A marine converter station water mist system pipe and attribute identification method thereof
By analyzing CAD drawings and constructing pipeline topology, combined with a graph neural network model, the problem of identifying densely intersecting pipelines in offshore converter stations was solved, achieving accurate matching and completion of pipeline attributes, and improving the accuracy and efficiency of identification.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- POWERCHINA HUADONG ENG CORP LTD
- Filing Date
- 2026-01-19
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies face significant challenges in identifying pipelines in offshore converter stations' fine water mist systems due to the unique nature of the scenarios, particularly the issues of densely intersecting pipelines and mismatched attribute labels.
By analyzing CAD drawings, denoising, constructing pipeline topology relationships, and using graph neural network models, combined with the consistency judgment of pipe diameter identification and pipeline direction, accurate matching and completion of pipeline attributes can be achieved.
It has enabled accurate identification of densely intersecting pipelines at offshore converter stations, improved the accuracy and reliability of attribute matching, reduced manual intervention, shortened the identification cycle, and improved the efficiency of digital modeling.
Smart Images
Figure CN122157296A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of digital and intelligent identification technology for engineering facilities, specifically to an automatic identification method for pipelines and their attributes in a fine water mist system of an offshore converter station, involving CAD drawing parsing, pipeline topology analysis, and graph-based data modeling and analysis technology. Background Technology
[0002] Offshore converter stations are the core hubs for offshore wind power grid connection, and fine water mist fire suppression systems are crucial for ensuring the safety of their equipment. Pipelines, as the core carriers for delivering the fire extinguishing medium, require precise identification of their shape and properties, which is fundamental to system design, construction, and operation and maintenance. In existing technologies, for conventional pipelines with simple layouts, CAD drawing analysis techniques (such as layer filtering and regular geometric primitive matching) or basic computer vision techniques can be used to extract pipeline locations and identify their contours, meeting the needs of low-complexity scenarios.
[0003] However, due to the limited space in the cabins of offshore converter stations, the fine water mist pipes are generally characterized by a large number of intersecting arrangements, and the attribute labeling is blurred due to the congestion of space and the overlap of pipes. The existing identification of fine water mist system pipes mainly relies on simple spatial distance to achieve automatic matching of attributes and pipes, which is prone to erroneous associations. Existing technologies do not consider the pipe topology relationship, while this invention solves the problem of identifying densely intersecting pipes by constructing a pipe topology network. Summary of the Invention
[0004] This disclosure aims to address the shortcomings of existing technologies in the identification of pipelines and their attributes in offshore converter station fine water mist systems due to the specific characteristics of the scenario, specifically including:
[0005] 1. Existing technologies rely on CAD drawing analysis or simple computer vision, and are only suitable for simple pipe recognition. Due to the dense and intersecting fine water mist pipes and the issue of jumpers, recognition becomes very difficult.
[0006] 2. Existing technologies match pipes with attribute labels using "simple spatial distance". However, the pipe spacing in the fine water mist system of an offshore converter station is extremely small, and the consistent line width of pipes of different diameters makes it difficult to identify pipe attributes, resulting in mismatched pipe labels.
[0007] This disclosure is achieved through the following technical solution:
[0008] A method for identifying the pipeline and attributes of a fine water mist system in an offshore converter station includes the following steps:
[0009] Step S1: Obtain the CAD drawings of the target fine water mist system;
[0010] Step S2: Parse the CAD drawing, extract the pipe layer and text layer, filter the text objects, identify the pipe diameter identification text used to characterize the pipe diameter, and obtain the corresponding text content and geometric parameters to form a pipe diameter identification set;
[0011] Step S3: Denoise the pipe primitives in the pipe layer, and establish pipe topology relationships based on the denoised pipe primitives to obtain an initial pipe topology structure containing node and pipe connection relationships.
[0012] Step S4: Based on the initial pipeline topology, verify the connection relationship between adjacent pipeline segments, and fill in the missing pipeline connections according to the continuity of pipeline direction and node spacing, and update the pipeline topology.
[0013] Step S5: Based on the pipe diameter identifier set and the updated pipe topology, match the pipe diameter identifier with the pipe segment according to the directional consistency and geometric distance between the pipe diameter identifier text and the pipe segment, and determine the pipe diameter attribute corresponding to each pipe segment in combination with the pipe diameter design rules of the fine water mist system.
[0014] Step S6: Write the determined pipe diameter attribute into the pipe topology structure, and complete the pipe segments without pipe diameter attribute according to the pipe diameter attribute of the adjacent pipe segments, and output a pipe attribute data table containing pipe segment identifier, start and end coordinates and pipe diameter attribute.
[0015] Further, step S3 includes the following sub-steps:
[0016] Step S301: Perform noise reduction processing on the pipe primitives in the pipe layer and delete the repeatedly drawn pipe segments. The repeatedly drawn pipe segments are pipe primitives with an overlap greater than a preset threshold.
[0017] Step S302: Based on the denoised pipeline primitives, establish pipeline topology relationships, specifically including: defining the start and end coordinates of each pipeline segment as topology nodes, defining the pipeline segments as valid connection edges between nodes; when two topology nodes belong to adjacent pipeline segments, the connection relationship between the nodes is defined as a potential connection edge.
[0018] Further, step S4 includes the following sub-steps:
[0019] Step S401: Based on the pipeline topology established in step S3, a graph neural network model is used to model the nodes and their connection relationships in the pipeline topology to construct an initial pipeline topology graph, wherein the pipeline topology graph includes a set of nodes and the connection relationships between nodes;
[0020] Step S402: Perform continuity judgment on the potential connection relationships in the initial pipeline topology diagram. Based on the continuity of the direction of adjacent pipeline segments and the geometric distance between corresponding nodes, identify the pipeline connections that need to be completed, and update the completed pipeline connections as valid connection relationships to improve the pipeline topology diagram.
[0021] Further, step S5 includes the following sub-steps:
[0022] Step S501: Perform a direction consistency judgment on the pipe diameter identifier text in the pipe diameter identifier set and each pipe segment in the pipe topology. The direction consistency judgment includes judging based on the angle between the text direction of the pipe diameter identifier text and the direction of the corresponding pipe segment. When the preset direction condition is met, it is determined that the pipe diameter identifier text and the pipe segment constitute a candidate matching relationship.
[0023] Step S502: For the candidate matching relationship, calculate the geometric distance between the pipe diameter identification text and the corresponding pipe segment, and filter the candidate matching relationship according to the preset distance conditions;
[0024] Step S503: For the filtered candidate matching relationships, based on the topological relationship between pipe segments in the pipe topology structure and combined with the pipe diameter design rules of the fine water mist system, perform logical verification to determine the optimal matching pipe segment corresponding to the pipe diameter identification text.
[0025] Further, step S6 includes the following sub-steps:
[0026] Step S601: For pipe segments with determined pipe diameter attributes, write the corresponding pipe diameter attributes into the pipe topology structure, and based on the pipe diameter attributes of adjacent pipe segments in the pipe topology structure, complete the pipe diameter attributes for pipe segments without marked pipe diameter attributes.
[0027] Step S602: Based on the pipeline topology structure with completed pipe diameter attribute annotation, generate a pipeline attribute data table containing pipeline segment identifiers, start and end coordinates, and pipe diameter attributes.
[0028] The beneficial effects of this disclosure are as follows: By denoising the pipe elements in CAD drawings and constructing the pipe topology, this disclosure achieves accurate identification of dense and intersecting pipes in the fine water mist system of an offshore converter station; by judging the consistency between pipe diameter identification and pipe direction, and by combining pipe topology relationships and pipe diameter design rules for multiple verifications, the accuracy and reliability of pipe attribute matching are effectively improved; at the same time, by adopting an automated pipe attribute identification process, manual intervention is reduced, the pipe attribute identification cycle is significantly shortened, and the efficiency of digital modeling of the fine water mist system is improved. Attached Figure Description
[0029] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0030] Specific embodiments of this disclosure are provided below. Those skilled in the art should understand that the embodiments are for illustrative purposes only and should not be construed as limiting the scope of this disclosure as defined by the claims in any way.
[0031] Figure 1 The flowchart of the method disclosed herein is as follows: Figure 1 As shown, a method for identifying the pipeline and attributes of a fine water mist system in an offshore converter station is disclosed, including the following steps:
[0032] Step S1: Obtain the dwg file of the CAD drawing of the target fine water mist system.
[0033] Step S2: Parse the CAD drawing, extract text objects from the pipe layer and text layer, obtain the text content and corresponding geometric information, and filter the text objects, retaining the pipe diameter identification text used to characterize the pipe diameter, such as DN25, DN80, and removing irrelevant text, forming a candidate pipe diameter identification set T={T1,T2,...,Tn}, where each This corresponds to a pipe diameter text and its geometric parameters that meet the conditions.
[0034] Step S301: Denoise the pipe elements in the pipe layer and delete the repeatedly drawn pipe segments, wherein the repeatedly drawn pipe segments are pipe elements with an overlap greater than a preset threshold.
[0035] Step S302: Based on the denoised pipeline primitives, mark the starting coordinates (Sx, Sy) and ending coordinates (Ex, Ey) of each pipeline segment, and establish the topological relationship between pipeline segments. Define the starting coordinates and ending coordinates as topological nodes, and define the pipeline segments as valid connection relationships between nodes. When two nodes belong to adjacent pipeline segments, define the corresponding connection relationship as a potential connection relationship, thereby forming the initial pipeline topology structure.
[0036] Step S401: Based on the initial pipeline topology, the GraphSAGE algorithm of the graph neural network model (GNN) is used to model the nodes and their connection relationships in the pipeline topology, and construct the initial pipeline topology graph G=(V,E), where V is the set of nodes and E is the set of edges.
[0037] Step S402: The continuity of potential connections in the initial pipeline topology diagram is determined. Based on the continuity of adjacent pipeline segments and the geometric distance between corresponding nodes, pipeline connections that need to be completed are identified, and the completed pipeline connections are updated as valid connections to improve the pipeline topology. In this embodiment, when the geometric distance between adjacent nodes is less than or equal to 500 mm, the corresponding potential connection is determined to be a pipeline connection that needs to be completed.
[0038] Step S501: Based on the candidate set of pipe diameter identifiers and the updated pipe topology, perform a geometric matching judgment on the pipe diameter identifier text and the pipe segment set P={P1,P2,...,Pn}. The geometric matching judgment includes directional consistency judgment and spatial proximity judgment. The directional consistency judgment includes judging based on the angle between the text direction of the pipe diameter identifier text and the direction of the corresponding pipe segment. When the angle is ≤5°, it is determined that the directions are consistent, and a candidate matching pair is obtained. ;
[0039] Step S502: For the candidate matching pairs retained in step S501 Further filtering is performed based on the geometric distance between the pipe diameter identification text and the corresponding pipe segment;
[0040] Step S503: For Geometry of adjacent pipes satisfying conditions S501 and S502 Call the updated pipeline topology map G=(V,E) in step S402, obtain the topological relationship of each pipeline segment, and perform logical verification in combination with the pipe diameter design logic of the fine water mist system to obtain the optimal matching pipeline.
[0041] Step S601: For each matched pipe segment, add the pipe diameter attribute to the node attributes of the topology graph G, and complete the nodes without pipe diameter attributes based on the pipe diameter attributes of adjacent nodes.
[0042] Step S602: Output a complete pipe attribute table consisting of pipe segment ID, start and end coordinates, and pipe diameter, to provide data support for the subsequent digitization of the fine water mist system.
Claims
1. A method for identifying the pipeline and attributes of a fine water mist system in an offshore converter station, characterized in that... Includes the following steps: Step S1: Obtain the CAD drawings of the target fine water mist system; Step S2: Parse the CAD drawing, extract the pipe layer and text layer, filter the text objects, identify the pipe diameter identification text used to characterize the pipe diameter, and obtain the corresponding text content and geometric parameters to form a pipe diameter identification set; Step S3: Denoise the pipe primitives in the pipe layer, and establish pipe topology relationships based on the denoised pipe primitives to obtain an initial pipe topology structure containing node and pipe connection relationships. Step S4: Based on the initial pipeline topology, verify the connection relationship between adjacent pipeline segments, and fill in the missing pipeline connections according to the continuity of pipeline direction and node spacing, and update the pipeline topology. Step S5: Based on the pipe diameter identifier set and the updated pipe topology, match the pipe diameter identifier with the pipe segment according to the directional consistency and geometric distance between the pipe diameter identifier text and the pipe segment, and determine the pipe diameter attribute corresponding to each pipe segment in combination with the pipe diameter design rules of the fine water mist system. Step S6: Write the determined pipe diameter attribute into the pipe topology structure, and complete the pipe segments without pipe diameter attribute according to the pipe diameter attribute of the adjacent pipe segments, and output a pipe attribute data table containing pipe segment identifier, start and end coordinates and pipe diameter attribute.
2. The method according to claim 1, characterized in that, Step S3 includes the following sub-steps: Step S301: Perform noise reduction processing on the pipe primitives in the pipe layer and delete the repeatedly drawn pipe segments. The repeatedly drawn pipe segments are pipe primitives with an overlap greater than a preset threshold. Step S302: Based on the denoised pipeline primitives, establish pipeline topology relationships, specifically including: defining the start and end coordinates of each pipeline segment as topology nodes, defining the pipeline segments as valid connection edges between nodes; when two topology nodes belong to adjacent pipeline segments, the connection relationship between the nodes is defined as a potential connection edge.
3. The method according to claim 1, characterized in that, Step S4 includes the following sub-steps: Step S401: Based on the pipeline topology established in step S3, a graph neural network model is used to model the nodes and their connection relationships in the pipeline topology to construct an initial pipeline topology graph, wherein the pipeline topology graph includes a set of nodes and the connection relationships between nodes; Step S402: Perform continuity judgment on the potential connection relationships in the initial pipeline topology diagram. Based on the continuity of the direction of adjacent pipeline segments and the geometric distance between corresponding nodes, identify the pipeline connections that need to be completed, and update the completed pipeline connections as valid connection relationships to improve the pipeline topology diagram.
4. The method according to claim 1, characterized in that, Step S5 includes the following sub-steps: Step S501: Perform a direction consistency judgment on the pipe diameter identifier text in the pipe diameter identifier set and each pipe segment in the pipe topology. The direction consistency judgment includes judging based on the angle between the text direction of the pipe diameter identifier text and the direction of the corresponding pipe segment. When the preset direction condition is met, it is determined that the pipe diameter identifier text and the pipe segment constitute a candidate matching relationship. Step S502: For the candidate matching relationship, calculate the geometric distance between the pipe diameter identification text and the corresponding pipe segment, and filter the candidate matching relationship according to the preset distance conditions; Step S503: For the filtered candidate matching relationships, based on the topological relationship between pipe segments in the pipe topology structure and combined with the pipe diameter design rules of the fine water mist system, perform logical verification to determine the optimal matching pipe segment corresponding to the pipe diameter identification text.
5. The method according to claim 1, characterized in that, Step S6 includes the following sub-steps: Step S601: For pipe segments with determined pipe diameter attributes, write the corresponding pipe diameter attributes into the pipe topology structure, and based on the pipe diameter attributes of adjacent pipe segments in the pipe topology structure, complete the pipe diameter attributes for pipe segments without marked pipe diameter attributes. Step S602: Based on the pipeline topology structure with completed pipe diameter attribute annotation, generate a pipeline attribute data table containing pipeline segment identifiers, start and end coordinates, and pipe diameter attributes.