High-voltage switch design drawing semantic recognition method and device, and storage medium

Abstract

The application provides a high-voltage switch design drawing semantic recognition method and device and a storage medium, and particularly relates to the technical field of mechanical design drawing recognition and auditing. The method comprises the following steps: acquiring drawing data corresponding to a high-voltage switch design drawing; analyzing the drawing data to obtain geometric primitives corresponding to the high-voltage switch in the drawing data and primitive information; regarding each geometric primitive as an independent node and constructing node features corresponding to each node based on the primitive information; constructing an edge between two nodes having a connection relationship based on the primitive coordinates; obtaining a local feature map corresponding to each edge based on the node features of the two end nodes of each edge; constructing a geometric topology graph corresponding to the high-voltage switch based on the local feature map and the node features; and taking the geometric topology graph as input data of a preset graph neural network model to obtain semantic information of each geometric primitive. The accuracy and completeness of the semantic understanding of the geometric primitive are improved.

Description

Technical Field

[0001] This application relates to the field of mechanical design drawing recognition and review, and in particular to a semantic recognition method, device and storage medium for high voltage switch design drawings. Background Technology

[0002] As a critical piece of equipment in power systems, the review of high-voltage switchgear design drawings is crucial for preventing design flaws and avoiding rework later. Compared to general electrical drawings, high-voltage switchgear design drawings are extremely complex, typically consisting of dozens of large components and hundreds of smaller components. Traditional drawing review relies heavily on manual experience, requiring senior designers to visually inspect hundreds of checkpoints, perform secondary calculations, and conduct multiple checks. Figure 1 Consistency comparison. This process not only consumes a lot of design resources, but is also very susceptible to errors or omissions due to human negligence, affecting the quality and efficiency of the review.

[0003] Against this backdrop, automated drawing review has become a technological hotspot in the field of drawing design. Semantic understanding of drawings is a key step in achieving automated review, aiming to assign semantic information such as connection relationships, structural features, and component affiliations to various geometric elements (such as lines and arcs) in drawings. High-voltage switchgear design drawings are characterized by numerous components, complex dimensioning, and dense information, leading to insufficient accuracy and completeness in the semantic understanding of geometric elements in existing automated drawing review models. Summary of the Invention

[0004] In view of this, this application provides a semantic recognition method, device and storage medium for high-voltage switch design drawings, which can improve the accuracy and completeness of semantic understanding of geometric elements corresponding to high-voltage switches.

[0005] The present application is described below from multiple aspects, and the implementation methods and beneficial effects of these aspects can be referred to each other.

[0006] Firstly, this application provides a semantic recognition method for high-voltage switch design drawings, including:

[0007] Obtain the drawing data corresponding to the high-voltage switch design drawings;

[0008] The drawing data is parsed to obtain the geometric elements corresponding to the high-voltage switch in the drawing data, as well as the element information corresponding to each geometric element. The element information includes at least the element coordinates.

[0009] Each geometric primitive is treated as an independent node, and node features corresponding to each node are constructed based on primitive information.

[0010] Based on primitive coordinates, construct an edge between two nodes that are connected.

[0011] Based on the node features of the two ends of each edge, a local feature map corresponding to each edge is obtained. The local feature map is used to characterize the spatial relationship between the two ends of the edge.

[0012] Based on local feature maps and node features, a geometric topology map corresponding to the high-voltage switch is constructed.

[0013] Using the geometric topology graph as input data for a pre-defined graph neural network model, semantic information of each geometric primitive is obtained.

[0014] According to this scheme, geometric primitives and primitive information containing primitive coordinates are obtained by parsing high-voltage switch design drawing data. Geometric primitives are used as nodes, and node features are constructed. Edges between nodes are built based on primitive coordinates, and local feature maps representing spatial relationships are generated based on the node features of the nodes at both ends of the edges. After constructing a geometric topology graph by combining node features and local feature maps, the graph is input into a graph neural network model to obtain primitive semantic information, which can completely capture the attributes of the geometric primitives themselves and the spatial relationships between primitives. This approach is suitable for the characteristics of high-voltage switch drawings, which have numerous components and dense drawing information, effectively improving the accuracy and completeness of geometric primitive semantic recognition.

[0015] In one possible implementation of the first aspect described above, a geometric topology map corresponding to the high-voltage switch is constructed based on local feature maps and node features, including:

[0016] Based on the local feature map corresponding to each edge and the node features of the two ends of each edge, spatial relationship reasoning is performed on the edges to obtain the relationship determination result;

[0017] The relation determination results are vectorized to obtain the edge features corresponding to the edges;

[0018] Based on nodes, node features, edges, and edge features, a geometric topology graph is obtained.

[0019] According to this scheme, by reasoning about the spatial relationships between nodes and vectorizing them into edge features, the constructed geometric topology graph contains both node attributes and quantified spatial relationship information, which improves the accuracy of the topology graph in representing the structure of high-voltage switch drawings and provides a more accurate and comprehensive input basis for subsequent semantic recognition.

[0020] In one possible implementation of the first aspect above, an edge between two nodes with a connection relationship is constructed based on primitive coordinates, including:

[0021] Calculate the distance between every two nodes based on the coordinates of the primitives;

[0022] If the distance is no greater than a preset threshold, it can be determined that there is a connection between two nodes, and then an edge can be constructed between the two nodes.

[0023] According to this scheme, the connection relationship between nodes is determined and edges are constructed by using a method based on the distance between primitive coordinates and a preset threshold. While ensuring the accuracy of connection relationship identification, the algorithm complexity is significantly reduced and the processing efficiency of edge construction is improved.

[0024] In one possible implementation of the first aspect above, a geometric topological graph is obtained based on nodes, node features, edges, and edge features, including:

[0025] By using nodes, node features, edges, and edge features as input data for a graph processing library, a geometric topological graph is obtained.

[0026] According to this scheme, a graph processing library is used to generate a geometric topology graph. By leveraging its natural adaptability to graph structures, a unified and efficient storage management of node, feature, and edge relationships is achieved. At the same time, the process of connecting with subsequent graph neural networks is simplified, and the efficiency of topology graph construction and application is improved.

[0027] In one possible implementation of the first aspect above, based on primitive information, node features corresponding to each node are constructed, including:

[0028] The primitive information is vectorized to obtain node features.

[0029] According to this scheme, node features are constructed by directly vectorizing primitive information, which realizes the standardized numerical conversion of unstructured primitive attributes, unifies the feature format of heterogeneous primitives, and provides efficient and reliable node-level input for subsequent topology construction and semantic recognition.

[0030] In one possible implementation of the first aspect above, spatial relationship reasoning is performed on the edges based on the local feature map corresponding to each edge and the node features of the two endpoints of each edge to obtain a relationship determination result, including:

[0031] The local feature map corresponding to each edge and the node features of the two ends of each edge are used as input data for a pre-defined multimodal model to obtain the relationship determination result.

[0032] According to this scheme, a multimodal model is used to fuse local feature maps and node features for spatial relationship reasoning. This can simultaneously utilize spatial visual association and primitive attribute constraint information, significantly improving the accuracy of complex spatial relationship determination and providing a reliable foundation for subsequent edge feature generation.

[0033] In one possible implementation of the first aspect above, the drawing data is parsed to obtain geometric elements corresponding to the high-voltage switch in the drawing data, and element information corresponding to each geometric element, including:

[0034] The drawing data is parsed using the first open-source library to obtain geometric primitives and primitive information;

[0035] Based on the node features of the two endpoints of each edge, a local feature map corresponding to each edge is obtained, including:

[0036] The node features of the two ends of each edge are used as input data for the second open-source library to obtain a local feature map.

[0037] According to the scheme, the first and second open-source libraries are used to complete the drawing parsing and local feature map generation respectively. Relying on mature open-source resources reduces the development difficulty and cost, improves the efficiency and stability of parsing and feature generation, and accelerates the engineering implementation of the technical solution.

[0038] Secondly, this application provides a semantic recognition device for high-voltage switch design drawings, comprising:

[0039] The data acquisition module is used to acquire drawing data corresponding to the high-voltage switch design drawings;

[0040] The parsing module is used to parse the drawing data to obtain the geometric elements corresponding to the high-voltage switch in the drawing data, as well as the element information corresponding to each geometric element. The element information includes at least the element coordinates.

[0041] The feature construction module is used to treat each geometric primitive as an independent node and construct node features corresponding to each node based on the primitive information.

[0042] The edge construction module is used to construct edges between two nodes that have a connection relationship, based on primitive coordinates.

[0043] The feature map drawing module is used to obtain a local feature map corresponding to each edge based on the node features of the two end nodes of each edge. The local feature map is used to represent the spatial relationship between the two end nodes of the edge.

[0044] The topology building module is used to construct a geometric topology map corresponding to the high-voltage switch based on local feature maps and node features;

[0045] The semantic understanding module is used to take the geometric topology graph as input data for a pre-defined graph neural network model and obtain the semantic information of each geometric primitive.

[0046] Thirdly, this application provides an electronic device including a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by the processor to implement the high-voltage switch design drawing semantic recognition method disclosed in the first aspect and any possible implementation thereof.

[0047] Fourthly, this application provides a computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the semantic recognition method for high-voltage switch design drawings disclosed in the first aspect and any possible implementation thereof.

[0048] Fifthly, this application provides a computer program product comprising: computer instructions that, when executed on an electronic device, cause the electronic device to perform the high-voltage switch design drawing semantic recognition method disclosed in the first aspect and any possible implementation thereof.

[0049] The beneficial effects of the second to fourth aspects can be found in the first aspect and the beneficial effects of any possible implementation of the first aspect, and will not be repeated here. Attached Figure Description

[0050] Figure 1 A flowchart for semantic recognition of pressure switch design drawings provided in this application embodiment;

[0051] Figure 2 This is a schematic diagram illustrating the geometric topology graph construction process provided in an embodiment of this application;

[0052] Figure 3 This is a diagram of the graph convolutional network model structure provided in the embodiments of this application;

[0053] Figure 4 This is a flowchart illustrating the construction of a geometric topology graph provided in an embodiment of this application.

[0054] Figure 5 This is a structural block diagram of the semantic recognition device provided in the embodiments of this application;

[0055] Figure 6 A block diagram of an electronic device provided in an embodiment of this application;

[0056] Figure 7 This is a block diagram of a SoC (system on chip) provided in an embodiment of this application.

[0057] Figure label:

[0058] 1000. Semantic recognition device for high-voltage switch design drawings; 1010. Data acquisition module; 1020. Parsing module; 1030. Feature construction module; 1040. Edge construction module; 1050. Feature graph drawing module; 1060. Topology construction module; 1070. Semantic understanding module. Detailed Implementation

[0059] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] The technical problems to be solved by the embodiments of this application will be described below.

[0061] As described in the background technology section above, semantic understanding of drawings is a core and crucial step in achieving automated review of high-voltage switchgear design drawings. Its core objective is to assign key semantic information such as connection relationships, structural features, and the components to which they belong to to various geometric elements in the drawings, including lines and arcs. However, due to the inherent characteristics of high-voltage switchgear design drawings—numerous components, complex dimensioning, and dense information—existing automated review models cannot effectively adapt to the complex scenarios of such drawings. They suffer from technical deficiencies in the accuracy and completeness of semantic understanding of geometric elements, making it difficult to support high-quality and efficient automated review work.

[0062] To address the aforementioned technical problems, this application provides a semantic recognition method for high-voltage switch design drawings. First, drawing data corresponding to the high-voltage switch design drawings is acquired. This data is then parsed to obtain the geometric primitives corresponding to the high-voltage switch and primitive information, including at least primitive coordinates. Next, each geometric primitive is treated as an independent node. Based on the primitive information, node features corresponding to each node are constructed, and edges are built between two nodes with a connection relationship based on the primitive coordinates. Then, local feature maps representing the spatial relationships between nodes are generated based on the node features of the nodes at both ends of each edge. These local feature maps and node features are combined to construct a geometric topology graph corresponding to the high-voltage switch. Finally, this geometric topology graph is input into a preset graph neural network model, and the semantic information of each geometric primitive is output.

[0063] This application embodiment constructs geometric primitives into a geometric topology graph that integrates local spatial relationships and inputs it into a graph neural network, thereby achieving automated and accurate recognition of semantic information of geometric primitives in complex design drawings of high-voltage switches, which significantly improves the accuracy, completeness and efficiency of automatic drawing review.

[0064] To better understand the semantic recognition method for high-voltage switch design drawings according to embodiments of this application, the following description is provided in conjunction with the appendix to the specification. Figures 1-4 The following is a detailed description of a semantic understanding method for high-voltage switch design drawings in the embodiments of this application.

[0065] refer to Figures 1-3 , Figure 1 This application illustrates a flowchart of semantic recognition of pressure switch design drawings in an embodiment of the present application. Figure 2 A schematic diagram of the geometric topology graph construction process in an embodiment of this application is shown. Figure 3 The diagram shows the structure of the graph convolutional network model in an embodiment of this application.

[0066] like Figure 1 As shown in the embodiment of this application, a semantic recognition method for high-voltage switch design drawings includes steps S10-S70.

[0067] S10 involves acquiring drawing data corresponding to the high-voltage switch design drawings.

[0068] It is understandable that the above drawing data can be CAD drawing data in DWG format.

[0069] S20: Analyze the drawing data to obtain the geometric elements corresponding to the high-voltage switch in the drawing data, as well as the element information corresponding to each geometric element.

[0070] It should be noted that the above-mentioned graphic element information may include other features such as graphic element coordinates, graphic element type, and color.

[0071] In some embodiments, the drawing data can be parsed by directly calling an open-source library (i.e., the first open-source library mentioned above) through an interface set in the computer program, so as to obtain the geometric elements corresponding to the high-voltage switch and the element information corresponding to the geometric elements.

[0072] For example, the drawing data can be parsed using the EZDXF or ODA open-source library to obtain the aforementioned geometric primitives and primitive information.

[0073] Taking a straight line primitive as an example, after processing by the above open-source library, the output primitive information is represented as follows: {"type": "LINE", "start": [0.0, 0.0], "end": [10.0, 10.0], "layer": "0", "color": "5", "linetype": "CONTINUOUS", "lineweight": "50"}, where type represents the primitive type, start and end represent the starting and ending coordinates of the line, layer represents the layer, color represents the color, linetype represents the line type, and lineweight represents the line weight.

[0074] S30 treats each geometric primitive as an independent node and constructs node features corresponding to each node based on the primitive information.

[0075] In some embodiments, node features can be obtained by vectorizing primitive information.

[0076] Understandably, by directly vectorizing primitive information to generate node features, a standardized conversion of unstructured primitive information in drawings to structured numerical features is achieved. This method can completely preserve the original attribute information of geometric primitives, while unifying the feature representation format of different types of geometric primitives and eliminating data format differences between heterogeneous primitive information.

[0077] Furthermore, the vectorization processing logic is simple and clear, with low computational complexity, and can efficiently complete the feature construction of large-scale primitives in high-voltage switch drawings, providing a unified and computable node-level data foundation for subsequent determination of inter-node connection relationships, edge feature generation, and semantic recognition processing of graph neural networks.

[0078] S40: Based on primitive coordinates, construct an edge between two nodes that have a connection relationship.

[0079] In some embodiments, constructing an edge between two nodes with a connection relationship based on primitive coordinates includes: calculating the distance between every two nodes based on primitive coordinates; if the distance is not greater than a preset threshold, it can be determined that there is a connection relationship between the two nodes, and then an edge is constructed between the two nodes.

[0080] It is understandable that establishing edges between two nodes with a connection relationship achieves a preliminary topological association between different geometric primitives, restores the basic connection structure of each component in the drawing, and enables discrete independent primitives to form a whole with an association relationship, providing a topological framework for subsequent spatial relationship reasoning.

[0081] S50: Based on the node features of the two endpoints of each edge, a local feature map corresponding to each edge is obtained.

[0082] It should be noted that the local feature map is used to represent the spatial relationship between the two ends of an edge.

[0083] In some embodiments, the node features of the two ends of each edge are used as input data to the open-source library (i.e., the second open-source library mentioned above) to obtain a local feature map.

[0084] For example, local feature maps can be constructed by calling the EZDXF or ODA open-source libraries through interfaces in computer programs.

[0085] Understandably, a local feature map is a type of local geometric topology subgraph data that is extracted and constructed from the original drawing using a CAD geometry engine (such as EZDXF or ODA) and represents the spatial relationship between the nodes at both ends of a single edge.

[0086] S60 constructs a geometric topology map corresponding to the high-voltage switch based on local feature maps and node features.

[0087] Understandably, by deeply integrating the individual attributes of graphic elements with the spatial relationships between them, a graph structure data that can comprehensively and accurately reflect the actual structure of high-voltage switch drawings can be constructed, perfectly adapting to the input requirements of graph neural network models.

[0088] like Figure 2 As shown, the schematic diagram of the geometric topology graph construction process includes a DWG format drawing on the left and a geometric topology graph on the right. The DWG format drawing on the left contains 19 geometric primitives, including lines and arcs, with the node set V = {v1, v2, ..., v19}. Among these 19 nodes, there are 28 pairs of nodes with connections, resulting in the edge set E = {e1, e2, ..., e28}.

[0089] After automatically generating the drawing based on the DWG format drawing through steps S10-S60, the geometric topology diagram on the right is obtained. Each node in the geometric topology diagram corresponds to a different geometric primitive. Furthermore, there is an edge between two nodes that are connected, i.e., a connecting line between the two nodes.

[0090] S70 uses the geometric topology graph as input data for a pre-defined graph neural network model to obtain the semantic information of each geometric primitive.

[0091] For example, a graph neural network can be one of a graph convolutional network (GCN), a graph attention network (GAT), a gated graph convolutional network (GatedGCN), or other graph neural networks.

[0092] like Figure 3 As shown, in some embodiments, a graph attention network is used as the graph neural network model described above. The graph attention network includes three graph attention layers, a fully connected layer, and an output layer.

[0093] The three graph attention layers are graph attention layer 1, graph attention layer 2, and graph attention layer 3. The output layer uses the Softmax function as the classification function. Taking the geometric topology graph as input, each geometric primitive carries initial geometric features. First, the features are aggregated layer by layer through the three graph attention layers. Each layer calculates neighbor weights through an attention mechanism and normalizes them using Softmax, gradually fusing the features of each primitive with the topological context within 1-hop, 2-hop, and 3-hop ranges. Then, a fully connected layer maps the aggregated geometric features to the semantic space, obtaining scores for each category. Finally, the Softmax output layer transforms the scores into node probability distributions, using the semantic category with the highest probability as the semantic information of that geometric primitive.

[0094] For example, the semantic categories mentioned above can be openings, grooves, contours, etc.

[0095] To adapt the graph attention network (GNN) for semantic recognition of high-voltage switchgear design drawings, it needs to be trained. First, the collected high-voltage switchgear design drawings are converted into geometric topological graphs. These graphs are then divided into training and testing sets at a ratio of 7:3 or 8:2. The GNN is then trained multiple times using the training set, and finally, its recognition accuracy is verified using the testing set. During training, optimizers (such as Adam or AdamW) can be used to optimize the GNN. Cross-entropy is chosen as the loss function, and mechanisms such as batch normalization, Dropout, L1 regularization, or L2 regularization are introduced to improve training performance. Detailed training procedures can be found in existing technologies and will not be elaborated upon here.

[0096] The trained graph attention network is applied to the reasoning process to identify primitive categories based on the input geometric topology graph, thereby achieving semantic understanding of the drawing.

[0097] The following is a detailed explanation of step S60.

[0098] refer to Figure 4 , Figure 4 A flowchart illustrating the geometric topology graph construction process in an embodiment of this application is shown.

[0099] like Figure 4 As shown in the embodiment of this application, a geometric topology map corresponding to the high-voltage switch is constructed based on the local feature map and node features, including steps S61-S63.

[0100] In S61, based on the local feature map corresponding to each edge and the node features of the two ends of each edge, spatial relationship reasoning is performed on the edges to obtain the relationship determination result.

[0101] In some embodiments, the local feature map corresponding to each edge and the node features of the two endpoints of each edge are used as input data for a preset multimodal model to obtain the relationship determination result.

[0102] For example, the aforementioned multimodal model can be a general model, a lightweight fine-tuning model, or a specialized model. The specific model selection depends on the complexity of the drawing and the inference performance, such as the BLIP-2 model, MobileCLIP, and visually enhanced graph attention networks. This paper does not limit the multimodal model.

[0103] S62, the relation determination result is vectorized to obtain the edge features corresponding to the edge.

[0104] It should be noted that the beneficial effects here can be found in the section on constructing node features above, and will not be repeated here.

[0105] S63, based on nodes, node features, edges, and edge features, obtains a geometric topology graph.

[0106] In some embodiments, nodes, node features, edges, and edge features are used as input data for a graph processing library to obtain a geometric topology graph.

[0107] For example, optional graph processing libraries include NetworkX, PyTorch Geometric, etc.

[0108] Understandably, by combining the local feature graphs corresponding to the edges with the node features of the two end nodes to perform spatial relationship reasoning, the actual spatial relationship between nodes can be accurately determined. Then, the relationship determination result is vectorized into edge features, so that the constructed geometric topology graph simultaneously contains quantitative information on the node's own attributes and the spatial relationship between nodes. This improves the representation accuracy of the geometric topology graph on the actual structure of the high-voltage switch drawing, and provides a more accurate and comprehensive input basis for the semantic recognition of the subsequent graph neural network.

[0109] Therefore, compared with the prior art, the semantic recognition method for high-voltage switch design drawings provided in this application, by constructing a geometric topology graph that integrates the attributes of geometric primitives with spatial relationships and combining it with a graph neural network to achieve a complete process of semantic recognition, not only effectively adapts to the complex scenarios of high-voltage switch drawings and significantly improves the accuracy and completeness of semantic recognition, but also greatly reduces development costs and algorithm complexity, unifies data formats, and improves the accuracy of spatial relationship determination and the representation accuracy of the topology graph by adopting mature open-source libraries, distance-based edge construction, primitive information vectorization, multimodal spatial relationship reasoning, and graph database-generated topology graphs.

[0110] refer to Figure 5 , Figure 5 A block diagram of the semantic recognition device in an embodiment of this application is shown.

[0111] like Figure 5 As shown in the embodiment of this application, a semantic recognition device 1000 for high-voltage switch design drawings includes: a data acquisition module 1010, a parsing module 1020, a feature construction module 1030, an edge construction module 1040, a feature map drawing module 1050, a topology construction module 1060, and a semantic understanding module 1070.

[0112] The data acquisition module 1010 is used to acquire drawing data corresponding to the high-voltage switch design drawings.

[0113] The parsing module 1020 is used to parse the drawing data to obtain the geometric elements corresponding to the high-voltage switch in the drawing data, as well as the element information corresponding to each geometric element. The element information includes at least the element coordinates.

[0114] The feature construction module 1030 is used to treat each geometric primitive as an independent node and construct node features corresponding to each node based on the primitive information.

[0115] Edge construction module 1040 is used to construct edges between two nodes with a connection relationship based on primitive coordinates;

[0116] The feature map drawing module 1050 is used to obtain a local feature map corresponding to each edge based on the node features of the two end nodes of each edge. The local feature map is used to characterize the spatial relationship between the two end nodes of the edge.

[0117] The topology building module 1060 is used to construct a geometric topology map corresponding to the high-voltage switch based on local feature maps and node features;

[0118] The semantic understanding module 1070 is used to take the geometric topology graph as input data for a preset graph neural network model and obtain the semantic information of each geometric primitive.

[0119] It should be noted that the specific functions and corresponding technical effects of the semantic recognition device for the high-voltage switch design drawings can be found in the above embodiments. Figures 1-4 The methods explained will not be repeated here.

[0120] Now for reference Figure 6 The diagram shown is a block diagram of an electronic device 1200 according to an embodiment of this application. The electronic device 1200 may include one or more processors (corresponding to...) coupled to a controller hub 1203. Figure 6 The first processor 1201 is described above. In at least one embodiment, the controller hub 1203 communicates with the first processor 1201 via a multi-branch bus such as a front-side bus (FSB), a point-to-point interface such as a quick path interconnect (QPI), or a similar connection. The first processor 1201 executes instructions that control general types of data processing operations. In one embodiment, the controller hub 1203 includes, but is not limited to, a graphics memory controller hub (GMCH) (not shown) and an input / output hub (IOH) (which may be on a separate chip) (not shown), wherein the GMCH includes memory and a graphics controller and is coupled to the IOH.

[0121] Electronic device 1200 may also include a coprocessor coupled to controller hub 1203 (corresponding to...) Figure 6 The first coprocessor 1202 and memory 1204 are integrated within the processor (as described in this application). Alternatively, one or both of the memory and GMCH can be integrated within the processor (as described in this application), with memory 1204 and the first coprocessor 1202 directly coupled to the first processor 1201 and the controller hub 1203, which is located on a single chip with the IOH. Memory 1204 can be, for example, dynamic random access memory (DRAM), phase change memory (PCM), or a combination of both. In one embodiment, the first coprocessor 1202 is a dedicated processor, such as a high-throughput MIC processor (many integerized core, MIC), a network or communication processor, a compression engine, a graphics processor, a general-purpose computing on GPU (GPGPU), or an embedded processor, etc. Optional properties of the first coprocessor 1202 are indicated by dashed lines. Figure 6 middle.

[0122] As a computer-readable storage medium, memory 1204 may include one or more tangible, non-transitory computer-readable media for storing data and / or instructions. For example, memory 1204 may include any suitable non-volatile memory such as flash memory and / or any suitable non-volatile storage device such as one or more hard-disk drives (HDDs), one or more compact disc (CD) drives, and / or one or more digital versatile disc (DVD) drives.

[0123] In one embodiment, electronic device 1200 may further include a network interface controller (NIC) 1206. Network interface 1206 may include a transceiver for providing a radio interface for electronic device 1200 to communicate with any other suitable device, such as a front-end module, antenna, etc. In various embodiments, network interface 1206 may be integrated with other components of electronic device 1200. Network interface 1206 can implement the functions of the communication unit in the above embodiments.

[0124] Electronic device 1200 may further include input / output (I / O) device 1205. I / O device 1205 may include: a user interface designed to enable a user to interact with electronic device 1200; a peripheral component interface designed to enable peripheral components to also interact with electronic device 1200; and / or sensors designed to determine environmental conditions and / or location information related to electronic device 1200.

[0125] It is worth noting that, Figure 6 This is merely an example. That is, although... Figure 6 The electronic device 1200 shown includes multiple devices such as a first processor 1201, a first coprocessor 1202, a controller hub 1203, and a memory 1204. However, in actual applications, devices using the methods of this application may include only a portion of the devices in the electronic device 1200. For example, it may include only the first processor 1201 and the network interface 1206. Figure 6 The properties of the optional devices are shown in dashed lines. According to some embodiments of this application, the memory 1204, which is a computer-readable storage medium, stores instructions that, when executed on a computer, cause the electronic device 1200 to perform the high-voltage switch design drawing semantic recognition method according to the above embodiments. For details, please refer to the method in the above embodiments, which will not be repeated here.

[0126] Now for reference Figure 7 The diagram shown is a block diagram of a SoC (system on chip) 1300 according to an embodiment of this application. Figure 7 In the diagram, similar components share the same reference numerals. Additionally, dashed boxes are an optional feature for more advanced SoCs. Figure 7 In the SoC1300, interconnect unit 1350 is coupled to the processor (corresponding to...). Figure 7 The system includes a second processor 1310, a system agent unit 1380, a bus controller unit 1390, an integrated memory controller unit 1340, and one or more coprocessors (corresponding to...). Figure 7 The second coprocessor 1320 may include integrated graphics logic, an image processor, an audio processor, and a video processor; a static random access memory (SRAM) unit 1330; and a direct memory access (DMA) unit 1360. In one embodiment, the second coprocessor 1320 includes a dedicated processor, such as a network or communication processor, a compression engine, a GPGPU, a high-throughput MIC processor, or an embedded processor.

[0127] The static random access memory (SRAM) cell 1330 may include one or more computer-readable media for storing data and / or instructions. The computer-readable storage medium may store instructions, specifically, temporary and permanent copies of those instructions. These instructions may include, when executed by at least one unit in the processor, causing the SoC 1300 to perform the high-voltage switch design drawing semantic recognition method according to the above embodiments, as detailed in the methods described above, which will not be repeated here.

[0128] Various embodiments of the mechanisms disclosed in this application can be implemented in hardware, software, firmware, or combinations of these implementation methods. Embodiments of this application can be implemented as computer programs or program code executable on a programmable system, the programmable system including at least one processor, a storage system (including volatile and non-volatile memory and / or storage elements), at least one input device, and at least one output device.

[0129] Program code can be applied to input instructions to execute the functions described in this application and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, the processing system includes any system having a processor such as a digital signal processor (DSP), microcontroller, application-specific integrated circuit (ASIC), or microprocessor.

[0130] The program code can be implemented using a high-level procedural language or an object-oriented programming language to communicate with the processing system. Assembly language or machine language can also be used when needed. In fact, the mechanisms described in this application are not limited to any particular programming language. In either case, the language can be a compiled language or an interpreted language.

[0131] In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried or stored thereon on one or more temporary or non-temporary machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or through other computer-readable media. Therefore, machine-readable media may include any mechanism for storing or transmitting information in a machine-readable (e.g., computer-readable) form, including but not limited to floppy disks, optical disks, CD-ROMs, compact disc read-only memory (CD-ROMs), magneto-optical disks, read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic cards or optical cards, flash memory, or tangible machine-readable storage for transmitting information (e.g., carrier waves, infrared signals, digital signals, etc.) using the Internet in the form of electrical, optical, acoustic, or other forms of propagated signals. Therefore, machine-readable media include any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a machine-readable (e.g., computer-readable) form.

[0132] In the accompanying drawings, some structural or methodological features may be shown in a specific arrangement and / or order. However, it should be understood that such a specific arrangement and / or order may not be necessary. Rather, in some embodiments, these features may be arranged in a manner and / or order different from that shown in the accompanying drawings. Furthermore, including structural or methodological features in a particular figure does not imply that such features are required in all embodiments, and in some embodiments, these features may be omitted or may be combined with other features.

[0133] This application also provides a computer-readable storage medium, the specific functions and corresponding technical effects of which can be referred to the above embodiments. Figures 1-4 The methods explained will not be elaborated here.

[0134] This application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are loaded and executed by a processor, they implement the high-voltage switch design drawing semantic understanding method described in the above embodiments. The specific functions and corresponding technical effects can be found in the above embodiments. Figures 1-4 The methods explained will not be elaborated here.

[0135] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0136] It should be noted that all units / modules mentioned in the device embodiments of this application are logical units / modules. Physically, a logical unit / module can be a physical unit / module, a part of a physical unit / module, or a combination of multiple physical units / modules. The physical implementation of these logical units / modules themselves is not the most important factor; the combination of functions implemented by these logical units / modules is the key to solving the technical problems proposed in this application. Furthermore, to highlight the innovative aspects of this application, the above-described device embodiments of this application have not introduced units / modules that are not closely related to solving the technical problems proposed in this application. This does not mean that the above-described device embodiments do not contain other units / modules.

[0137] It should be noted that in the examples and description of this application, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0138] Although this application has been illustrated and described with reference to certain preferred embodiments thereof, those skilled in the art should understand that various changes in form and detail may be made thereto without departing from the spirit and scope of this application.

Claims

1. A semantic recognition method for high-voltage switch design drawings, characterized in that, include: Obtain the drawing data corresponding to the high-voltage switch design drawings; The drawing data is parsed to obtain geometric elements corresponding to the high-voltage switch in the drawing data, and element information corresponding to each geometric element, wherein the element information includes at least element coordinates; Each geometric primitive is treated as an independent node, and node features corresponding to each node are constructed based on the primitive information. Based on the coordinates of the primitives, construct an edge between two nodes that have a connection relationship; Based on the node features of the two end nodes of each edge, a local feature map corresponding to each edge is obtained, and the local feature map is used to characterize the spatial relationship between the two end nodes of the edge; Based on the local feature map and the node features, a geometric topology map corresponding to the high-voltage switch is constructed, including: Based on the local feature map corresponding to each edge and the node features of the two end nodes of each edge, spatial relationship reasoning is performed on the edges to obtain relationship determination results; The relationship determination result is vectorized to obtain the edge features corresponding to the edge; The geometric topology graph is obtained based on the nodes, node features, edges, and edge features; The geometric topology graph is used as input data for a preset graph neural network model to obtain the semantic information of each geometric primitive.

2. The method according to claim 1, characterized in that, The step of constructing an edge between two nodes with a connection relationship based on the primitive coordinates includes: Based on the coordinates of the primitives, calculate the distance between every two nodes; If the distance is not greater than a preset threshold, it can be determined that there is a connection between the two nodes, and then the edge can be constructed between the two nodes.

3. The method according to claim 1, characterized in that, The process of obtaining the geometric topology graph based on the nodes, node features, edges, and edge features includes: The nodes, node features, edges, and edge features are used as input data for a graph processing library to obtain the geometric topology graph.

4. The method according to claim 1, characterized in that, The step of constructing node features corresponding to each node based on the primitive information includes: The primitive information is vectorized to obtain the node features.

5. The method according to claim 1, characterized in that, The method of performing spatial relationship reasoning on the edges based on the local feature map corresponding to each edge and the node features of the two endpoints of each edge to obtain a relationship determination result includes: The relationship determination result is obtained by using the local feature map corresponding to each edge and the node features of the two end nodes of each edge as input data of a preset multimodal model.

6. The method according to claim 1, characterized in that, The step of parsing the drawing data to obtain the geometric elements corresponding to the high-voltage switch in the drawing data, and the element information corresponding to each geometric element, includes: The drawing data is parsed using the first open-source library to obtain the geometric primitives and their information; The process of obtaining a local feature map corresponding to each edge based on the node features of the two endpoints of each edge includes: The node features of the two ends of each edge are used as input data for the second open-source library to obtain the local feature map.

7. A semantic recognition device for high-voltage switch design drawings, characterized in that, include: The data acquisition module is used to acquire drawing data corresponding to the high-voltage switch design drawings; The parsing module is used to parse the drawing data to obtain geometric elements corresponding to the high-voltage switch in the drawing data, and element information corresponding to each geometric element, wherein the element information includes at least element coordinates; The feature construction module is used to treat each geometric primitive as an independent node and construct node features corresponding to each node based on the primitive information. An edge construction module is used to construct an edge between two nodes that have a connection relationship based on the coordinates of the primitive; The feature map drawing module is used to obtain a local feature map corresponding to each edge based on the node features of the two end nodes of each edge, and the local feature map is used to characterize the spatial relationship between the two end nodes of the edge; A topology construction module is used to construct a geometric topology map corresponding to the high-voltage switch based on the local feature map and the node features, including: Based on the local feature map corresponding to each edge and the node features of the two end nodes of each edge, spatial relationship reasoning is performed on the edges to obtain relationship determination results; The relationship determination result is vectorized to obtain the edge features corresponding to the edge; The geometric topology graph is obtained based on the nodes, node features, edges, and edge features; The semantic understanding module is used to take the geometric topology graph as input data for a preset graph neural network model to obtain the semantic information of each geometric primitive.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction or at least one program, which is loaded and executed by a processor to implement the semantic recognition method for high-voltage switch design drawings as described in any one of claims 1-6.

9. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the processor, they implement the semantic recognition method for high-voltage switch design drawings as described in any one of claims 1-6.

Images