Transformer substation secondary circuit three-dimensional panoramic visual modeling system adopting convolutional neural network

Abstract

The invention relates to the technical field of visual modeling, in particular to a transformer substation secondary circuit three-dimensional panoramic visual modeling system adopting a convolutional neural network, which comprises a structure attribute mapping module, a spatial layout generation module, a spatial conflict screening module, a target patch selection module and a modeling instruction output module. According to a logic structure between nodes and equipment attributes, equipment geometric morphology and spatial relationship are automatically analyzed, entity occupation and connection states in a three-dimensional environment are deduced, spatial intersection and direction matching in equipment layout are comprehensively judged, and the target position is optimized through spatial overlapping conflict screening and target position optimization. Equipment nodes under three-dimensional coordinates are deployed in a system space in a coordination mode, a space mapping instruction of an equipment model is generated in a dynamic parameter driving mode, full-automatic three-dimensional modeling output of a complex loop is supported, the system space consistency and the structure reduction capacity are improved, and digital management and scene expansion are facilitated.

Description

Technical Field

[0001] This invention relates to the field of visualization modeling technology, and in particular to a three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks. Background Technology

[0002] The field of visualization modeling mainly involves modeling methods that intuitively express physical or logical structures through graphics. It encompasses the visual presentation and 3D reconstruction of information such as real-world scenes, system structures, and operational states, and is widely used in industrial simulation, power system management, building information modeling, and virtual reality. Among these, the traditional 3D panoramic visualization modeling system for substation secondary circuits refers to modeling the connection relationships between secondary equipment such as relay protection, automatic control, signal acquisition, and measuring instruments within the substation, and then manually constructing a 3D model using graphical software to achieve a visual representation of the spatial structure of the substation's secondary system.

[0003] In existing technologies for 3D visualization modeling of substation secondary circuits, manual modeling is often used to represent equipment layout and circuit relationships. Due to limitations in manual processing capabilities and manual operation procedures, it is difficult to simultaneously complete the spatial positioning of secondary equipment and the processing of circuit connection relationships. When encountering scenarios with a large number of devices and complex wiring, overlapping and occupancy of equipment space may occur, connection paths may be difficult to sort out, the model structure may lack integrity and consistency, spatial conflicts may not be efficiently identified, and the rationality of model layout may be limited, affecting the intuitiveness and effectiveness of subsequent maintenance, management, and system operation. Summary of the Invention

[0004] To address the technical problems existing in the prior art, this invention provides a 3D panoramic visualization modeling system for substation secondary circuits using a convolutional neural network. The technical solution is as follows:

[0005] On the one hand, a 3D panoramic visualization modeling system for substation secondary circuits using convolutional neural networks is provided, including:

[0006] The structural attribute mapping module is based on the substation secondary circuit. It analyzes the relationship between the node logical number and the bus number, compares the functional category code with the equipment size parameters, determines the type of equipment associated with each node, maps it to the node number, and obtains the node structural parameter set.

[0007] Based on the node structure parameter set, the spatial layout generation module determines the geometric relationship between the device size and the spatial coordinate origin, calculates the three-dimensional spatial bounding boundary of each node device, and obtains the spatial layout envelope set.

[0008] The spatial conflict screening module analyzes the overlap between the current node and the area enclosed by the adjacent node based on the spatial layout envelope set, determines the spatial coverage and directional relationship of the enclosed area, compares the spatial overlap characteristics, and obtains the node spatial conflict identifier.

[0009] The target patch selection module filters out non-overlapping patches based on the node spatial conflict identifier, analyzes the directional consistency of the patch and the upstream node connection, judges the patch boundary and standard size matching, and obtains the node projection center parameters.

[0010] The modeling instruction output module analyzes the compatibility between spatial coordinates and functional category templates based on the node projection center parameters, combines the patch center coordinates and equipment model information, and obtains the three-dimensional model layout instruction.

[0011] On the other hand, the node structure parameter set includes category identifier, shape data and interface information; the spatial layout envelope set includes position coordinates, spatial volume and attitude information; the node spatial conflict identifier includes conflict location, conflict type and association number; the node projection center parameters include three-dimensional position, patch number and relative distance to the generatrix; and the three-dimensional model deployment instructions include deployment order, model type and coordinate parameters.

[0012] On the other hand, the structure attribute mapping module includes:

[0013] The logic mapping construction submodule is based on the substation secondary circuit. It analyzes the field content in the node logic number and the bus number, compares the attribution relationship of each group of numbers in the character structure, and determines whether the circuit unit in the logic number is consistent with the main line identifier of the bus number, thus obtaining the attribution mapping dataset.

[0014] The function code comparison submodule analyzes the category definition and structural fields in the standard equipment size parameters based on the attribution mapping dataset, identifies the corresponding combinations between the function category code and the size parameters, determines whether each combination meets the classification conditions, and obtains the function size matching set.

[0015] The structural attribute organization submodule analyzes the category, shape, and interface parameters of the node-associated devices based on the functional size matching set, adjusts the classification method of various structural information, and classifies and organizes the node device structural data to obtain the node structural parameter set.

[0016] On the other hand, the spatial layout generation module includes:

[0017] The size relationship judgment submodule compares the relative positions of each node device with the spatial coordinate origin based on the node structure parameter set, determines the distribution of the three-dimensional size boundary of the device in the spatial coordinate system, and optimizes the initial spatial layout of the nodes by combining the geometric distance relationship between the device center point and the coordinate origin, thus obtaining the spatial association parameter set.

[0018] Based on the spatial association parameter set, the spatial boundary calculation submodule calculates the three-dimensional enclosing boundary of each node device, analyzes the boundary distribution range of the node on the coordinate axis, adjusts the center coordinates of the spatial enclosing area of ​​the node, optimizes the boundary limit distribution, determines the boundary attributes of the node in space, and obtains the three-dimensional boundary parameter set.

[0019] Based on the three-dimensional boundary parameter set, the layout information summarization submodule adjusts the arrangement order of the enclosing regions of each node in the spatial coordinate system, optimizes the spatial distribution between devices, and integrates the spatial enclosing information and layout mapping parameters of the node devices to obtain a spatial layout envelope set.

[0020] On the other hand, the space conflict screening module includes:

[0021] The spatial overlap analysis submodule, based on the spatial layout envelope set, compares the overlap of the enclosing regions of the current node and the spatially adjacent nodes in the three-dimensional coordinate system, determines the overlapping position of the node enclosing boundaries, identifies the coordinate range of the spatially overlapping part, and summarizes the intersection characteristics between each pair of adjacent nodes to obtain spatial overlap relationship data.

[0022] The directional relationship judgment submodule, based on the spatial overlap relationship data, filters the enclosure parameters of each spatially related device in the control cabinet structure, analyzes the spatial coverage of the enclosed area of ​​each node, compares the arrangement direction between the enclosed areas of the nodes, judges the correspondence between spatial coverage and direction, and obtains the spatial coverage direction parameters.

[0023] The overlap feature comparison submodule compares the area and volume distribution of the spatial intersection of each node based on the spatial coverage direction parameters, analyzes the spatial overlap features within the area surrounded by the nodes, summarizes the key parameters of each spatial intersection, determines the node number and corresponding area where a spatial conflict occurs, and obtains the node spatial conflict identifier.

[0024] On the other hand, the target patch selection module includes:

[0025] The patch filtering submodule filters patches that do not overlap spatially based on the node spatial conflict identifier, analyzes the overlap between each patch and the spatial boundary of adjacent devices, determines the independent boundary of the patch in three-dimensional space, and removes patches that overlap spatially or have intersecting boundaries to obtain an independent patch parameter set.

[0026] The directional consistency analysis submodule analyzes the connection direction between all patches and upstream nodes based on the independent patch parameter set, compares the spatial consistency of the connection between the patch center point and the upstream node center, determines the correspondence between the spatial arrangement of patches and the route direction, identifies patches with consistent arrangement directions, and obtains directional consistency distribution groups.

[0027] The spatial difference comparison submodule compares the spatial distance from the center point of each patch to the center of the busbar based on the directional consistency distribution group, analyzes the adaptation of each patch boundary to the standard size, judges the degree of matching between the patch and the design size, identifies patches with small spatial differences and matching sizes, and obtains the node projection center parameters.

[0028] On the other hand, the modeling instruction output module includes:

[0029] Based on the node projection center parameters, the spatial coordinate analysis submodule determines the geometric relationship between the three-dimensional spatial coordinates of each node and the origin of the modeling space, compares the distribution order of the center points of each node patch in space, summarizes the spatial arrangement state, and obtains the coordinate arrangement parameters.

[0030] The template adaptation judgment submodule compares the spatial coordinates with the device function category based on the coordinate layout parameters, analyzes the structural information of the three-dimensional template corresponding to each node, judges the adaptability between the template and the spatial coordinates, identifies the node index that matches the spatial coordinates and the template structure, and obtains the template adaptation index group.

[0031] The data instruction generation submodule, based on the template adaptation index group, combines the three-dimensional spatial coordinates of the nodes, the center point information of the facets, and the equipment model information, calculates the arrangement order of each node in the three-dimensional space, adjusts the correspondence between the equipment type and the spatial coordinates, and outputs the parameters required for spatial layout to obtain the three-dimensional model layout instruction.

[0032] On the other hand, the substation secondary circuit refers to the logical circuit of each signal and control path in the substation secondary system, including the lines of measurement and control, protection and automation functions. The node logical number refers to the number used to identify the functional unit in each circuit. The bus number refers to the unique number representing the main electrical bus of the substation, which determines the main circuit affiliation of each secondary device in the system.

[0033] On the other hand, the origin of the spatial coordinates refers to the reference point of the three-dimensional modeling space, which is the starting point set for modeling, and the three-dimensional spatial enclosing boundary refers to the smallest envelope of the space occupied by the device in the three-dimensional coordinate system.

[0034] On the other hand, the current node refers to the secondary equipment or loop node that is being analyzed or is about to be deployed, and the overlapping feature refers to the area, volume or overlapping shape of the spatial intersection.

[0035] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0036] Based on the logical structure between nodes and device attributes, the system automatically analyzes the geometric shape and spatial relationship of devices, infers the entity occupancy and connection status in the 3D environment, comprehensively judges the spatial intersection and directional matching in the device layout, and deploys device nodes in the system space in a coordinated manner through spatial overlap conflict screening and target position optimization. It uses a dynamic parameter-driven approach to generate spatial mapping instructions for device models, supports fully automatic 3D modeling output of complex loops, improves the system's spatial consistency and structural restoration capabilities, and facilitates digital management and scenario expansion. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a schematic diagram of the system of the present invention;

[0039] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0040] Figure 3 This is a flowchart of the structural attribute mapping module of the present invention;

[0041] Figure 4 This is a flowchart of the spatial layout generation module of the present invention;

[0042] Figure 5 This is a flowchart of the spatial conflict screening module of the present invention;

[0043] Figure 6 This is a flowchart of the target patch selection module of the present invention;

[0044] Figure 7 This is a flowchart of the modeling instruction output module of the present invention. Detailed Implementation

[0045] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0046] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0047] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0048] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0049] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0050] This invention provides a 3D panoramic visualization modeling system for substation secondary circuits using convolutional neural networks, such as... Figure 1 As shown, the system includes:

[0051] The structural attribute mapping module is based on the substation secondary circuit. It analyzes the correspondence between node logical numbers and bus numbers, compares the matching between equipment function category codes and standard equipment size parameters, determines the type of equipment associated with each node, sorts out the structural attribute parameters of the nodes, maps them to the corresponding node numbers, and obtains the node structural parameter set.

[0052] The spatial layout generation module, based on the node structure parameter set, determines the spatial geometric relationship between the equipment size parameters and the spatial coordinate origin, calculates the three-dimensional spatial bounding boundary of each node device, optimizes the center point coordinates and boundary range of the bounding area of ​​each node, summarizes the spatial layout information of the equipment, and obtains the spatial layout envelope set.

[0053] The spatial conflict screening module is based on the spatial layout envelope set. It analyzes the geometric overlap between the current node and the area surrounded by adjacent nodes, identifies the spatially related equipment surrounding data within the control cabinet structure, judges the spatial coverage and directional relationship between the areas surrounded by each node, compares the overlapping features of the spatial intersection, and obtains the node spatial conflict identifier.

[0054] The target patch selection module filters patches that do not overlap spatially based on node spatial conflict identifiers, analyzes the consistency of spatial direction between the patch and the upstream node connection, compares the spatial distance from the patch center point to the generatrix center, judges the matching of the patch boundary with the standard size, selects the patch with the smallest spatial difference, and obtains the node projection center parameters.

[0055] The modeling instruction output module analyzes the corresponding spatial coordinates based on the node projection center parameters, determines the compatibility of the coordinates with the 3D template of the device function category, combines the center coordinates of the patch with the device model information, and outputs the data instructions for 3D spatial layout to obtain the 3D model layout instructions.

[0056] The node structure parameter set includes category identifier, shape data and interface information; the spatial layout envelope set includes position coordinates, spatial volume and attitude information; the node spatial conflict identifier includes conflict location, conflict type and association number; the node projection center parameters include 3D position, patch number and relative distance to the generatrix; and the 3D model deployment instructions include deployment order, model type and coordinate parameters.

[0057] In the structural attribute mapping module, the substation secondary circuit refers to the logical circuit of each signal and control path in the substation secondary system, covering lines with functions such as measurement and control, protection, and automation; the node logical number refers to the number used to identify the functional unit (such as circuit breaker, protection unit, measurement and control point, etc.) in each circuit, facilitating system identification and indexing; the bus number refers to the unique number representing the main electrical bus of the substation, determining the main circuit affiliation of each secondary device in the system; the correspondence method refers to the affiliation or association mapping relationship between the node logical number and the bus number, used to establish the connection relationship between secondary equipment and primary main electrical lines; the equipment function category code refers to the unique function code set for different types of secondary equipment (such as protection, measurement and control, communication, etc.), used to distinguish the functional attributes of the equipment; the standard equipment size parameters refer to the size data set during equipment production or design, including structural parameters such as length, width, and height, used for subsequent three-dimensional spatial mapping; the associated equipment type refers to clarifying the type of secondary equipment represented by the node according to the equipment function category code and node affiliation; the structural attribute parameters refer to the structural information such as the functional category, size, interface, and connection direction of the equipment, serving as the basic data for subsequent three-dimensional modeling and layout.

[0058] In the spatial layout generation module, the spatial coordinate origin refers to the reference point of the 3D modeling space, usually the starting point set by the modeling system, serving as the reference zero point for the position coordinates of all devices; the 3D spatial enclosing boundary refers to the smallest envelope of the space occupied by the device in the 3D coordinate system, generally a cuboid or cube boundary, covering all physical entities of the device; the node enclosing region refers to the physical space area covered by the enclosing body of a specific secondary device in 3D space; the center point coordinates refer to the coordinate values ​​of the center point of the device's enclosing body in 3D space, serving as a reference point for device positioning and subsequent layout; the boundary range refers to the outermost coordinate values ​​of the device's enclosing body in the X, Y, and Z directions, defining the outer limit of the space occupied by the device in 3D space; and the spatial layout information refers to the set of information such as the enclosing relationship, relative position, and layout order of all secondary devices (or nodes) in the 3D coordinate space.

[0059] In the spatial conflict screening module, the current node refers to a specific secondary equipment or loop node that is being analyzed or is about to be deployed; spatially adjacent nodes refer to other equipment nodes that are physically close to the current node in three-dimensional space and have a spatial proximity relationship; geometric overlap refers to the intersection of the areas enclosed by two or more equipment in three-dimensional space, that is, the state of physical overlap or partial overlap in space; spatially related equipment enclosure data refers to the detailed data of the areas enclosed by other equipment that have spatial conflicts with the current node within specific structures such as control cabinets; spatial coverage refers to the parts of the equipment enclosures that overlap or are covered by each other in space, to determine whether there is a space occupation conflict between the two equipment; directional relationship refers to the orientation or arrangement between the enclosures, to determine the rationality and spatial coordination of the equipment layout by analyzing the equipment orientation; overlap characteristics describe the area, volume, and overlapping shape of the spatial intersection, to quantitatively determine the degree of conflict.

[0060] In the target surface selection module, surfaces without spatial overlap refer to those that, after spatial conflict screening, are confirmed not to have spatial overlap with their respective equipment and are valid placement surfaces for the equipment. Upstream node connection refers to the upstream node directly connected to the current equipment node in the loop, used for direction determination during equipment deployment. Spatial direction consistency refers to comparing the consistency between the connection direction between the surface and the upstream node and the loop logic direction to optimize the route and equipment layout sequence. Spatial distance refers to the three-dimensional coordinate distance from the center point of the surface to the center of the busbar (or its reference point), used to evaluate the compactness or rationality of the equipment layout. Surface boundary refers to the physical boundary of the surface in three-dimensional space, i.e., the geometric range of the placement surface. Standard size refers to the reference size used to check whether the surface matches the equipment design size. Matching condition refers to judging the degree of deviation between the surface boundary and the standard size, used to screen the most suitable equipment placement position. Surface with the smallest difference refers to the surface whose size and position are closest to the design standard among multiple candidate surfaces.

[0061] In the modeling instruction output module, the 3D template refers to the standard 3D digital model corresponding to the equipment type, which is used to directly call the modeling system for equipment layout; the equipment model information refers to the set of information such as the 3D structure, interface, and occupied space of the equipment model, which is the basis for generating spatial layout instructions; the data instructions refer to the set of parameters assembled by the patch coordinates and the equipment 3D model information, which can be directly read and executed by the modeling engine, and are used to automatically complete the placement of 3D equipment.

[0062] like Figure 2 and Figure 3 As shown, the structure attribute mapping module includes:

[0063] The logic mapping construction submodule is based on the substation secondary circuit. It analyzes the field content in the node logic number and the bus number, compares the attribution relationship of each group of numbers in the character structure, and determines whether the circuit unit in the logic number is consistent with the main line identifier of the bus number, thus obtaining the attribution mapping dataset.

[0064] First, the logical numbers of each node are extracted item by item. The characters at each position in the node number string are separated, and subfields such as the device function identifier segment, tag number segment, and sequence segment are extracted. Simultaneously, the bus number list is read, and the prefix and main line segments of the bus numbers are extracted character by character. During the comparison, a one-to-one comparison is performed between the first three characters of the node number and the first three characters of the bus number. If all three characters are identical, they are considered to have a logical correspondence. Further comparison is made between the middle three characters of the node number and the middle three characters of the bus number to determine if there is consistency in the tag number. If there is inconsistency, the degree of overlap of the last characters is determined. The number of identical characters is counted digit by digit, and the similarity is calculated by dividing the number of identical characters by the total number of digits. If they are similar... If the similarity reaches 60% or higher, it is considered to have a weak attribution attribute. To further confirm, the information of the main equipment connected to the same node number is retrieved, and the upstream equipment number of the node is searched in the wiring configuration table. The main bus number connected to it is then checked to see if it is the same number. This serves as an auxiliary criterion for judging the attribution relationship. In a real scenario, for a node numbered "PTF20306", the first three digits "PTF" are the same as the bus "PTF101", the middle three digits "203" are different from "101", and the last digit "06" does not match "01". However, if it is found that the upstream equipment of this node is connected to "PTF101", it can be mapped as the attribution node of the bus "PTF101". The relationship between all logical numbers and bus numbers, which has been confirmed through multi-layer character analysis and connection, is summarized into a set of attribution mapping datasets.

[0065] The function code matching submodule analyzes the category definition and structural fields in the standard equipment size parameters based on the attribution mapping dataset, identifies the corresponding combinations between the function category code and the size parameters, determines whether each combination meets the classification conditions, and obtains the function size matching set.

[0066] Extract the functional category code field corresponding to each node. This field is usually located in the prefix segment of the logical number. Based on this, search the standard equipment parameter table item by item to find the equipment type definition field corresponding to each functional code. Then, extract the structural parameter fields of this equipment type in the standard database, including length, width, and height values, and record the structural morphological characteristics of this type of equipment as a reference. Subsequently, compare the structural parameters of the actual deployed equipment with the standard data item by item. Perform a difference comparison for each item of length, width, and height to determine whether the actual size falls within the upper and lower fluctuation range set by the standard size. If any size exceeds the fluctuation range, it is judged as a mismatch. If the three structural parameters... If all values ​​fall within the floating range, the function category is considered to be related to the current device size, and the combination is included in the function size matching set. If multiple size combinations are found to correspond to a certain category during the identification process, the frequency of all sizes under that category is further counted, and the size group with the higher frequency is selected as the representative. In actual use, if the function category "PRT" corresponds to multiple devices in a certain site, most of the device sizes are concentrated in the range of 600×450×220, while a very small number of 670×460×230 specifications appear. In this case, the former is taken as the representative specification for the classification and matching between function and size. Common combinations are included in the result set, while uncommon combinations are not included.

[0067] The structural attribute organization submodule analyzes the category, shape and interface parameters of the node-related equipment based on the functional size matching set, adjusts the classification method of various structural information, classifies and organizes the node equipment structural data, and obtains the node structural parameter set.

[0068] For each functional category, the relevant equipment's external parameters and interface configuration data are extracted sequentially. First, the functional codes are identified to confirm the equipment's category. Then, the structural data table under the corresponding equipment number is read, and the actual length, width, and height dimensions of the equipment are extracted and categorized by functional category. Upper and lower limits are set for the size range of each equipment category. By statistically analyzing the maximum and minimum values ​​of each dimension across all equipment, the boundary range occupied by this type of equipment in the deployment space is defined. Further analysis of the equipment's interface characteristics is conducted, extracting parameters such as the number of interfaces, interface type, and interface distribution method. Statistics on interface field values ​​are performed, and each interface category is recorded. The frequency of interface type occurrence is determined by the number of instances. If the interface types are consistent but the number of instances differs, the most frequently occurring instance is used as the standard configuration. If the interface types differ significantly, the main interface types are defined proportionally and recorded. For example, if a certain type of "PRT" device has a large number of samples with four RJ45 interfaces and a small number of samples with two RS485 interfaces, the former is defined as the main interface configuration of this type of device. The structural parameters and interface definitions of each type of device are summarized and written into the node structural parameter set in a unified format. Each record includes information such as category identifier, length range, width range, height range, main interface type, and number of typical interfaces.

[0069] like Figure 2 and Figure 4 As shown, the spatial layout generation module includes:

[0070] The size relationship judgment submodule compares the relative positions of each node device with the spatial coordinate origin based on the node structure parameter set, determines the distribution of the three-dimensional dimension boundary of the device in the spatial coordinate system, and optimizes the initial spatial layout of the nodes by combining the geometric distance relationship between the device center point and the coordinate origin, thus obtaining the spatial association parameter set.

[0071] First, the size field of each device in the structural parameter set is processed. A ternary combination calibration is performed on each set of length, width, and height data. Simultaneously, the current center coordinates of each node are extracted as reference points for the node's spatial position. Then, the triaxial difference between the coordinate center point and the origin of the modeling space is calculated. The coordinate differences in the X, Y, and Z directions are used as the relative position vectors of the nodes to the origin. This vector is then fed into a two-dimensional image matrix representation. By setting a fixed pixel step size, the original coordinates are mapped to a standard image size matrix to construct a spatial distribution image. This image serves as the input image for the CNN to extract the geometric position feature distribution pattern between each node and the origin. Simultaneously, a size border mask is constructed in the device size nested image and superimposed onto the original image as a channel input. The CNN extracts the feature response intensity maps of different nodes in the spatial image through multi-layer convolution operations. The center point and boundary responses are fused to form the initial spatial distribution. The image is first mapped, and then a feature map inverse mapping is performed to project the regions with high response values ​​in the image onto a three-dimensional spatial coordinate system. The peak response points are selected as potential initial deployment locations. The candidate deployment coordinates of each node are sorted by response intensity, and the coordinates of the top five nodes with the highest response values ​​are used as the initial selection list. Then, the distance relationship between the size boundary and the origin coordinates in the three axes is combined to determine the relative spatial distribution region of the current device and the origin. The initial layout priority is adjusted according to the spatial similarity image output by the CNN. In a real scenario, when a device has a size of 600×400×220 and its center point is located at (150, 250, 300), its convolution response peak is close to the coordinate origin and there is a boundary projection overlap region in the Y-axis direction. Then, this node is determined to be a node that should be deployed closer to the origin. This process is repeated to complete the spatial distribution map assignment between all nodes and the origin, and a spatial association parameter set is obtained.

[0072] The spatial boundary calculation submodule calculates the three-dimensional bounding boundary of each node device based on the spatial association parameter set, analyzes the boundary distribution range of the node on the coordinate axis, adjusts the center coordinates of the spatial bounding area of ​​the node, optimizes the boundary limit distribution, determines the boundary attributes of the node in space, and obtains the three-dimensional boundary parameter set.

[0073] First, the 3D center coordinates and size triplet of each node are extracted from the associated parameters. The device boundary is then extended by half the dimension length along the X, Y, and Z axes, using the center point as the reference, to construct a bounding box boundary point set. All boundary point sets are then fed into a 3D spatial matrix for structured processing. The minimum and maximum points of the node boundary range on each coordinate axis are registered to generate the original 3D boundary intervals. This interval data is then projected onto a 3D raster image to construct a visible boundary pattern. The boundary pattern is fed into a CNN for feature recognition, identifying high-response regions generated by the convolutional kernel in the boundary images of each node. Regions where multiple node boundaries spatially intersect are marked with high response overlap intensity blocks. Simultaneously, points with high response overlap rates between node boundaries are marked in the feature map extracted by the CNN and displayed in the spatial map. These point boundary intervals are identified as areas requiring optimization, and responses are extracted from these areas. For nodes with prominent values, an offset operation is performed on their center coordinates. The offset direction is determined by the direction of the weaker response value in the boundary map. Each coordinate adjustment does not exceed 10% of the device's long side. After adjustment, the bounding box is regenerated and fed into the CNN again for feature map verification until the highly overlapping area of ​​the response value is effectively diluted. For example, when two devices are located at (400, 300, 100) and (460, 300, 100) respectively, with sizes of 600×400×220 and 620×420×240 respectively, and the X-axis boundary of the two devices overlaps by 80mm, and the intensity of the overlapping area of ​​the response in the CNN boundary map is continuously greater than the set boundary threshold, the center point of the second device is adjusted to a new coordinate that increases by 100mm in the X-axis direction. The boundary image is rebuilt for all adjusted nodes, and the coordinates of each node, the start and end points of the boundary, and its coverage in the convolutional response map are recorded to obtain the three-dimensional boundary parameter set.

[0074] The layout information summarization submodule, based on the three-dimensional boundary parameter set, adjusts the arrangement order of the enclosing regions of each node in the spatial coordinate system, optimizes the spatial distribution between devices, integrates the spatial enclosure information of node devices and layout mapping parameters, and obtains the spatial layout envelope set.

[0075] The boundary start coordinates, center point coordinates, and functional classification labels of each node are extracted. All nodes in the space are initially sorted according to their Z-axis height, prioritizing lower-level nodes. Nodes in the same layer are then arranged in ascending order according to their X-axis coordinates to generate a preliminary spatial arrangement order. Subsequently, all nodes are surrounded by their boundaries and projected onto the overall spatial view image in the form of a 3D bounding box, constructing a 3D device arrangement distribution map. This image is then fed into a CNN to perform spatial sparsity determination, identifying the distribution differences between low-frequency and dense response regions in the convolutional layers. Low-frequency response regions are identified as blank areas, and dense regions are marked as conflict areas. Nodes are sorted based on their response heat values ​​in the CNN feature map, identifying nodes with poor spatial utilization and reassigning their spatial coordinates. The convolutional response is analyzed during the coordinate adjustment process. The positional differences of devices along the Y-axis in the image determine whether a column-wise rearrangement is needed. Nodes in the column whose space exceeds the preset spacing are moved sequentially to the next row. Convolutional response template matching analysis is performed on nodes with the same functional labels to identify whether they have a common layout direction requirement. If nodes with the same label in the CNN image are found to have a consistent channel response trajectory, then the layout direction of such nodes is forcibly aligned. For example, communication nodes are arranged in a concentrated manner along the Y-axis. The coordinates of the center points of all nodes after rearrangement are summarized, and the start and end values ​​of the boundary of each node and its arrangement order index are recorded. The node classification number of the convolutional layer in the response image is also bound together to generate a spatial layout envelope set formed by node spatial coordinates, arrangement position, label classification, and boundary combination.

[0076] like Figure 2 and Figure 5 As shown, the space conflict screening module includes:

[0077] The spatial overlap analysis submodule is based on the spatial layout envelope set. It compares the overlap of the enclosing regions of the current node and the spatially adjacent nodes in the three-dimensional coordinate system, determines the overlapping position of the node enclosing boundary, identifies the coordinate range of the spatially overlapping part, and summarizes the intersection characteristics between each pair of adjacent nodes to obtain spatial overlap relationship data.

[0078] Extract the 3D bounding region boundary values ​​of the current node, using its starting and ending coordinates along the X, Y, and Z axes as spatial intervals. Then, compare the boundary data of all adjacent nodes within the same coordinate range. For each node pair, perform a coordinate range comparison operation to determine if there exists a situation where the starting value is less than the ending value of the other node and the ending value is greater than the starting value of the other node in any of the X, Y, or Z dimensions. If all three axes meet this condition, it is considered a spatial overlap relationship, and the matching node pairs are added to the overlap analysis queue. Subsequently, extract the overlapping segment of the node pair in 3D space. By taking the maximum starting coordinate and minimum ending coordinate of the intersection interval, the spatial coordinate range of the overlapping cube is obtained. Simultaneously, a 3D overlapping region image is constructed, and a node overlap heatmap is built through pixelation. This image is used as one of the input images for the CNN. Geometric feature values ​​of the overlapping edge regions are extracted using convolutional kernels to determine the severity of boundary overlap. In the response map of the CNN, edge segments with prominent response values ​​are marked as major conflict areas. If the pixel area of ​​the high response area exceeds 30% of the total overlapping image area, it is recorded as a severely overlapping pair. In practical applications, when the boundary of node A is X: 200-600, Y: 300-700, Z: 100-400, and the boundary of node B is X: 500-800, Y: 500-850, Z: 200-450, the X-axis overlap range of the two nodes is 500-600, the Y-axis is 500-700, and the Z-axis is 200-400. The coordinate range of the overlapping volume is X: 500-600, Y: 500-700, Z: 200-400. After the overlapping spatial voxel image is processed by the CNN, the response is concentrated in the lower boundary area of ​​the Z-axis. The coordinates of this intersection are marked and written into the node overlap information set to form the spatial overlap relationship data between the current node and its adjacent nodes.

[0079] The directional relationship judgment submodule filters the enclosure parameters of each spatially related device within the control cabinet structure based on spatial overlap relationship data, analyzes the spatial coverage of the enclosed area of ​​each node, compares the arrangement direction between the enclosed areas of the nodes, judges the correspondence between spatial coverage and direction, and obtains the spatial coverage direction parameters.

[0080] First, for each identified node pair, spatial affiliation within the control cabinet structure is filtered, extracting the cabinet number to which the node belongs. Then, it is checked whether the coordinates of the node's location fall within the physical dimensions of the same cabinet. For node combinations within the same cabinet, they are recorded as spatially associated devices. Next, the boundary coordinates of the areas enclosed by each node are extracted, and a spatial direction vector map is constructed. A node direction reference vector is built by connecting the node center points. This vector is compared with the cabinet layout direction vector by angle difference. If the angle between the two vectors is less than 20 degrees, they are considered to be in the same direction; if the angle is between 20 and 70 degrees, they are considered to be parallel offset; and if the angle is greater than 70 degrees, they are considered to be opposing layouts. Simultaneously, the spatial arrangement image extracted from the CNN feature map is also considered. Identify the node relationships on the high-response linear arrangement path. If the convolutional layer response is continuously distributed in the same directional channel and the response intensity is stable, then confirm that the arrangement direction is an effective continuous layout path. Then analyze the angle between the projection direction of the overlapping area formed by the node and the arrangement direction. If the directions overlap, the relationship is recorded as "same-direction overlap"; otherwise, it is recorded as "opposite-direction overlap". For example, if the center point of node A is (300, 500, 100) and node B is (500, 500, 100), the connection vector is in the X-axis direction, the cabinet direction is parallel to the X-axis direction, and the angle between the two is 0 degrees, then the directions are consistent. If the overlapping area also falls in the X-axis direction, then a same-direction overlap relationship is formed. All such data are recorded in the spatial coverage direction parameter set.

[0081] The overlap feature comparison submodule compares the area and volume distribution of the spatial intersection of each node based on the spatial coverage direction parameter, analyzes the spatial overlap features within the area surrounded by the nodes, summarizes the key parameters of each spatial intersection, determines the node number and corresponding area where spatial conflict occurs, and obtains the node spatial conflict identifier.

[0082] First, the intersection coordinate range of each pair of spatially overlapping nodes is retrieved. Boundary difference calculations are then performed on the intersection spatial volume. The length, width, and height of the overlapping region are calculated as the basis for volume determination. For area distribution, a two-dimensional projection map is constructed, and pixelation mapping is performed on the XY, YZ, and XZ planes to build a two-dimensional planar coverage map of the intersection. The area range of each projection plane is statistically analyzed. Intersection regions with an area range below 50,000 square millimeters are marked as slight overlap, those between 50,000 and 150,000 square millimeters are marked as moderate overlap, and those exceeding 150,000 square millimeters are marked as severe overlap. In actual deployment, if an intersection is found with X: 500-700, Y: 600-800, and Z: 200-350, the overlapping volume is 200. ×200×150=6,000,000 cubic millimeters, corresponding to area projections of 40,000 square millimeters, 36,000 square millimeters, and 30,000 square millimeters, which are determined to be moderately overlapping regions. At the same time, the intersection projection map is fed into a CNN for edge response analysis. The density of overlapping boundary feature points is identified in the CNN convolution output image. If the density of edge response points continuously exceeds 50% of the image boundary, the node pair is recorded as a high-intensity overlap. Subsequently, the node number, intersection spatial coordinates, volume level, orientation consistency label, and CNN response intensity index in the overlapping feature data are combined to form a spatial conflict label entry. The node number is then bound to the conflict label result set to complete the generation of node spatial conflict labels.

[0083] like Figure 2 and Figure 6 As shown, the target patch selection module includes:

[0084] The patch filtering submodule filters patches that do not overlap spatially based on node spatial conflict identifiers, analyzes the overlap between each patch and the spatial boundary of adjacent devices, determines the independent boundary of the patch in three-dimensional space, and removes patches that overlap spatially or have intersecting boundaries to obtain an independent patch parameter set.

[0085] First, extract the node numbers of all nodes marked as not having spatial overlap, and retrieve their associated patch sets. Then, sequentially read the boundary coordinates of each patch, registering the minimum and maximum coordinates of the six boundaries of each patch in the X, Y, and Z directions as boundary values. Next, perform a cross-comparison of the coordinate ranges of the spatial boundaries of adjacent devices, determining whether the patch boundary falls within the bounded boundary range of adjacent devices on any axis. If boundary values ​​overlap in any axis direction, the patch is marked as a potentially overlapping patch. Further analysis is then performed to determine if the patch exhibits boundary overlap behavior. By analyzing the convolution response maps of the boundary envelope and other patch boundaries on the spatial image, the patch boundary map is used as the input image and fed into the CN (Convolutional Language Processing) system. The N network extracts edge features, using three convolutional layers to extract abrupt change regions at the corners. If the boundary of a facet in the CNN response image shows a high-frequency cross-response at multiple points, it is marked as an invalid facet with an interlacing phenomenon and removed. In a practical example, facet A has boundaries X: 100-400, Y: 200-500, Z: 50-150, and facet B has boundaries X: 350-600, Y: 480-700, Z: 100-200. The two facets overlap in the X-axis range of 350-400, the Y-axis range of 480-500, and the Z-axis range of 100-150. They are determined to have overlapping boundaries on the three axes, so facet A is excluded from the candidate facet set. The remaining unmarked facets are indexed and numbered to form an independent facet parameter set.

[0086] The directional consistency analysis submodule analyzes the connection direction between all patches and upstream nodes based on the independent patch parameter set, compares the spatial consistency of the connection between the patch center point and the upstream node center, determines the correspondence between the spatial arrangement of patches and the route direction, identifies patches with consistent arrangement direction, and obtains the directional consistency distribution group.

[0087] First, the 3D coordinates of the center point of each patch are extracted one by one, and the upstream logical connection node number of its corresponding node is retrieved simultaneously. Then, the center point coordinates of the upstream node are extracted to construct a spatial connection vector between the patch and the upstream node. Subsequently, the connection vector is processed by three-axis interpolation to generate a set of patch direction vectors. Then, the cable path direction field in the loop to which the current node belongs is read to generate a logical wiring direction vector. The logical direction vector and the physical patch direction vector are compared by direction angle calculation to obtain the spatial difference angle between the two vectors. If the absolute value of the angle is less than 15 degrees, it is determined that the direction is consistent; 15 to 45 degrees is considered slightly consistent; and greater than 45 degrees is considered that the direction is deviated. At the same time, a patch center direction image is constructed and input into CN. In the N-convolutional structure, directional channel activation detection is performed. The CNN analyzes multiple receptive fields and outputs a heatmap of the orientation of the face. If the orientation channel corresponding to the face shows a continuous activation region in the heatmap and the corresponding upstream node has a consistent orientation response channel, then the orientation of the face is marked as consistent with the orientation of the upstream node. In actual deployment, if the center point of the face is (300, 400, 100) and the center point of the upstream node is (100, 400, 100), and the logical direction vector points to the positive X-axis direction, then the angle between the two is 0 degrees, which meets the orientation consistency condition. In the CNN output response map, a strong X-direction response channel can be observed at the same time, which confirms that it meets the orientation consistency condition. All face numbers and spatial coordinates that meet the orientation consistency condition are written into the orientation consistency distribution group.

[0088] The spatial difference comparison submodule is based on the directional consistency distribution group. It compares the spatial distance from the center point of each patch to the center of the busbar, analyzes the adaptation of each patch boundary to the standard size, judges the degree of matching between the patch and the design size, identifies patches with small spatial differences and consistent size, and obtains the node projection center parameters.

[0089] First, the 3D coordinates of the center point of each patch and the coordinates of the center point of the generatrix are extracted. The differences in the X, Y, and Z directions between the two points are recorded separately. Then, the sum of the absolute values ​​of each difference is calculated as a spatial distance index. Next, the boundary dimensions of the current patch are read, the long and short side dimension fields are extracted, and the differences are compared with the preset standard equipment dimension values. If the difference between the patch boundary and the standard dimension in both the length and width directions is less than 20mm, the patch is considered to have a high degree of size fit. If the difference exceeds 50mm, it is judged to deviate from the standard. The size fit range is set as 0 to 20mm for a perfect match, 20 to 50mm for a partial fit, and more than 50mm for a misfit. Based on this, a spatial difference analysis diagram is constructed to reflect the distance information between the patch boundary and the generatrix. The projection is a two-dimensional matrix input into the CNN network, where feature activation detection is performed to identify the matching degree output map of the center region of the bus response. If the center point of a patch in the CNN output map falls in the high-frequency region of the bus response and its boundary projection has an overlap rate of more than 80% with the response of the standard size contour map, then the patch is determined to be a matching patch. In the example, the center of a patch is (400, 200, 100), the center of the bus is (420, 220, 100), the difference between each axis is no more than 30mm, the size is 620×400, the difference between the patch and the standard size 600×400 is 20mm in the long side direction and 0 in the wide side direction. It is registered as a candidate patch with small spatial difference and matching size. The number, center coordinates, and difference data of this type of patch are output and summarized to form the node projection center parameters.

[0090] like Figure 2 and Figure 7 As shown, the modeling instruction output module includes:

[0091] The spatial coordinate analysis submodule determines the geometric relationship between the three-dimensional spatial coordinates of each node and the origin of the modeling space based on the node projection center parameters, compares the distribution order of the center points of each node patch in space, summarizes the spatial arrangement state, and obtains the coordinate arrangement parameters.

[0092] Extract the 3D spatial coordinates of each node, obtaining its X, Y, and Z coordinate values. Calculate the coordinate difference between these values ​​and the reference origin coordinates (0, 0, 0) in the modeling space to obtain the spatial displacement distance of each node along each coordinate axis. Then, sort the nodes by X-axis difference from smallest to largest to form a horizontal arrangement sequence, by Y-axis difference to form a depth arrangement sequence, and by Z-axis to form a hierarchical arrangement sequence. Next, compare the sorted node indices with the original layout order to mark the relative arrangement level of each node in space. Subsequently, analyze the coordinates of all nodes one by one against the coordinates of their center patch. For nodes with a deviation exceeding 100mm, record their offset and retrieve the layout direction information of such nodes, determining the angle difference between the node's orientation and the patch's pointing direction. When the angle exceeds 45 degrees, it is recorded as "uncoordinated arrangement state". If the angle is less than 15 degrees and the spatial distance is less than 300mm, it is recorded as "priority arrangement node". Then, all node spatial projection points are visualized and mapped in the modeling coordinate system to construct an arrangement distribution map. This arrangement map is fed into a convolutional neural network to extract spatial density and layout path features. The convolutional kernel in the CNN network identifies the concentrated and sparse areas of the arrangement map. The effectiveness of node ranking is analyzed by the difference in response intensity of each node in the response map. If a node ranks high in the dense response area but is actually placed low in the actual layout order, it is recorded as a "candidate node for order adjustment". The ranking number, spatial arrangement category, three-axis difference from the origin, and offset value from the surface are recorded as coordinate arrangement parameters.

[0093] The template adaptation judgment submodule compares spatial coordinates with device function categories based on coordinate layout parameters, analyzes the structural information of the 3D template corresponding to each node, judges the adaptability between the template and spatial coordinates, identifies the node index that matches the spatial coordinates and template structure, and obtains the template adaptation index group.

[0094] Extract the spatial coordinates of each node and its corresponding device function category code. Compare the response intensity level of the area where the node coordinates are located in the layout diagram with the spatial layout requirements of the corresponding template in the standard template library. Call the structural fields such as template size, interface direction, and device mounting surface, and check the alignment relationship between the corresponding coordinate point and the template structure item by item to determine whether the current node meets the three-dimensional position conditions required by the template. The judgment criteria are: whether the area where the coordinates are located has the template mounting direction, whether the surface connected to the coordinate point meets the template interface boundary, and whether the space size is sufficient to place the template device outline. Label the above three judgments. If all three are met, it is recorded as "fully adapted". If two are met and one deviation is within the threshold of 50mm, it is recorded as "partially adapted". If any item is not met or the deviation exceeds 100mm, it is marked as "incompatible". In actual deployment, if the node's X coordinate is 200, the function category is "PRT", the corresponding template structure is a long-side wall-mounted type with the interface facing down, the installation panel width requirement is 400mm, the current actual panel width is 420mm, and the interface is located at the center of the lower boundary of the panel, then this coordinate position meets the template structure boundary requirements, and it is identified as an compatible node. An index relationship is established between this coordinate and the "PRT" category template. At the same time, the response intensity of the node's position is compared with the response of other nodes of the same type in the layout diagram image identified by CNN. If the response intensity of this node is in the top 30%, it is assigned a high compatibility level label and written into the index table to generate a template compatibility index group.

[0095] The data instruction generation submodule is based on the template adaptation index group, combines the three-dimensional spatial coordinates of nodes, the center point information of patches and the equipment model information, calculates the arrangement order of each node in three-dimensional space, adjusts the correspondence between equipment type and spatial coordinates, and outputs the parameters required for spatial layout to obtain the three-dimensional model layout instruction.

[0096] Extract the 3D spatial coordinates, center patch coordinates, node number, and template category fields for each node. Retrieve the 3D equipment structure model matching the node category from the template library, and construct a model call path and placement parameter comparison table. Then, sort the nodes according to the placement order field, assign numbers to all compatible nodes sequentially, and repeatedly verify the spatial positions of adjacent nodes. If the 3D spatial distance between adjacent nodes is less than 200mm, reverse the placement order. Next, categorize the device types of the nodes in the current space, and calculate the number of devices of each type and their average X, Y, and Z coordinates. Use this result to assist in subsequent spatial block placement. Finally, compare the center coordinates of each node's patch with the reference points of the template model. The system performs line matching and sets the matching point as the device insertion anchor point. The X, Y, and Z values ​​of the nodes in the 3D spatial coordinates are written into the deployment parameter field, and the template number and device category are written into the model type field. The generated number, device category, template call path, 3D spatial coordinates, center patch coordinates, and deployment sequence number are structurally combined to construct the device deployment data instruction set. During this process, the above instruction set is fed into the CNN deployment feedback network structure. The generated image is detected by the convolutional layer to verify whether there are duplicate anchor points or overlapping responses in the template insertion area in the device deployment image. If the overlap rate of the response at a certain position in the CNN response image exceeds 80%, the coordinates of that node are offset and the deployment coordinates are regenerated. The 3D model deployment instructions for all nodes are output.

[0097] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks, characterized in that, The system includes: The structural attribute mapping module is based on the substation secondary circuit. It analyzes the relationship between the node logical number and the bus number, compares the functional category code with the equipment size parameters, determines the type of equipment associated with each node, maps it to the node number, and obtains the node structural parameter set. Based on the node structure parameter set, the spatial layout generation module determines the geometric relationship between the device size and the spatial coordinate origin, calculates the three-dimensional spatial bounding boundary of each node device, and obtains the spatial layout envelope set. The spatial conflict screening module analyzes the overlap between the current node and the area enclosed by the adjacent node based on the spatial layout envelope set, determines the spatial coverage and directional relationship of the enclosed area, compares the spatial overlap characteristics, and obtains the node spatial conflict identifier. The target patch selection module filters out non-overlapping patches based on the node spatial conflict identifier, analyzes the directional consistency of the patch and the upstream node connection, judges the patch boundary and standard size matching, and obtains the node projection center parameters. The modeling instruction output module analyzes the compatibility between spatial coordinates and functional category templates based on the node projection center parameters, combines the patch center coordinates and equipment model information, and obtains the three-dimensional model layout instruction.

2. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The node structure parameter set includes category identifier, shape data and interface information; the spatial layout envelope set includes position coordinates, spatial volume and attitude information; the node spatial conflict identifier includes conflict location, conflict type and association number; the node projection center parameters include three-dimensional position, patch number and relative distance to the generatrix; and the three-dimensional model deployment instructions include deployment order, model type and coordinate parameters.

3. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The structural attribute mapping module includes: The logic mapping construction submodule is based on the substation secondary circuit. It analyzes the field content in the node logic number and the bus number, compares the attribution relationship of each group of numbers in the character structure, and determines whether the circuit unit in the logic number is consistent with the main line identifier of the bus number, thus obtaining the attribution mapping dataset. The function code comparison submodule analyzes the category definition and structural fields in the standard equipment size parameters based on the attribution mapping dataset, identifies the corresponding combinations between the function category code and the size parameters, determines whether each combination meets the classification conditions, and obtains the function size matching set. The structural attribute organization submodule analyzes the category, shape, and interface parameters of the node-associated devices based on the functional size matching set, adjusts the classification method of various structural information, and classifies and organizes the node device structural data to obtain the node structural parameter set.

4. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The spatial layout generation module includes: The size relationship judgment submodule compares the relative positions of each node device with the spatial coordinate origin based on the node structure parameter set, determines the distribution of the three-dimensional size boundary of the device in the spatial coordinate system, and optimizes the initial spatial layout of the nodes by combining the geometric distance relationship between the device center point and the coordinate origin, thus obtaining the spatial association parameter set. Based on the spatial association parameter set, the spatial boundary calculation submodule calculates the three-dimensional enclosing boundary of each node device, analyzes the boundary distribution range of the node on the coordinate axis, adjusts the center coordinates of the spatial enclosing area of ​​the node, optimizes the boundary limit distribution, determines the boundary attributes of the node in space, and obtains the three-dimensional boundary parameter set. Based on the three-dimensional boundary parameter set, the layout information summarization submodule adjusts the arrangement order of the enclosing regions of each node in the spatial coordinate system, optimizes the spatial distribution between devices, and integrates the spatial enclosing information and layout mapping parameters of the node devices to obtain a spatial layout envelope set.

5. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The space conflict screening module includes: The spatial overlap analysis submodule, based on the spatial layout envelope set, compares the overlap of the enclosing regions of the current node and the spatially adjacent nodes in the three-dimensional coordinate system, determines the overlapping position of the node enclosing boundaries, identifies the coordinate range of the spatially overlapping part, and summarizes the intersection characteristics between each pair of adjacent nodes to obtain spatial overlap relationship data. The directional relationship judgment submodule, based on the spatial overlap relationship data, filters the enclosure parameters of each spatially related device in the control cabinet structure, analyzes the spatial coverage of the enclosed area of ​​each node, compares the arrangement direction between the enclosed areas of the nodes, judges the correspondence between spatial coverage and direction, and obtains the spatial coverage direction parameters. The overlap feature comparison submodule compares the area and volume distribution of the spatial intersection of each node based on the spatial coverage direction parameters, analyzes the spatial overlap features within the area surrounded by the nodes, summarizes the key parameters of each spatial intersection, determines the node number and corresponding area where a spatial conflict occurs, and obtains the node spatial conflict identifier.

6. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The target patch selection module includes: The patch filtering submodule filters patches that do not overlap spatially based on the node spatial conflict identifier, analyzes the overlap between each patch and the spatial boundary of adjacent devices, determines the independent boundary of the patch in three-dimensional space, and removes patches that overlap spatially or have intersecting boundaries to obtain an independent patch parameter set. The directional consistency analysis submodule analyzes the connection direction between all patches and upstream nodes based on the independent patch parameter set, compares the spatial consistency of the connection between the patch center point and the upstream node center, determines the correspondence between the spatial arrangement of patches and the route direction, identifies patches with consistent arrangement directions, and obtains directional consistency distribution groups. The spatial difference comparison submodule compares the spatial distance from the center point of each patch to the center of the busbar based on the directional consistency distribution group, analyzes the adaptation of each patch boundary to the standard size, judges the degree of matching between the patch and the design size, identifies patches with small spatial differences and matching sizes, and obtains the node projection center parameters.

7. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The modeling instruction output module includes: Based on the node projection center parameters, the spatial coordinate analysis submodule determines the geometric relationship between the three-dimensional spatial coordinates of each node and the origin of the modeling space, compares the distribution order of the center points of each node patch in space, summarizes the spatial arrangement state, and obtains the coordinate arrangement parameters. The template adaptation judgment submodule compares the spatial coordinates with the device function category based on the coordinate layout parameters, analyzes the structural information of the three-dimensional template corresponding to each node, judges the adaptability between the template and the spatial coordinates, identifies the node index that matches the spatial coordinates and the template structure, and obtains the template adaptation index group. The data instruction generation submodule, based on the template adaptation index group, combines the three-dimensional spatial coordinates of the nodes, the center point information of the facets, and the equipment model information, calculates the arrangement order of each node in the three-dimensional space, adjusts the correspondence between the equipment type and the spatial coordinates, and outputs the parameters required for spatial layout to obtain the three-dimensional model layout instruction.

8. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The substation secondary circuit refers to the logical circuit of each signal and control path in the substation secondary system, including the lines for measurement and control, protection and automation functions. The node logical number refers to the number used to identify the functional unit in each circuit. The bus number refers to the unique number representing the main electrical bus of the substation, which determines the main circuit affiliation of each secondary device in the system.

9. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The origin of the spatial coordinates refers to the reference point of the three-dimensional modeling space, which is the starting point set for modeling. The three-dimensional spatial enclosing boundary refers to the smallest envelope of the space occupied by the device in the three-dimensional coordinate system.

10. The three-dimensional panoramic visualization modeling system for substation secondary circuits using convolutional neural networks as described in claim 1, characterized in that, The current node refers to the secondary equipment or loop node that is being analyzed or is about to be deployed, and the overlapping feature refers to the area, volume, or overlapping shape of the spatial intersection.

Images