, and load margins of each visible partition are obtained; the detail level is dynamically scheduled according to the load margins and the partition-level rendering priority sequence, and a real-time rendering result is output. The technical scheme can keep the main device recognizable display and reduce invalid rendering load under the frame time budget constraint.

Description

Technical Field

[0001] This application relates to the field of 3D visualization rendering technology, specifically a method and system for dynamic scheduling of detail levels in real-time rendering of 3D scenes in power distribution networks. Background Technology

[0002] 3D distribution network scenes have been used for substation inspection, feeder display, transformer area equipment management, and distribution network digital twin operation and maintenance. These scenes typically include objects such as transformers, circuit breakers, disconnect switches, busbar sections, instrument transformers, surge arresters, cabinet shells, brackets, bases, covers, and signs. Maintenance personnel will frequently change their viewpoint and observation direction while browsing.

[0003] In large-scale 3D distribution network scenarios, when maintenance personnel switch from the overall feeder view to the view of switchgear bays, pole-mounted transformer areas, or the interior of ring main units, core electrical equipment, auxiliary electrical equipment, and non-electrical structural components will appear simultaneously in the same frame. Current real-time rendering processes are prone to display imbalances, where insufficient detail in critical electrical equipment coexists with excessive detail in auxiliary structural components. This makes it difficult to maintain stable identification of the types and operating status of core objects such as circuit breakers, busbar sections, transformers, and disconnect switches under target frame rate constraints.

[0004] In view of this, this application proposes a method and system for dynamic scheduling of detail levels in real-time rendering of three-dimensional scenes of power distribution networks. Summary of the Invention

[0005] To achieve the above objectives, this application provides a method and system for dynamic scheduling of detail levels in real-time rendering of 3D scenes of power distribution networks. The specific technical solution is as follows:

[0006] Firstly, this application provides a method for dynamic scheduling of detail levels in real-time rendering of 3D scenes of power distribution networks, including:

[0007] Based on the electrical topology and spatial distribution of each device model in the 3D scene of the power distribution network, the 3D scene is spatially partitioned according to the functional units in the electrical topology, and the spatial boundaries of each partition and the device index within the partition are established.

[0008] Within each spatial partition, the device model is divided into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology. Rendering priorities are assigned to each device according to the node level to construct a partition-level rendering priority sequence.

[0009] The set of visible partitions is determined based on the current viewpoint position and the range of the view frustum. The rendering load is estimated by combining the total number of model patches in each visible partition. The rendering load is compared with the frame time budget to obtain the load margin of each visible partition.

[0010] Based on the load margin and partition-level rendering priority sequence, the detail level of the main device set is kept no lower than a preset threshold in each visible partition, and the model simplification and replacement are performed on the auxiliary device set in order of rendering priority from low to high to complete the real-time rendering output.

[0011] Preferably, the spatial partitioning includes: reading the electrical topology data of the distribution network, performing a depth-first traversal along the electrical topology path, and determining the boundary nodes according to the functional unit boundary rules; grouping the equipment nodes between two adjacent boundary nodes into the same functional unit group, and grouping the topology end equipment nodes into the functional unit group where the nearest boundary node is located; in the ring network structure, using the node where the ring network interconnection switch is located as the boundary node; forming spatial boundaries according to the vertex coordinate range of the equipment models in the coordinate direction within the functional unit group, and generating an intra-zone equipment index that records the unique identifier of the equipment model, the equipment type code, and the reference of the grid resources.

[0012] Preferably, the main device set and the auxiliary device set include: querying node information in the electrical topology for device models within a partition; assigning device models on the main link of the electrical topology to the main device set; assigning auxiliary electrical devices on branch links and non-electrical structural components without node identity to the auxiliary device set; each device model in the same partition belongs to only one set.

[0013] Preferably, the rendering priority includes: assigning a value according to the rule that the higher the electrical topology node level, the greater the rendering priority value; the rendering priority value is jointly determined by the normalized device category weight score and the node level score, and the device category weight adjustment coefficient and the node level score adjustment coefficient are used to determine the proportion; after mapping processing, the rendering priority value of the auxiliary electrical equipment in the auxiliary equipment set is lower than the lowest rendering priority value in the main equipment set and higher than the rendering priority value of non-electrical structural components.

[0014] Preferably, the partition-level rendering priority sequence includes: arranging all device models within a partition in descending order of rendering priority value to form an ordered array; the elements of the ordered array record the unique identifier of the device model, the set tag to which it belongs, and the rendering priority value; the set tag to which it belongs is used to distinguish between the main device set and the auxiliary device set; the partition-level rendering priority sequence is regenerated when the electrical topology changes.

[0015] Preferably, the set of visible partitions includes: determining the current viewpoint position based on the world coordinates of the virtual camera, and determining the range of the view frustum based on the vertical field of view of the virtual camera, the aspect ratio of the image, the distance to the near clipping plane, and the distance to the far clipping plane; performing an intersection test between the spatial boundaries of each partition and the view frustum; marking partitions that are not completely excluded by the clipping plane of the view frustum as visible partitions and adding them to the set of visible partitions.

[0016] Preferably, the total number of model faces includes: pre-generating multiple detail level versions for each device model and storing the number of faces for each detail level version in the scene database; reading the number of faces of all device models in the visible partition at the current detail level; and summing the number of faces of all device models in the visible partition to obtain the total number of model faces in the visible partition.

[0017] Preferably, the rendering load includes: rendering a standard test scene with a known number of faces on the target graphics hardware and recording the rendering time to obtain the average rendering time per face; generating an estimated rendering time for the visible partition based on the total number of model faces in the visible partition, the average rendering time per face, and the fixed overhead of the rendering pipeline; and using the estimated rendering time as the rendering load for the visible partition.

[0018] Preferably, the load margin includes: comparing the rendering load of all visible partitions with the frame time budget to generate a global load margin; allocating the global load margin according to the proportion of the estimated rendering time of each visible partition in the estimated rendering time of all visible partitions; and using the allocation result as the load margin of each visible partition.

[0019] Preferably, the preset threshold includes: power distribution maintenance technicians conducting identification tests on the rendered images of the main equipment model at each level of detail; recording the identification accuracy of equipment type and operating status at each level of detail; and determining the level before the level of detail whose identification accuracy first falls below the preset identification accuracy standard as the preset threshold for the equipment model in the main equipment set.

[0020] Preferably, the model simplification replacement includes: when the partition load margin is negative, reducing the level of detail starting from the device model with the lowest rendering priority in the auxiliary device set; if the partition load margin is still negative after the same device model is reduced to the lowest level of detail version, the partition-level rendering priority sequence is switched to the next device model with a higher rendering priority; if the partition load margin is still negative after all auxiliary devices are reduced to the lowest level of detail version, the level of detail is reduced for the main device set, and the level of detail of the device models in the main device set is not lower than a preset threshold.

[0021] Secondly, this application also provides a detailed hierarchical dynamic scheduling system for real-time rendering of distribution network three-dimensional scenes, which is used to implement the detailed hierarchical dynamic scheduling method for real-time rendering of distribution network three-dimensional scenes, including: a topology partitioning module, a priority construction module, a load estimation module, and a hierarchical scheduling module.

[0022] The topology partitioning module is used to spatially partition the three-dimensional scene based on the electrical topology relationship and spatial distribution of each device model in the distribution network three-dimensional scene, using the functional units in the electrical topology as the dividing criteria, and to establish the spatial boundaries of each partition and the device index within the partition.

[0023] The priority construction module is used to divide the device model into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology within each spatial partition, assign rendering priority to each device according to the node level, and construct a partition-level rendering priority sequence.

[0024] The load estimation module is used to determine the set of visible partitions based on the current viewpoint position and the range of the view frustum, estimate the rendering load by combining the total number of model patches in each visible partition, and compare the rendering load with the frame time budget to obtain the load margin of each visible partition.

[0025] The hierarchical scheduling module is used to maintain the detail level of the main device set at no less than a preset threshold in each visible partition according to the load margin and the partition-level rendering priority sequence, and to perform model simplification and replacement on the auxiliary device set in order of rendering priority from low to high to complete real-time rendering output.

[0026] The beneficial effects of this application are:

[0027] This application achieves a reduction in model traversal scale when performing rendering scheduling at the partition level by dividing the 3D scene into spatial partitions based on functional units in the electrical topology and establishing partition spatial boundaries and device indexes within the partitions, while maintaining the coordination and consistency of devices within the functional units in adjusting details.

[0028] This application divides devices into a set of primary devices and a set of auxiliary devices based on the electrical topology node hierarchy and constructs a partition-level rendering priority sequence. This integrates the semantic information of the importance of electrical functions into the rendering scheduling priority system, enabling differentiated protection of device detail levels in an orderly manner according to the importance of electrical functions when rendering resources are limited.

[0029] This application achieves quantitative evaluation and balanced distribution of the rendering load status of each visible partition by performing view frustum intersection tests at the partition level and estimating the rendering load of each partition based on the total number of faces and the rendering time per unit face. The global load margin is then distributed according to the estimated rendering time ratio of each partition, thereby realizing the quantitative evaluation and balanced distribution of the rendering load status of each visible partition.

[0030] This application achieves reliable assurance of display quality of core electrical equipment and maintenance of visual continuity of the screen under frame time budget constraints by performing model simplification and replacement of auxiliary equipment sets in order of rendering priority from low to high according to partition load margin, while keeping the detail level of the main equipment set no lower than a preset threshold, and suppressing frame-by-frame switching by combining a hysteresis judgment mechanism.

[0031] When the overall technical solution of this application is applied to the 3D operation and maintenance inspection system of the power distribution network, it can stably maintain the rendering frame rate near the target frame rate during viewpoint movement and scene content changes. At the same time, it ensures that the core electrical equipment responsible for power transmission and on / off control always maintains a display quality that can be identified by operation and maintenance personnel, providing real-time rendering guarantee for the 3D operation and maintenance inspection of the power distribution network that takes into account both interactive smoothness and the reliability of key equipment identification. Attached Figure Description

[0032] Figure 1 Flowchart of the method for dynamic scheduling of detail levels in real-time rendering of the 3D scene of the power distribution network provided in this application;

[0033] Figure 2 This application provides a schematic diagram showing the correspondence between the functional unit equipment composition and spatial partitioning.

[0034] Figure 3 This is a schematic diagram illustrating the classification of electrical topology node levels and device sets provided in this application;

[0035] Figure 4 A schematic diagram of the intersection test between the view frustum and the partition bounding box provided in this application;

[0036] Figure 5 A comparative diagram of the meshes of the device model with multiple levels of detail provided in this application;

[0037] Figure 6 The structural diagram of the dynamic scheduling system for real-time rendering of the distribution network 3D scene provided in this application. Detailed Implementation

[0038] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0040] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0041] Example 1

[0042] Reference Figures 1 to 5This is the first embodiment of the present application, such as Figure 1 As shown, a detailed level dynamic scheduling method for real-time rendering of 3D scenes in power distribution networks is provided.

[0043] In a 3D scene of a power distribution network, there are typically multiple functional units, such as substations, switching stations, ring main units, prefabricated substations, and pole-mounted transformer substations. The electrical equipment and non-electrical structural components within each functional unit are stored as meshes in the scene database of the rendering engine. With the expansion of the power distribution network's coverage and the improvement of 3D modeling accuracy, the number of equipment models and the size of single-model patches in the scene continue to grow, correspondingly increasing the geometric processing load and rendering scheduling load faced by the rendering engine when generating images frame by frame. This application achieves adaptive control of the detail level of the 3D scene of the power distribution network under frame time budget constraints through spatial partitioning based on electrical topology relationships, rendering priority construction driven by topology semantics, rendering load estimation based on view frustum association, and dynamic scheduling of detail levels driven by load margin.

[0044] Step S1: Based on the electrical topology and spatial distribution of each device model in the 3D scene of the power distribution network, the 3D scene is spatially partitioned according to the functional units in the electrical topology, and the spatial boundaries of each partition and the device index within the partition are established.

[0045] After constructing the 3D scene of the power distribution network, all equipment models in the scene are stored in the form of meshes. Each mesh contains vertex coordinate data, face index data, and material attribute information. The equipment model types in the scene include transformers, circuit breakers, disconnect switches, busbar segments, surge arresters, current transformers, voltage transformers, cable terminations, drop-out fuses, equipment mounting brackets, base foundations, cabinet exterior panels, cable trench covers, and signage. When the total number of equipment models in the scene is large, the rendering engine needs to traverse all model objects when performing view frustum clipping and rendering scheduling each frame. An excessively large traversal range will lead to unnecessary scheduling computation overhead. To reduce the scale of model traversal during frame-by-frame scheduling and provide partitioned management units for subsequent rendering scheduling, this step spatially partitions the 3D scene based on the electrical topology relationships and spatial distribution of each equipment model in the 3D scene of the power distribution network, using functional units in the electrical topology as the dividing criteria. The electrical topology relationships are used to determine the equipment composition of the functional units, and the spatial distribution is used to determine the spatial boundary range of the corresponding partitions for each functional unit.

[0046] Establishing spatial partitioning first requires reading the electrical topology data of the distribution network. This data records the electrical connections between various devices in the scenario, including feeder connections, switch connections, busbar connections, high and low voltage transformer side connections, and grounding connections. In the electrical topology, a functional unit refers to a combination of devices that performs independent operational or protective functions during electrical operation. For example, in a 10kV switchgear bay, a functional unit consists of an incoming disconnect switch, incoming circuit breaker, busbar section, outgoing circuit breaker, outgoing disconnect switch, and their associated current transformers and surge arresters. Similarly, in a pole-mounted transformer substation, a functional unit consists of a pole-mounted transformer, drop-out fuse, surge arrester, and mounting bracket. The delineation of functional units references the operating units and protection coordination sections in the distribution network operation regulations. Devices within the same functional unit have close coordination in operation and protection actions.

[0047] Figure 2 The equipment composition and corresponding spatial partition boundaries within a 10kV switchgear bay functional unit are shown. Figure 2 Solid rectangles represent various electrical equipment, including the top horizontal busbar section, the left-side incoming disconnect switch and incoming circuit breaker, the right-side outgoing circuit breaker and outgoing disconnect switch, and the current transformer and surge arrester in the lower middle section. Dashed rectangles represent non-electrical structural components such as mounting brackets, cabinet outer panels, base foundations, and signage. The entire functional unit is enclosed by an axis-aligned bounding box represented by a long dashed rectangle, which serves as the spatial partition boundary defined based on electrical topology. Figure 2 The document also indicates the representation methods for three types of graphic elements: electrical equipment, non-electrical structural components, and partition boundaries.

[0048] During functional unit identification, a depth-first traversal is performed along the electrical topology path starting from the feeder exit node. Boundary nodes on the traversal path are determined according to preset functional unit boundary rules. These rules are based on operational and protection boundaries in the distribution network. Boundary nodes are typically located at the connection points between feeder sections and substation equipment, i.e., the boundary node where one side of the feeder section is the line section and the other side is the substation equipment section. Each boundary node is assigned to the functional unit group upstream of its power source. During the traversal, all equipment nodes between two adjacent boundary nodes are grouped into the same functional unit group. For equipment nodes at the topology's end that do not form a closed loop, the topology path is traced upwards to the nearest boundary node, and the end-point equipment node is assigned to the functional unit group containing that boundary node. When a ring network structure appears in the topology, the node containing the ring network tie switch is used as the boundary node. The tie switch itself is assigned to the functional unit group on its power supply side, and the remaining equipment on both sides of the tie switch is assigned to its respective functional unit group.

[0049] After completing the functional unit grouping, the spatial boundaries of each partition are established based on the spatial distribution of all equipment models within each group. The vertex coordinates of each equipment model mesh within a group are extracted. The minimum and maximum values ​​of all vertex coordinates are taken in the X, Y, and Z axes, respectively. The minimum value in each direction forms the minimum corner point, and the maximum value forms the maximum corner point. The minimum and maximum corner points determine the axis-aligned bounding box of the partition, which serves as the spatial boundary of the partition. To avoid the partition bounding box fitting too tightly to the outer contour of the equipment models within the partition, leading to frequent overlap between adjacent partition bounding boxes in densely populated equipment areas, an expansion margin is applied to both the positive and negative sides of the bounding box in the X, Y, and Z axes. This is achieved by subtracting the minimum corner point from the minimum value in each axis coordinate. Add the maximum corner point to each axis coordinate. Extended margin is defined as follows: ,in This refers to the expansion margin in each axial direction. This represents the maximum side length of the bounding box in the X, Y, and Z axes. This is the expansion factor. The value of the expansion factor needs to balance the compactness of the bounding box with the sufficiency of the spacing between adjacent zones. For example, in a substation scenario, when the minimum net spacing between functional units is approximately 0.15 times the maximum span of the equipment, the expansion factor is... Setting it to 0.05 ensures that the bounding boxes of adjacent partitions do not overlap under most typical layout conditions.

[0050] After establishing the spatial boundaries of each partition, an intra-partition device index is constructed for each partition. The intra-partition device index is stored in an array structure, where each element records the unique identifier of the device model, the device type code, and the mesh resource reference of the device model in the rendering engine's scene database. The device type code is pre-defined according to the power distribution network equipment classification standard, covering all aforementioned electrical equipment types and non-electrical structural component types. Through the intra-partition device index, when the rendering engine needs to perform a rendering scheduling operation on a specific partition, it can directly read the index array to obtain the mesh resource references of all device models within that partition, without needing to traverse the entire scene's global model list, thus reducing the retrieval time for a single scheduling operation.

[0051] Spatial partitioning is based on electrical topology functional units, ensuring that devices within each partition are electrically related, and devices within the same functional unit are managed as a whole within the same partition. When rendering scheduling is performed on a partition-by-partition basis, devices within a functional unit can participate in detail-level evaluation and adjustment as a whole, which helps maintain visual consistency in detail-level transitions between devices within a functional unit and avoids excessive scheduling computation overhead caused by per-model traversal.

[0052] Step S2: Within each spatial partition, the device model is divided into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology. Rendering priorities are assigned to each device according to the node level to construct a partition-level rendering priority sequence.

[0053] After completing the spatial partitioning, it is necessary to prioritize the rendering of equipment models within each partition to provide differentiated levels of detail for different devices when rendering resources are limited. In the operation and maintenance of a 3D power distribution network scenario, operators pay varying degrees of visual attention to different types of equipment. When inspecting a switchgear bay, operators pay close attention to equipment directly related to judging the electrical operating status, such as the circuit breaker body, the position of the disconnector's moving contacts, and the busbar connections, while paying less attention to auxiliary structural components such as the cabinet's outer panel, mounting brackets, bolt assemblies, and identification signs.

[0054] The node level of equipment in the electrical topology reflects its position within the functional links of the distribution network, and the node level corresponds to the level of attention required for operation and maintenance. In the distribution network electrical topology, equipment on the backbone link undertakes power transmission and on / off control functions, including transformers, circuit breakers, load switches, and busbar sections; these devices occupy upper-level nodes in the electrical topology. Auxiliary electrical equipment connected to the backbone link nodes, such as voltage transformers, current transformers, surge arresters, grounding switches, and drop-out fuses, perform measurement, protection, or grounding functions and occupy lower-level nodes in the electrical topology. Non-electrical structural components that do not participate in electrical circuits but exist as geometric components in the three-dimensional scene, such as mounting brackets, base foundations, cabinet outer panels, covers, and signs, do not have node status in the electrical topology and are at the lowest level.

[0055] During implementation, the node information of each device within a partition is queried in the electrical topology. Based on the device's node level in the electrical topology, the device model is divided into a main device set and an auxiliary device set. Devices with node status in the electrical topology are categorized according to their link location, while non-electrical structural components without node status are directly assigned to the auxiliary device set. Device models on the main links of the electrical topology are assigned to the main device set, while auxiliary electrical devices and all non-electrical structural components on branch links are assigned to the auxiliary device set. Device models in the main device set have priority during detail-level scheduling, while device models in the auxiliary device set are allowed to have their detail level reduced when rendering resources are insufficient. Within the same partition, there is no overlap between the main device set and the auxiliary device set; each device model belongs to only one set.

[0056] Figure 3Using the electrical topology within a functional unit as an example, this diagram illustrates the node hierarchy of devices within the topology and their corresponding set classifications. The vertical axis, from top to bottom, is labeled from level 5 to level 1. Solid circle nodes represent devices on the main trunk link, including incoming disconnect switches, circuit breakers, busbar sections, outgoing circuit breakers, and outgoing disconnect switches, all classified as main equipment. Hollow circle nodes represent auxiliary electrical equipment connected to the main trunk nodes, including current transformers, surge arresters, and voltage transformers. Square markers at the bottom indicate non-electrical structural components without topology node status, with a level score of zero. The main trunk link is traversed by thick connecting lines, while branch devices are connected to their corresponding main trunk nodes by thin connecting lines. Figure 3 It expresses the rule that the node hierarchy decreases from the power supply side to the end, as well as the boundary logic between main equipment and auxiliary equipment.

[0057] After the set is partitioned, rendering priorities are assigned to each device according to node level. The rendering priority values ​​follow the rule that higher-level electrical topology nodes have higher rendering priority values. For device models in the main device set, the rendering priority value is determined based on the device's position in the main topology link; main devices closer to the power supply side have higher rendering priority than those closer to the feeder end. For device models in the auxiliary device set, after calculating the rendering priority value according to the formula, it is checked whether there are any auxiliary device models whose rendering priority value is not lower than the lowest rendering priority value in the main device set. If so, the rendering priority values ​​of all auxiliary electrical devices in the auxiliary device set are scaled proportionally. The highest rendering priority value among auxiliary electrical devices is mapped to the lowest rendering priority value in the main device set minus 0.01, and the lowest rendering priority value among auxiliary electrical devices is mapped to the highest rendering priority value among non-electrical structural components plus 0.01. The relative order of auxiliary electrical devices within the auxiliary device set remains unchanged, thus ensuring that the rendering priority value of auxiliary electrical devices in the auxiliary device set is strictly lower than the device model with the lowest rendering priority in the main device set and strictly higher than non-electrical structural components. Non-electrical structural components have the lowest rendering priority among all devices in the partition because their node level score is 0 and their equipment category weight score is the lowest among all equipment types.

[0058] The rendering priority value is represented as ,in For the first in the partition The rendering priority value of each device model. For the first The device category weight score for each device model. For the first The node-level score of each device model in the electrical topology. This is an adjustment factor for the weight of the equipment category. This is an adjustment coefficient for the node-level score. Equipment category weights are pre-set according to equipment type; the weight of main equipment such as transformers and circuit breakers is higher than that of auxiliary electrical equipment such as instrument transformers and surge arresters, and the weight of auxiliary electrical equipment is higher than that of non-electrical structural components such as supports and panels. The node-level score is obtained by hierarchically numbering the electrical topology, assigning the highest number to the feeder exit node and decreasing layer by layer along the topology path towards the end. Adjustment coefficient and This is used to balance the impact of device category differences and topology location differences on the rendering priority value; the sum of the two is 1.

[0059] To ensure comparability of device category weight scores and node level scores on a numerical scale, both are normalized to the range of 0 to 1 before being used in rendering priority calculations. The normalization method for device category weight scores is as follows: An initial category weight score is pre-defined for each device type; the highest initial category weight score among all device types is used as the normalization benchmark. The normalization values ​​for each device model are then... This is equal to the original category weight score of the equipment type divided by the normalized baseline value. For example, the original category weight score is set to 10 for transformers and circuit breakers, 9 for load switches, 8 for busbar sections, 6 for current transformers and voltage transformers, 5 for surge arresters and grounding switches, 4 for drop-out fuses, 2 for mounting brackets and bases, and 1 for cabinet exterior panels, covers, and identification plates; using the highest original category weight score of 10 as the normalized baseline value, the transformer's... The value is 1.0 for the surge arrester. The value is 0.5, for the sign. The value is 0.1. The normalization method for node level scores is as follows: assign the highest level number value to the feeder exit node, and decrease it layer by layer along the topology path towards the end to 1. Use the highest level number value as the normalization benchmark value. This is equal to the level number of the node where the device is located divided by the normalized reference value. For example, in a functional unit partition containing 5 topology levels, the level number of the device at the feeder exit node is 5, and the level number of the terminal device is 1. The normalized reference value for the device at the feeder exit node is... Version 1.0, end device The score is 0.2. For non-electrical structural components that do not have node status in the electrical topology, since they do not participate in the electrical circuit, the node level score is 0.2. The value is assigned to 0.

[0060] For example, when the impact of differences in device category on the level of attention paid to operation and maintenance is greater than the impact of differences in topology location, Take 0.6, Setting it to 0.4 makes the device category have a higher weight in the rendering priority calculation than the topology position.

[0061] All device models within a partition are sorted in descending order of rendering priority value to construct a partition-level rendering priority sequence. This sequence is stored as an ordered array, where each element contains a unique identifier for the device model, a set identifier, and a rendering priority value. The set identifier distinguishes whether a device model belongs to the primary or secondary device set. The partition-level rendering priority sequence remains fixed after generation during partition creation and does not require recalculation during subsequent frame-by-frame scheduling. It is only regenerated when the electrical topology changes.

[0062] The rendering priority division process integrates the semantic information of the electrical topology of the power distribution network into the priority system of rendering scheduling. This allows the detail level to be adjusted in an orderly manner from low-priority devices according to the importance of electrical functions when rendering resources are limited. The backbone devices that undertake the functions of power transmission and on / off control are always at the forefront in the partition-level rendering priority sequence, thus ensuring the rendering quality of the devices that operation and maintenance personnel focus on in terms of visual effects.

[0063] Step S3: Determine the set of visible partitions based on the current viewpoint position and the range of the view frustum. Estimate the rendering load by combining the total number of model patches in each visible partition. Compare the rendering load with the frame time budget to obtain the load margin of each visible partition.

[0064] After the partition-level rendering priority sequence for each partition is constructed, the rendering engine needs to determine the currently visible partition set and estimate the rendering load during the preparation phase of each rendering frame. Real-time rendering of the 3D scene is performed frame-by-frame by the rendering engine, and the generation of each frame must be completed within the frame time budget. The frame time budget is determined by the target frame rate. For example, when the target frame rate is set to 30 frames per second, the frame time budget is 33.3 milliseconds; when the target frame rate is set to 60 frames per second, the frame time budget is 16.7 milliseconds. 3D operation and maintenance inspection scenarios for power distribution networks typically require a target frame rate of no less than 30 frames per second to ensure smooth interaction during viewpoint translation and rotation operations.

[0065] The current viewpoint position is determined by the world coordinates of the virtual camera in the 3D scene. The view frustum is defined by the virtual camera's vertical field of view, aspect ratio, near clipping plane distance, and far clipping plane distance. The view frustum forms a frustum region in 3D space, bounded by six planes: the top clipping plane, bottom clipping plane, left clipping plane, right clipping plane, near clipping plane, and far clipping plane. Objects located inside or intersecting with the frustum region are potentially visible objects; objects outside the frustum region do not need to be rendered in the current frame.

[0066] To determine the set of visible partitions, the axis-aligned bounding boxes of each partition established in step S1 are intersected with the view frustum based on the current viewpoint position and the range of the view frustum. The intersection test uses a separation axis determination method, checking each of the six clipping planes of the view frustum to determine whether the axis-aligned bounding boxes of the partitions are completely outside the clipping planes. Specifically, the determination method is as follows: the signed distance values ​​of the eight vertices of the bounding box relative to the normal vectors of the clipping planes are calculated. If all eight signed distance values ​​are negative, it indicates that the bounding box is completely outside the clipping planes and is determined to be invisible. If the axis-aligned bounding box of a partition is not completely excluded by any of the six clipping planes in the individual checks, the corresponding partition is marked as visible and added to the set of visible partitions. The intersection test between the view frustum and the bounding box is performed at the partition level rather than per device model. The number of partitions is much smaller than the total number of device models, thus reducing the computational load of the intersection test.

[0067] Figure 4 The spatial relationship between the view frustum and the bounding boxes aligned with each partition axis, as well as the intersection test results, are shown from a top-down perspective. The vertex of the triangle on the left side of the figure marks the position of the virtual camera. The trapezoidal area extending to the right from the camera represents the top-down projection of the view frustum. The near clipping plane and the far clipping plane are marked with dashed lines. Figure 4 Six bounding boxes are distributed in the center: Partitions A and C are completely inside the view frustum; Partition B intersects the upper boundary of the view frustum but is not completely excluded; all three are considered visible and are represented by solid-line boxes. Partition D is located below and outside the view frustum; Partition E is located above and outside the view frustum; Partition F is located near the clipping plane; all three are considered invisible and are represented by dashed-line boxes. Figure 4 This embodies a spatial filtering mechanism that performs intersection tests at the partition level.

[0068] After determining the set of visible partitions, the rendering load is estimated based on the total number of model faces within each visible partition. In the 3D scene, multiple detail level versions are pre-generated for each device model, with different detail level versions corresponding to different numbers of faces. High detail level versions contain more faces to present the device's fine geometry, while low detail level versions contain fewer faces to reduce the geometric processing load on the rendering pipeline. The number of detail level versions and the number of faces for each version of each device model are generated using a mesh simplification algorithm during the model preprocessing stage and stored in the scene database. The number of detail level versions is determined based on the geometric complexity of the device model. For example, the transformer model has a complex geometry, so four detail level versions are set, with face numbers of 12000, 5000, 2000, and 800 for levels 1 to 4, respectively; the surge arrester model has a relatively simple geometry, so three detail level versions are set, with face numbers of 3000, 1200, and 400 for levels 1 to 3, respectively. The more complex the geometry of the device model, the more detail levels can be set to provide more selectable patch precision levels.

[0069] No. The total number of model faces in a visible partition is the sum of the number of faces of all device models within that partition at their respective current detail levels, expressed as: ,in For the first The number of device models within each visible partition For the first The detail level number of the current device model. For the first Within the visible partition, the first A device model at its current level of detail The number of dough pieces below.

[0070] When converting the total number of model faces into estimated rendering time to quantify the rendering load, it is necessary to obtain the average rendering time per face on the graphics hardware in the current deployment environment. The average rendering time per face is obtained by performing benchmark tests on the target graphics hardware. The benchmark test is performed as follows: several standard test scenes with a known number of faces are rendered on the target graphics hardware, and the rendering time for each group is recorded. A linear regression is then performed with the number of faces as the independent variable and the rendering time as the dependent variable. The slope of the regression line is the average rendering time per face. The intercept of the regression line is the global baseline estimate of the fixed overhead of the rendering pipeline. This intercept reflects the global average fixed overhead under the standard test scene. In reality, due to differences in material configuration and rendering call structure, the fixed overhead of each partition is measured using the partition-specific values ​​described below. The benchmark test is executed once when the 3D scene is deployed to the target hardware platform, and the test results are stored and directly referenced in subsequent frame-by-frame scheduling. For example, when the benchmark test is executed on a workstation equipped with a mid-range discrete graphics card, the rendering times are 0.15 ms, 0.52 ms, 0.98 ms, 1.91 ms, and 4.72 ms respectively when the number of polygons in the standard test scene is 10000, 50000, 100000, 200000, and 500000. Linear regression fitting yields the average rendering time per polygon. The rendering time is approximately 0.0094 microseconds per facet, meaning the average rendering time per facet is approximately 9.4 nanoseconds.

[0071] No. The rendering load of each visible partition is characterized by the estimated rendering time, expressed as: ,in For the first Estimated rendering time for each visible partition, The average rendering time per unit area. For the first The fixed overhead of the rendering pipeline for each visible partition. The fixed overhead of the rendering pipeline includes time-consuming components such as draw call initiation, rendering state switching, and texture resource binding, which are not directly linearly related to the number of faces. The fixed overhead of the rendering pipeline varies between partitions due to the number of material types on the device within that partition; more material types result in more rendering state switching and higher fixed overhead. The fixed overhead of the rendering pipeline for each partition was obtained during benchmarking by measuring the rendering startup time of each partition when the number of faces approached zero. Specifically, this was done by rendering an alternative scene with very few faces using the actual material configuration and draw call structure of each partition and recording the time taken. For example, the fixed overhead of a partition containing 8 materials and 12 draw calls is... Approximately 0.35 milliseconds; the partition contains 3 materials and 4 draw calls, with a fixed rendering pipeline overhead. It takes approximately 0.12 milliseconds.

[0072] The global load margin is obtained by comparing the rendering load of all visible partitions with the frame time budget. The global load margin is defined as follows: ,in For frame time budget, The number of visible partitions in the current frame. For the first Estimated rendering time for each visible partition, This represents the global load margin. A positive global load margin indicates that the total rendering time for all visible partitions under the current detail level configuration does not exceed the frame time budget, and the rendering engine has the capacity to maintain the current detail level or increase the detail level of some device models. A negative global load margin indicates that the total rendering time exceeds the frame time budget, and it is necessary to reduce the rendering time by lowering the detail level of some device models.

[0073] To distribute the global load margin among the visible partitions, it is allocated according to the proportion of the estimated rendering time of each visible partition to the total rendering time, thus obtaining the load margin for each visible partition. The load margin of each visible partition is expressed as: in For the first The load margin is obtained by allocating the load to each visible partition. Allocating the load based on the estimated rendering time means that partitions with longer rendering times bear a larger proportion of detail level adjustments when the global load margin is insufficient, while partitions with shorter rendering times receive smaller adjustments. This helps to avoid uneven image quality caused by some partitions being excessively downgraded while others maintain a high level of detail.

[0074] By determining the set of visible partitions and calculating partition-level load margins, the rendering engine obtains the quantified load status of each visible partition during the rendering preparation phase of each frame. The partition-level load margin value provides a clear quantitative basis for the dynamic adjustment of the detail level in step S4, enabling the decision to raise or lower the detail level to be dynamically determined based on the actual rendering load of the current frame, rather than relying on pre-fixed empirical parameters.

[0075] Step S4: Based on the load margin and the partition-level rendering priority sequence, maintain the detail level of the main device set at no less than the preset threshold in each visible partition, and perform model simplification and replacement on the auxiliary device set in order of rendering priority from low to high to complete the real-time rendering output.

[0076] After obtaining the load margin and partition-level rendering priority sequence for each visible partition, dynamic scheduling at the level of detail is performed within each visible partition based on the load margin and partition-level rendering priority sequence. The scheduling needs to maintain the level of detail of the device models in the primary device set at a level no lower than a preset threshold under frame time budget constraints, and to perform model simplification replacement on the auxiliary device set in ascending order of rendering priority. Model simplification replacement refers to switching the currently bound mesh resources of the device model to a lower level of detail version mesh with fewer faces, allowing the device to be represented in the image with fewer geometric faces, thereby reducing the geometry processing time of the rendering pipeline.

[0077] The preset threshold is the detail level threshold for the main equipment set, and it is set based on the recognizability of key geometric features of the equipment model at each detail level. In a 3D distribution network scene, the key geometric features of the main equipment refer to the geometric structures that reflect the equipment type and operating status, including the moving and stationary contact structure of circuit breakers, the winding outline and radiator fins of transformers, the cross-sectional shape and connection ends of busbar segments, and the shape of the disconnector arm of disconnectors. When the detail level of the equipment model is reduced to a point where the number of faces is insufficient to express the key geometric features, the equipment will be simplified in the image to a basic geometric shape whose equipment type or operating status cannot be identified. The preset threshold is the lowest detail level at which the key geometric features can still maintain recognizability.

[0078] The preset threshold is determined through manual visual evaluation. Power distribution maintenance technicians conduct identification tests on the rendered images of each main equipment model at various detail levels. The accuracy rate of the technicians' identification of equipment type and operating status at each detail level is recorded. The accuracy rate is checked level by level from level 1 to the lowest detail level. The level before the first detail level where the accuracy rate drops below the preset accuracy rate standard (i.e., the lowest detail level where the accuracy rate is still not lower than the preset standard) is determined as the preset threshold. The preset accuracy rate standard is selected based on the accuracy requirements of maintenance operations for equipment identification. For example, the preset accuracy rate standard is 90%. For a circuit breaker model with four detail levels, if the technicians' accuracy rate at level 1 and level 2 is not lower than 90%, and the accuracy rate drops to 72% at level 3, then the preset threshold for the circuit breaker model is determined to be level 2. For a transformer model with four levels of detail, if the accuracy rate of identification at levels 1 to 3 is no less than 90%, and the accuracy rate drops to 65% at level 4, then the preset threshold for the transformer model is determined to be level 3. Different preset thresholds are allowed for different master equipment models, depending on the geometric complexity of each model and its ability to preserve key features.

[0079] Figure 5 Using a transformer model as an example, the differences in mesh geometry across four detail levels are illustrated. From left to right, these represent levels 1 to 4, with 12,000, 5,000, 2,000, and 800 faces respectively. In level 1, the heatsink fins are presented piece by piece, the bushing structure is complete, and the mesh density is high. In level 2, the heatsink assembly is simplified, but the bushings are still identifiable. Level 3 is marked as a preset threshold level, retaining the outline and key geometric features but significantly reducing details; this is the lowest level the main device is allowed to be reduced to. Level 4 only retains the basic geometry, making the device type difficult to identify. Arrows indicate the direction of model simplification / replacement and the direction of detail level improvement between the versions. Figure 5 The preset threshold protection rules are also noted at the bottom.

[0080] The dynamic adjustment of the detail level is handled separately in each visible partition according to the positive or negative state of the partition load margin.

[0081] When the zone load margin When the value is positive, it indicates that the rendering load under the current detail level configuration within the partition is lower than the allocated frame time budget, and there is a rendering time margin available to upgrade the detail level of the device model. Under the condition that the load margin is positive, the device models in the auxiliary device set within the partition are evaluated for upgrades one by one from high to low according to the partition-level rendering priority sequence constructed in step S2. For auxiliary device models that are not currently at the highest detail level, the number of faces added after upgrading their detail level by one level is calculated. , The difference between the number of faces in the target detail level version and the number of faces in the current detail level version will be used to... Multiply by the average rendering time per unit area The additional rendering time required for a level upgrade is determined. If the additional rendering time does not exceed the partition's current remaining load margin, a level upgrade operation is performed. Simultaneously, the partition's remaining load margin is reduced by the corresponding additional rendering time value, and the same device model is further evaluated for upgrade potential. This process is repeated until the device model reaches the highest level of detail or the remaining load margin is insufficient to support the next level upgrade. Then, the same step-by-step upgrade evaluation is performed on the next device model in descending order of rendering priority, until all device models in the auxiliary device set have reached the highest level of detail or the partition's remaining load margin is insufficient to support any upgrade operation for the next device model. If all device models in the auxiliary device set are already at the highest level of detail and there is still remaining load margin, the same step-by-step upgrade evaluation process is further performed on the device models in the primary device set that are not at the highest level of detail, in descending order of rendering priority.

[0082] When the zone load margin A negative value indicates that the rendering load under the current detail level configuration within the partition exceeds the allocated frame time budget. In this case, model simplification and replacement needs to be performed on the auxiliary device set in ascending order of rendering priority to reduce rendering time. Model simplification and replacement begins with the device model in the auxiliary device set with the lowest rendering priority and proceeds sequentially from low to high along the partition-level rendering priority sequence. For each device model to be simplified and replaced, the number of faces reduced by lowering its detail level by one level is calculated. , The difference between the number of faces in the current level of detail version and the number of faces in the target level of detail version after reducing it by one level is used to calculate the difference. Multiply by the average rendering time per unit area The rendering time saved by performing the first-level model simplification replacement is obtained, and the saved rendering time is added to the partition load margin. If the partition load margin is still negative and the device model has not been reduced to the lowest level of detail version, the next level of model simplification replacement is performed on the same device model. This process is repeated until the device model is reduced to the lowest level of detail version or the partition load margin is restored from a negative value to zero or a positive value.

[0083] When the partition load margin returns to zero or a positive value, all model simplification and replacement operations cease. If the current device model has been reduced to the lowest detail level version and the partition load margin is still negative, the model simplification and replacement continues with the next higher-priority device model in the partition-level rendering priority sequence. During this process, the detail level of auxiliary device models is allowed to be reduced to the lowest detail level version. If all device models in the auxiliary device set have been reduced to the lowest detail level version and the partition load margin is still negative, model simplification and replacement begins with the device model with the lowest rendering priority in the main device set, but the detail level of the main device model after model simplification and replacement must not fall below a preset threshold. If all device models in the main device set have been reduced to their respective preset thresholds and the partition load margin is still negative, the current detail level configuration is maintained and model simplification and replacement is no longer performed; the rendering engine receives the result of the current frame time exceeding the budget. The preset threshold protection mechanism of the main device set ensures that devices performing core electrical functions in the power distribution network 3D scene maintain a display quality identifiable by maintenance personnel under any rendering load conditions.

[0084] To avoid frequent switching of the device model's detail level between consecutive rendering frames due to minor fluctuations in load margin, a lag-based decision-making mechanism for detail level switching is introduced into the scheduling process. This lag-based mechanism uses a combination of absolute margin dead zone and inter-frame variation as trigger control. Before executing the detail level adjustment decision for each frame, the load margin of the current frame partition is first determined. Does the absolute value not exceed the lag threshold? ,like If the partition load margin is within the steady-state allowable range, then the model simplification and replacement operation is not performed, and the detail level configuration of the previous frame is maintained. Furthermore, the current frame partition load margin value is compared with the load margin estimate after the scheduling decision was made in the previous frame. The comparison continues until the absolute value of the difference between the two values ​​exceeds the hysteresis threshold. At this time, model simplification and replacement operations are not performed. The load margin estimate after the scheduling decision in the previous frame is obtained by recalculating the partitioned estimated rendering time based on the detailed hierarchical configuration of each device model after the scheduling in the previous frame and subtracting it from the allocated frame time budget. This estimate is recorded and stored when the scheduling in the previous frame is completed. Only when Furthermore, the absolute value of the difference between the current frame's load margin and the estimated load margin after scheduling in the previous frame also exceeds [a certain value]. Detail level adjustments are only performed when the frame first enters the rendering loop or when a partition has just transitioned from an invisible to a visible state. Since there is no previous frame's load margin reference value, hysteresis determination is skipped, and detail level adjustments are performed directly. The hysteresis threshold is expressed as... ,in The hysteresis threshold, For frame time budget, The hysteresis coefficient is used to determine the optimal value for the hysteresis coefficient. A hysteresis coefficient that is too small will not be sufficient to suppress the jitter in the model's geometric appearance caused by frame-by-frame switching, while a hysteresis coefficient that is too large will lead to a slower scheduling response and a larger frame rate fluctuation range. Using an absolute margin dead zone condition can avoid unnecessary scheduling actions when the load margin is within the small deviation allowed by the frame time budget. The inter-frame variation condition can further suppress false triggers caused by calculation accuracy errors when the scene rendering load does not change substantially between consecutive frames. For example, when the frame time budget is 16.7 milliseconds, the hysteresis coefficient... A value of 0.08 corresponds to a hysteresis threshold of 1.34 milliseconds. Under normal viewpoint movement speeds, this can suppress frame-by-frame jumps in detail levels caused by minor fluctuations in load margin. Simultaneously, it can still trigger timely scheduling responses when the viewpoint undergoes significant displacement or rotation, resulting in changes to the visible partition set. The absolute margin dead zone condition means that when the partition rendering time exceeds or falls below the budget by no more than 1.34 milliseconds, the scheduling system tolerates this minor deviation to maintain visual continuity.

[0085] When rapid viewpoint movement causes changes in the set of visible partitions, the visibility state switching of partitions needs to be handled. When a partition changes from invisible to visible, the partition state cache is first checked for a valid record: if the partition state cache contains a historical detail level record for that partition, the previously saved detail level value is read from the cache and restored as the initial configuration; if the partition state cache does not contain a historical record for that partition (i.e., the partition has never entered the visible range in this rendering session), the initial detail level of the device models in the primary device set within the partition is set to the detail level version corresponding to their respective preset thresholds, and the initial detail level of the device models in the secondary device set is set to the lowest detail level version available to each device model. Regardless of whether cache restoration or initial initialization is used, the first frame in which a partition enters the visible state skips the hysteresis judgment and directly performs detail level adjustment based on the current frame's load margin to ensure that the restored or initialized detail level configuration does not cause a load overload under the current frame's time budget constraints. The primary device, starting with a preset threshold, ensures that it meets the identifiability requirements in the first frame upon entering the visible range. The auxiliary devices, by gradually increasing their detail level from the lowest level, avoid instantaneous rendering load spikes caused by all devices rendering at high detail levels when new areas enter the view frustum. When a partition changes from visible to invisible, the current detail level values ​​of each device model within that partition are saved to the partition state cache. The storage capacity of the partition state cache is proportional to the total number of partitions in the scene. Each partition's cache contains the device identifier and corresponding detail level value for each device model within that partition. The storage overhead is negligible in typical power distribution network scenarios.

[0086] After completing the detail level scheduling decision, the rendering engine updates the mesh resource references bound to the rendering pipeline for device models that need to undergo model simplification replacement, based on the scheduling results. Device models requiring model simplification replacement point their mesh resource references to the pre-generated mesh data corresponding to the target detail level version; device models not requiring model simplification replacement retain their existing mesh resource references. The mesh data for each detail level version has already been loaded into graphics memory during the model preprocessing stage. Switching mesh resource references only involves changing the reference pointers, which is completed during the rendering pipeline's preparation stage, without adding additional computation time to the rasterization and fragment processing stages. After completing the detail level updates for all device models within all visible partitions, the rendering engine submits all visible device models of the current frame to the rendering pipeline according to the updated detail level configuration. Real-time rendering output is then completed after vertex processing, rasterization, and fragment shading.

[0087] This embodiment employs a dynamic scheduling process at the detail level. Before each frame is generated, the rendering load of the 3D power distribution network scene is quantitatively evaluated, and the detail level of each device model is adjusted in an orderly manner based on the partition-level load margin and the partition-level rendering priority sequence driven by electrical topology semantics. The main device model maintains a identifiable display quality under a preset threshold protection. Auxiliary device models, based on load margin and rendering priority from low to high, perform model simplification and replacement in an orderly manner, switching between different detail level versions. This ensures that the rendering frame rate remains near the target frame rate during viewpoint movement and scene content changes. A hysteresis mechanism suppresses unnecessary switching of detail levels between consecutive frames, improving the visual continuity of the image. The partition-level scheduling granularity ensures that device models within the same functional unit remain coordinated during detail level adjustments, avoiding the uneven display quality that can occur within the same interval when scheduling is done independently at the individual device level.

[0088] Example 2

[0089] Reference Figure 6 This is the second embodiment of the present application, which provides a detailed level dynamic scheduling system for real-time rendering of a three-dimensional scene of a power distribution network.

[0090] The system includes: a topology partitioning module, a priority construction module, a load estimation module, and a hierarchical scheduling module.

[0091] The topology partitioning module is used to spatially partition the three-dimensional scene based on the electrical topology relationship and spatial distribution of each device model in the distribution network three-dimensional scene, using the functional units in the electrical topology as the dividing basis, and to establish the spatial boundaries of each partition and the device index within the partition.

[0092] The priority construction module is used to divide the device model into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology within each spatial partition, assign rendering priority to each device according to the node level, and construct a partition-level rendering priority sequence.

[0093] The load estimation module is used to determine the set of visible partitions based on the current viewpoint position and the range of the view frustum, estimate the rendering load by combining the total number of model patches in each visible partition, and compare the rendering load with the frame time budget to obtain the load margin of each visible partition.

[0094] The hierarchical scheduling module is used to maintain the detail level of the main device set at no less than a preset threshold in each visible partition according to the load margin and the partition-level rendering priority sequence, and to perform model simplification and replacement on the auxiliary device set in order of rendering priority from low to high to complete real-time rendering output.

[0095] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0096] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of this application without departing from the spirit and scope of protection of the claims. All of these variations are within the protection scope of this application.

[0097] It should be noted that the collection, storage, and use of all personal information involved in the technical solution of this invention must be carried out only after obtaining the explicit authorization and separate consent of the information subject. The processing of personal information-related data strictly complies with the requirements of currently effective national laws, regulations, and relevant standards and specifications. The collected personal information is limited to use within the specific purpose necessary to achieve the technical solution of this invention, and no processing beyond that purpose is performed. Necessary technical and management measures are taken to ensure the security of personal information.

Claims

1. A method for dynamic scheduling of detail levels in real-time rendering of a 3D scene of a power distribution network, characterized in that, include: Based on the electrical topology and spatial distribution of each device model in the 3D scene of the power distribution network, the 3D scene is spatially partitioned according to the functional units in the electrical topology, and the spatial boundaries of each partition and the device index within the partition are established. Within each spatial partition, the device model is divided into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology. Rendering priorities are assigned to each device according to the node level to construct a partition-level rendering priority sequence. The set of visible partitions is determined based on the current viewpoint position and the range of the view frustum. The rendering load is estimated by combining the total number of model patches in each visible partition. The rendering load is compared with the frame time budget to obtain the load margin of each visible partition. Based on the load margin and partition-level rendering priority sequence, the detail level of the main device set is kept no lower than a preset threshold in each visible partition, and the model simplification and replacement are performed on the auxiliary device set in order of rendering priority from low to high to complete the real-time rendering output.

2. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The spatial partitioning includes: Read the electrical topology data of the distribution network, perform a depth-first traversal along the electrical topology path, and determine the boundary nodes according to the functional unit boundary rules; The device nodes between two adjacent boundary nodes are grouped into the same functional unit group, and the device nodes at the end of the topology are grouped into the functional unit group where the nearest boundary node is located. In a ring network structure, the node where the ring network interconnection switch is located is used as the boundary node; Spatial boundaries are formed based on the vertex coordinate range of the device models within the functional unit group in the coordinate direction, and an intra-area device index is generated that records the unique identifier of the device model, the device type code, and the reference of the mesh resources.

3. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The main equipment set and the auxiliary equipment set include: Query node information in the electrical topology of the device model within the partition; The device models located on the electrical topology backbone are grouped into the main device set; Auxiliary electrical equipment located on branch links and non-electrical structural components that do not have node status are grouped into the auxiliary equipment set; Each device model in the same partition belongs to only one set.

4. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The rendering priority includes: The values ​​are assigned according to the rule that the higher the electrical topology node level, the greater the rendering priority value; The rendering priority value is determined by the normalized device category weight score and node level score. The device category weight adjustment coefficient and the node level score adjustment coefficient are used to determine the proportion. After mapping, the auxiliary electrical equipment in the auxiliary equipment set has a rendering priority value that is lower than the lowest rendering priority value in the main equipment set but higher than the rendering priority value of non-electrical structural components.

5. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The partition-level rendering priority sequence includes: Arrange all device models within the partition in descending order of rendering priority value to form an ordered array; The elements of the ordered array record the unique identifier of the device model, the set tag to which it belongs, and the rendering priority value; The set identifier is used to distinguish between the primary device set and the auxiliary device set; The partition-level rendering priority sequence is regenerated when the electrical topology changes.

6. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The set of visible partitions includes: The current viewpoint position is determined based on the world coordinates of the virtual camera, and the frustum range is determined based on the vertical field of view of the virtual camera, the aspect ratio of the image, the distance to the near clipping plane, and the distance to the far clipping plane. Intersection tests were performed between the spatial boundaries of each partition and the view frustum; Partitions that are not completely excluded by the clipping plane of the view frustum are marked as visible partitions and added to the set of visible partitions.

7. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The total number of model facets includes: Multiple detail level versions are generated in advance for each device model, and the number of faces in each detail level version is stored in the scene database; Read the number of faces of all device models within the visible partition at the current detail level; The total number of model faces in the visible partition is obtained by summing the number of faces of all device models within the visible partition.

8. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The rendering load includes: Render a standard test scene with a known number of faces on the target graphics hardware and record the rendering time to obtain the average rendering time per face. The estimated rendering time for the visible partition is generated by the total number of model faces in the visible partition, the average rendering time per face, and the fixed overhead of the rendering pipeline. The estimated rendering time is used as the rendering load for the visible partition.

9. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The load margin includes: The rendering load of all visible partitions is compared with the frame time budget to generate a global load margin; The global load margin is distributed according to the proportion of the estimated rendering time of each visible partition to the total estimated rendering time of all visible partitions. The allocation results are used as the load margin for each visible partition.

10. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The preset threshold includes: Power distribution maintenance technicians conducted identification tests on the rendered images of the main equipment model at various detail levels. Record the accuracy of device type and operating status identification at each detailed version level; The level preceding the detail level whose recognition accuracy first falls below the preset recognition accuracy standard is determined as the preset threshold for the device model in the master device set.

11. The method for dynamic scheduling of detail levels in real-time rendering of a three-dimensional scene of a power distribution network according to claim 1, characterized in that, The model simplification and replacement include: When the partition load margin is negative, start by rendering the device model with the lowest priority from the auxiliary device set and reduce the level of detail. If the partition load margin is still negative after the same device model is downgraded to the lowest detail level version, it will switch to the next device model with a higher rendering priority along the partition-level rendering priority sequence. If the partition load margin is still negative after all auxiliary equipment sets are reduced to the lowest level of detail, the level of detail of the primary equipment set is reduced, and the level of detail of the equipment models in the primary equipment set is not lower than the preset threshold.

12. A detailed-level dynamic scheduling system for real-time rendering of a three-dimensional scene in a power distribution network, used to implement the detailed-level dynamic scheduling method for real-time rendering of a three-dimensional scene in a power distribution network as described in any one of claims 1 to 11, characterized in that, include: The module includes a topology partitioning module, a priority building module, a load estimation module, and a hierarchical scheduling module. The topology partitioning module is used to spatially partition the three-dimensional scene based on the electrical topology relationship and spatial distribution of each device model in the distribution network three-dimensional scene, using the functional units in the electrical topology as the dividing criteria, and to establish the spatial boundaries of each partition and the device index within the partition. The priority construction module is used to divide the device model into a set of main devices and a set of auxiliary devices according to the node level of the device in the electrical topology within each spatial partition, assign rendering priority to each device according to the node level, and construct a partition-level rendering priority sequence. The load estimation module is used to determine the set of visible partitions based on the current viewpoint position and the range of the view frustum, estimate the rendering load by combining the total number of model patches in each visible partition, and compare the rendering load with the frame time budget to obtain the load margin of each visible partition. The hierarchical scheduling module is used to maintain the detail level of the main device set at no less than a preset threshold in each visible partition according to the load margin and the partition-level rendering priority sequence, and to perform model simplification and replacement on the auxiliary device set in order of rendering priority from low to high to complete real-time rendering output.

Images