Rapid three-dimensional modeling method and system for steam engine room pipe network of thermal power plant

Abstract

The invention discloses a thermal power plant steam engine room pipe network rapid three-dimensional modeling method and system, and belongs to the technical field of thermal power plant digital modeling, and the method specifically comprises the steps: obtaining pipe network point cloud data and position coordinate data through employing a laser scanning technology and a total station; denoising by adopting a filtering algorithm, splicing and fusing the point cloud data by adopting a point cloud registration algorithm to form a pipe network point cloud model, and aligning with a geographic coordinate system through a coordinate transformation algorithm; extracting pipe network characteristics by using a region growing algorithm, marking and classifying by using a template matching algorithm in combination with design drawing attribute data, and establishing a pipe network topological structure relationship; carrying out three-dimensional visual modeling on the pipe network through three-dimensional modeling software; and finally, checking and correcting to obtain a final rapid three-dimensional model of the steam engine room pipe network. According to the invention, the modeling efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital modeling of thermal power plants, and specifically relates to a method and system for rapid three-dimensional modeling of pipe networks in the turbine hall of a thermal power plant. Background Art

[0002] The pipe network system in the turbine hall of a thermal power plant is complex, including numerous pipelines such as steam pipelines, condensate pipelines, and feedwater pipelines. Therefore, an accurate three-dimensional model of the pipe network system in the turbine hall is of great significance for equipment management, fault diagnosis, renovation planning, etc. Traditional three-dimensional modeling methods often require a large amount of manpower, material resources, and time, and the modeling process is complex and prone to errors. For example, the surveying and mapping work of the pipe network may not be accurate enough, resulting in a deviation between the model and the actual situation; in the model construction stage, the parameter setting and connection relationship processing of each component may not be efficient enough, affecting the modeling speed and quality.

[0003] For example, the Chinese patent application with the publication number CN119089614A discloses a method, device, equipment, and medium for automatic three-dimensional pipe network GIS modeling of urban pipe networks, including: obtaining pipeline data of urban pipe networks, preprocessing the pipeline data of urban pipe networks to obtain original pipeline information data; identifying the pipe point table and pipeline table in the original pipeline information data, where the pipe point table includes key fields of several pipe points, and the pipeline table includes key fields of several pipelines; generating a modeling configuration file based on the key fields of several pipe points and the key fields of several pipelines; generating a modeling time configuration, and based on the modeling time configuration, importing the modeling configuration file into a preset modeling program to automatically model the urban pipe network to obtain a three-dimensional pipe network GIS model. This technical solution can be seamlessly docked with different modeling software platforms, and has the advantages of high modeling efficiency, low labor cost, and low operating cost such as fast data update.

[0004] For example, the Chinese patent with the authorization announcement number CN115795768B discloses a method and system for three-dimensional modeling and data update of pipe networks considering entity morphology, including: obtaining the area range of an underground pipeline construction project, exporting the current pipeline data of the area as a working base map to assist in the planning and design of pipeline projects; integrating and establishing quality inspection rules to review the planning and design scheme of pipeline projects; configuring quality inspection rule items according to the pipeline project completion survey data and generating a quality inspection report; judging the update status of the completion survey data that passes the quality inspection, and inserting the elements with the update status of deletion and attribute update into the historical database for data backup; adopting a range-level incremental update method to realize the update and storage of two-dimensional data; adopting an automatic parameter modeling method considering entity morphology to complete the construction of the three-dimensional model of underground pipelines for the two-dimensional data stored in the database; establishing an urban underground comprehensive pipe network information system to realize the management, application, and sharing of pipeline databases.

[0005] The above existing technologies all have the following problems: Limited by the functions and flexibility of the preset modeling program, the applicable range and flexibility are poor, and it is more suitable for large-scale and relatively fixed pipe network systems such as urban pipe networks; the data source and preprocessing complexity are relatively high, and the modeling efficiency and automation level are low. Summary of the Invention

[0006] In view of the deficiencies of the existing technologies, the present invention proposes a method and system for rapid three-dimensional modeling of the pipe network in the turbine building of a thermal power plant. Laser scanning technology and total station are used to obtain the pipe network point cloud data and position coordinate data; a filtering algorithm is adopted for denoising, and a point cloud registration algorithm is used to splice and fuse the point cloud data to form a pipe network point cloud model, which is aligned with the geographical coordinate system through a coordinate transformation algorithm; a region growing algorithm is used to extract the pipe network features, and a template matching algorithm is used for annotation and classification in combination with the attribute data of the design drawings to establish the topological structure relationship of the pipe network; three-dimensional visualization modeling of the pipe network is carried out through three-dimensional modeling software; finally, verification and correction are carried out to obtain the final rapid three-dimensional model of the pipe network in the turbine building; the present invention improves the modeling efficiency and accuracy.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for rapid three-dimensional modeling of the pipe network in the turbine building of a thermal power plant, comprising:

[0009] Step S1: Use laser scanning technology to perform an all-round scan of the pipe network in the turbine building to obtain pipe network point cloud data, and combine a total station to measure the core parts to obtain position coordinate data;

[0010] Step S2: Preprocess the collected pipe network point cloud data, and use a point cloud registration algorithm to splice and fuse the point cloud data obtained from multiple scans to form a pipe network point cloud model in the turbine building. Combining the position coordinate data, use a coordinate transformation algorithm to align the pipe network point cloud model in the turbine building with the actual geographical coordinate system;

[0011] Step S3: According to the geometric shape of the pipeline and the distribution characteristics of the point cloud data, use a region growing algorithm to automatically extract the pipe network features from the aligned pipe network point cloud model in the turbine building. Combining the attribute data in the design drawings, use a template matching algorithm to annotate and classify the extracted pipe network features, and establish the topological structure relationship of the pipe network. The pipe network features include the pipeline center line, pipe diameter information, pipe fitting type, position and attitude of the pipe fitting in space;

[0012] Step S4: According to the extracted pipe network features and topological structure, use three-dimensional modeling software to perform three-dimensional visualization modeling of the pipe network;

[0013] Step S5: Compare and verify the constructed 3D pipeline network model, and modify and improve the 3D model according to the verification results to obtain the final rapid 3D model of the pipeline network in the turbine building of the thermal power plant.

[0014] Specifically, the specific steps of step S2 include:

[0015] S2.1: Obtain the point cloud data of the pipeline network and p i =(x i , y i , z i ), i ∈ [1, n], and use to preprocess the collected point cloud data of the pipeline network to obtain the preprocessed point cloud data of the pipeline network P' = {p 1 ',..., p' n}, and p i '=(x i ', y i ', z i '), where p n represents the nth point cloud data point of the pipeline network, p' n represents the nth preprocessed point cloud data point of the pipeline network, x i , y i , z i respectively represent the x, y, and z coordinate values of the ith point cloud data point of the pipeline network, x i ', y i ', z i ' respectively represent the x, y, and z coordinate values of the ith preprocessed point cloud data point of the pipeline network, G(x j -x i , y j -y i , z j -z i ) represents the three-dimensional Gaussian weight function, G(x j -x i , y j -y i ) represents the two-dimensional Gaussian weight function, and N(i) represents the set of points within the filtering window around the point cloud data point p i ;

[0016] S2.2: For each point p 1 ' in the preprocessed point cloud data of the pipeline network P' = {p n ',..., p' i '}, determine a 3×3 neighborhood centered on p i ', and use the point feature histogram to calculate the geometric relationship of the points within the neighborhood of p i ' to obtain the feature descriptor of the point p i '.

[0017] Specifically, the specific steps of step S2 further include:

[0018] S2.3: Obtain the pipe network point cloud data and Use to calculate the similarity measure between all feature points in two point clouds and where d(p b,j , p c,l ) represents the Euclidean distance, p b,j represents a feature point in point cloud , p c,l represents a feature point in point cloud , p b,jr represents the coordinate value of feature point p b,j in the r dimension, p c,lr represents the coordinate value of feature point p c,l in the r dimension, r represents the dimension, and r = 1, 2, 3 correspond to the x, y, z coordinates;

[0019] S2.4: According to the similarity measure result, obtain the preliminary feature point pairs, randomly select m preliminary feature point pairs, and calculate a rigid body transformation model based on the m preliminary feature point pairs and in combination with the feature descriptor.

[0020] Specifically, the specific steps of step S2 further include:

[0021] S2.5: Calculate the transformation error of the remaining feature point pairs according to the rigid body transformation model, and compare the transformation error with a preset error threshold. If it is less than the error threshold, regard this feature point as an inlier and form an inlier set;

[0022] S2.6: Through iterative repetition, find the set of point pairs with the largest number of inliers as the final feature point pairs, and preliminarily align the point cloud according to the final feature point pairs.

[0023] Specifically, the specific steps of step S2 further include:

[0024] S2.7: Select control points from the position coordinate data in step S1. For each control point, substitute its coordinates into the formula to obtain the error equation, where respectively represent the x coordinate value, y coordinate value, and z coordinate value in the actual geographic coordinate system, k represents the scale parameter, respectively represent the rotation parameters on the x, y, and z coordinates, Δx i ′, Δy i ′, Δz i′ represent the translation parameters on the x, y, and z coordinates respectively;

[0025] S2.8: Combine the error equations of all control points into a matrix form to obtain an error equation matrix, and use the least squares method to solve the error equation matrix to obtain the estimated values of the seven parameters;

[0026] S2.9: Substitute the calculated seven parameters into the formula in S2.7 again, perform coordinate transformation on all points in the preliminarily aligned pipe network point cloud model to align it with the actual geographic coordinate system, and obtain the aligned pipe network point cloud model of the turbine building.

[0027] Specifically, the specific steps of step S3 include:

[0028] S3.1: Obtain the aligned pipe network point cloud model of the turbine building, analyze the distribution characteristics of the point cloud data therein, and combine with the geometric shape of the pipeline to select the starting end and branch point positions of the pipeline as seed points;

[0029] S3.2: Set the normal vector angle threshold according to the characteristics of the point cloud data and the geometric accuracy requirements of the pipe network.

[0030] S3.3: Starting from the seed points in S3.1, traverse the point cloud data in the aligned pipe network point cloud model of the turbine building according to the normal vector angle threshold set in S3.2. For each seed point, check the angle between the normal vector of its adjacent points and the normal vector of the seed point.

[0031] Specifically, the specific steps of step S3 further include:

[0032] S3.4: If the angle between the normal vector of the adjacent point and the normal vector of the seed point is less than the normal vector angle threshold, include the adjacent point in the same area as the seed point, and continue to check the adjacent points of the newly included points until there are no new adjacent points whose normal vector angles with the seed point are less than the normal vector angle threshold, and obtain the pipe network feature data;

[0033] S3.5: According to the attribute data in the design drawings, construct three-dimensional template models of different pipe fittings, and use a shape feature-based matching method to match the extracted pipe network feature data with the three-dimensional template models of the pipe fittings. If the matching is successful, label the attribute information in the three-dimensional template model of the pipe fitting on the corresponding pipe network features;

[0034] S3.6: Establish the topological structure relationship of the pipe network according to the labeled pipe network features and their connection relationships.

[0035] A rapid three-dimensional modeling system for the pipe network of a thermal power plant turbine building includes: a data acquisition module, a data processing module, a feature extraction and annotation module, a three-dimensional modeling module, a verification and correction module;

[0036] The data acquisition module is used to acquire the pipeline network data in the turbine hall;

[0037] The data processing module is used to denoise, splice and fuse the acquired data to form a point cloud model of the pipeline network in the turbine hall, and align it with the actual geographical coordinate system;

[0038] The feature extraction and annotation module is used to extract pipeline network features from the point cloud model of the pipeline network, perform annotation and classification, and establish the topological structure relationship of the pipeline network;

[0039] The 3D modeling module is used to perform 3D visualization modeling on the pipeline network using 3D modeling software to generate an intuitive and interactive 3D model of the pipeline network;

[0040] The verification and correction module is used to compare and verify the constructed 3D model of the pipeline network with the actual on-site situation, and correct and improve the 3D model according to the verification results.

[0041] Specifically, the data processing module includes: a filtering and denoising unit, a point cloud registration unit, and a coordinate transformation unit;

[0042] The filtering and denoising unit is used to perform denoising processing on the acquired point cloud data of the pipeline network using a filtering algorithm;

[0043] The point cloud registration unit is used to splice and fuse the point cloud data obtained from multiple scans using a point cloud registration algorithm, and form a point cloud model of the pipeline network in the turbine hall by finding the feature correspondence relationships between different scan data;

[0044] The coordinate transformation unit is used to align the point cloud model of the pipeline network in the turbine hall with the actual geographical coordinate system using a coordinate transformation algorithm, so that the point cloud model of the pipeline network in the turbine hall is accurately matched with the actual situation in terms of spatial position.

[0045] Specifically, the verification and correction module includes: a comparison and verification unit and a correction and improvement unit;

[0046] The comparison and verification unit is used to compare the 3D model of the pipeline network with the actual on-site situation to check the accuracy and integrity of the model;

[0047] The correction and improvement unit is used to correct and improve the 3D model according to the results of the comparison and verification.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] 1. The present invention proposes a rapid 3D modeling system for the pipeline network in the turbine hall of a thermal power plant, and has optimized and improved the architecture, operation steps and processes. The system has the advantages of simple process, low investment and operation costs, and low production work costs.

[0050] 2. The present invention provides a method for quickly three-dimensionally modeling the pipe network in the turbine hall of a thermal power plant. Through the precise measurement of laser scanning technology and total station, the three-dimensional point cloud data and position coordinate data of the pipe network in the turbine hall can be quickly and comprehensively obtained, thus ensuring the accuracy and integrity of the data; through preprocessing and point cloud registration algorithms, a point cloud model of the pipe network in the turbine hall can be efficiently generated and aligned with the actual geographic coordinate system.

[0051] 3. The present invention provides a method for quickly three-dimensionally modeling the pipe network in the turbine hall of a thermal power plant. By using the region growing algorithm and template matching algorithm, the features of the pipe network can be automatically extracted, labeled, and classified, thereby quickly establishing the topological structure relationship of the pipe network, which not only improves the modeling efficiency but also ensures the accuracy and precision of the model; finally, through visual modeling and comparison verification using three-dimensional modeling software, a complete three-dimensional model of the pipe network in the turbine hall of the thermal power plant is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Schematic diagram of a method for quickly three-dimensionally modeling the pipe network in the turbine hall of the present invention;

[0053] Figure 2 Principle flowchart of a method for quickly three-dimensionally modeling the pipe network in the turbine hall of the present invention;

[0054] Figure 3 System architecture diagram of a method for quickly three-dimensionally modeling the pipe network in the turbine hall of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] Embodiment 1

[0056] Please refer to Figure 1 and Figure 2 , an embodiment provided by the present invention: A method for quickly three-dimensionally modeling the pipe network in the turbine hall of a thermal power plant, comprising the following steps:

[0057] Step S1: Use laser scanning technology to scan the pipe network in the turbine hall comprehensively to obtain the point cloud data of the pipe network, and combine the total station to measure the core parts to obtain the position coordinate data;

[0058] It should be noted that when using laser scanning technology to conduct a full - range scan of the pipe network in the turbine hall, the obtained pipe network point cloud data is composed of multiple groups of point cloud data. Due to the spatial layout of the turbine hall and the complex structure of the pipe network, it is difficult to completely obtain all the point cloud information of the pipe network in a single scan. To obtain complete pipe network data, scans are usually carried out from multiple scanning stations. Each scanning station obtains a group of point cloud data, and these point cloud data cover the part of the pipe network that can be seen from the perspective of that station. In the subsequent data - processing process, such as constructing a complete three - dimensional model of the pipe network, these multiple groups of point cloud data need to be stitched and fused because each group of point cloud data only represents local information of the pipe network, and only by combining them can a complete point cloud model of the entire turbine hall pipe network be obtained.

[0059] Furthermore, the specific steps of step S1 include:

[0060] (1) Equipment preparation and parameter setting

[0061] Laser scanning equipment preparation: Select a laser scanner according to the size, complexity, and modeling accuracy requirements of the pipe network in the turbine hall. For example, if the pipe network in the turbine hall has a small space and dense pipes, a laser scanner with millimeter - level accuracy and high resolution needs to be selected;

[0062] Total station preparation: Calibrate the total station to ensure the accuracy of its measurements, including checking the levelness and verticality of the total station, and setting the measurement unit and angle unit;

[0063] (2) Scanning planning

[0064] According to the layout of the turbine hall and the distribution of the pipe network, plan the laser scanning path and stations. Generally speaking, it is necessary to ensure that the scanning stations can cover the entire pipe network system to avoid scanning blind spots. Methods such as circumferential scanning and layered scanning can be used. For complex pipe network intersection areas and pipes at high positions, special attention should be paid to the scanning angle and station position to obtain complete point cloud data;

[0065] (3) Laser scanning operation

[0066] Place the laser scanner at the planned station and start the scanning program. During the scanning process, record information such as the spatial coordinates and reflection intensity of each scanning point; the laser scanner emits a laser beam and receives the reflected light, calculates the distance from the scanning point to the scanner based on the flight time or phase difference of the laser, and then combines the spatial position and angle information of the scanner itself to determine the three - dimensional coordinates of the scanning point, thereby obtaining the pipe network point cloud data;

[0067] (4) Measurement of the core part of the total station

[0068] After completing the laser scanning, use a total station to measure the core parts of the pipe network. Among them, the core parts include key nodes such as pipe connection points, valve installation positions, and pipe branch points. Place the total station at a suitable position, aim at the target point, measure its horizontal angle, vertical angle, and inclined distance, and calculate the three-dimensional coordinates of the target point using trigonometric function relationships. Exemplarily, for a target point, let the horizontal angle measured by the total station be α, the vertical angle be β, and the inclined distance be s. The station coordinates of the total station are (X 0 , Y 0 , Z 0 ). Then the coordinate calculation formula for the target point is: where sin(·) represents the sine function and cos(·) represents the cosine function;

[0069] Record the position coordinate data of each measurement point in the core part. These position coordinate data will be combined with the pipe network point cloud data for subsequent data processing and model construction;

[0070] (5) Data recording and sorting

[0071] Sort out the pipe network point cloud data obtained by laser scanning and the position coordinate data measured by the total station. The data can be stored on the local hard disk or cloud server. Save the pipe network point cloud data in the LAS and PLY data formats, and save the position coordinate data in tabular form to ensure the integrity and accuracy of the data. At the same time, record relevant information such as the time, location, and equipment number of data collection for convenient subsequent data traceability and processing.

[0072] Step S2: Preprocess the collected pipe network point cloud data, and use the point cloud registration algorithm to splice and fuse the point cloud data obtained from multiple scans to form a pipe network point cloud model of the turbine building. Combine the position coordinate data and use the coordinate transformation algorithm to align the pipe network point cloud model of the turbine building with the actual geographic coordinate system;

[0073] Step S3: According to the geometric shape of the pipeline and the distribution characteristics of the point cloud data, use the region growing algorithm to automatically extract the pipeline network features from the aligned pipe network point cloud model of the turbine building. Combine the attribute data in the design drawings and use the template matching algorithm to label and classify the extracted pipeline network features to establish the topological structure relationship of the pipeline network. The pipeline network features include the pipeline centerline, pipe diameter information, pipe fitting type, position and attitude of the pipe fitting in space;

[0074] Step S4: According to the extracted pipeline network features and topological structure, use 3D modeling software to perform 3D visualization modeling of the pipeline network;

[0075] Furthermore, the specific steps of Step S4 include:

[0076] (1) In 3D modeling software, use its data import function to import the converted pipeline network feature data, including pipeline centerline coordinates, pipe diameter information, fitting types, position and attitude, and topological structure data, into the workspace;

[0077] (2) According to the imported pipeline centerline coordinates and pipe diameter information, use basic geometric shape creation tools in the software to generate pipeline models. For each section of the pipeline, create a cylinder with the centerline as the axis and the pipe diameter as the diameter;

[0078] (3) According to actual needs, perform detail processing on the pipeline model, such as adding textures to the pipeline surface and setting the transparency of the pipeline;

[0079] (4) For different types of fittings, such as elbows, tees, and valves, if there are compliant models in the software's built-in model library, they can be directly called; if not, they need to be created according to the geometric shape and size of the fittings; or fitting models can be imported from external model libraries, such as online 3D model resource websites or the enterprise's internal standard fitting model library;

[0080] (5) According to the imported fitting position and attitude information, accurately place the fittings in 3D space, align the connection points of the fittings with the corresponding pipeline endpoints, ensure that the connection between the fittings and the pipeline conforms to the topological structure relationship. At the same time, in the software, tools such as move, rotate, and scale can be used to adjust the position and attitude of the fittings;

[0081] (6) In the software, use its association tool or scripting function to construct the connection relationship between the pipeline and fitting models according to the imported topological structure data.

[0082] Step S5: Compare and verify the constructed 3D pipeline network model, and modify and improve the 3D model according to the verification results to obtain the final rapid 3D model of the steam turbine house pipeline network in the thermal power plant.

[0083] Furthermore, the specific steps of step S5 include:

[0084] (1) Compare and verify the constructed 3D model with the actual pipeline network through on-site measurement to check the accuracy of the model, including whether the pipeline orientation, position, size, etc. are consistent with the actual situation;

[0085] (2) Modify and improve the 3D model according to the results of the comparison and verification. If connection errors are found, such as incorrect alignment between fittings and pipelines or incorrect connection order, make corrections in a timely manner;

[0086] (3) After modification and improvement, obtain the final rapid 3D model of the steam turbine house pipeline network in the thermal power plant, and conduct final inspections and tests on the model to ensure the stability and usability of the model.

[0087] Exemplarily, in the project of modeling the pipe network in the steam turbine house of a thermal power plant, first, select appropriate laser scanners and total stations to comprehensively scan the pipe network in the steam turbine house and measure key parts. Among them, the scanning accuracy of the laser scanner is set to 0.5 mm, and the resolution is 1 mm to ensure that detailed and accurate point cloud data can be obtained. At the same time, the total station accurately measures important positions such as pipe connection points, and the measurement error is controlled within ±0.1 mm.

[0088] Then, import the collected data into data processing software, perform denoising processing using the Gaussian filtering algorithm, set the standard deviation of the filter to 0.3 mm, effectively remove noise points, and use the curvature-based reduction method to reduce the data, reducing the number of point clouds by 70% while retaining the key features of the pipe network. After calibration, the total station data is fused with the point cloud data to form a complete data set.

[0089] Secondly, use the region growing algorithm to extract the centerlines of the pipes. The growth threshold is set to 0.05 m according to the actual situation of the pipes to accurately identify the centerlines of each pipe. By measuring and calculating on the point cloud cross-section, the accurate dimensions and wall thickness information of each section of the pipe with a diameter range of 0.2 m - 1 m are obtained; for pipe fittings such as valves, elbows, and tees, use the shape analysis algorithm to identify and extract parameters. More than 50 types of pipe fittings are identified in total, and their accurate installation positions and connection directions are determined.

[0090] In addition, in professional 3D modeling software, create a pipe model according to the extracted centerlines and parameters, use the built-in pipe fitting library of the software to construct pipe fitting models, and accurately connect them to the pipe model. Assign metal material properties to the model, adjust the color to silver-gray, the glossiness to 0.6, and the reflectivity to 0.4 to make the model have good visualization effects.

[0091] Finally, compare and verify the constructed 3D model. By comparing the measured dimensions of 10 key parts on-site with the model, the errors are all controlled within ±2 mm, ensuring the accuracy and reliability of the model.

[0092] The specific steps of step S2 include:

[0093] S2.1: Obtain pipe network point cloud data and p i =(x i ,y i ,z i ), i ∈ [1, n], and use to preprocess the collected pipe network point cloud data to obtain the preprocessed pipe network point cloud data P' = {p 1 ',..., p' n}, and pi ′ = (x i ′, y i ′, z i ′), where p n represents the nth pipe network point cloud data point, p′ n represents the nth preprocessed pipe network point cloud data point, x i , y i , z i respectively represent the x, y, and z coordinate values of the ith pipe network point cloud data point, x i ′, y i ′, z i ′ respectively represent the x, y, and z coordinate values of the ith preprocessed pipe network point cloud data point, G(x j - x i , y j - y i , z j - z i ) represents a three-dimensional Gaussian weight function, G(x j - x i , y j - y i ) represents a two-dimensional Gaussian weight function, N(i) represents the set of points within the filtering window around the pipe network point cloud data point p i ;

[0094] It should be understood that when using laser scanning technology to perform a full - range scan of the pipe network in the turbine building, the obtained pipe network point cloud data is an overall point cloud, and this point cloud contains a large number of points, and these points are a set of three - dimensional coordinate points sampled by the laser scanning device on the surface of the pipe network in space; and the feature points are representative points extracted from this overall point cloud.

[0095] It should also be understood that during the process of point cloud registration, two or more point clouds are involved. For example, assuming that the pipe network in the turbine building is scanned twice to obtain point cloud b and point cloud c, in order to register point cloud b and point cloud c, corresponding points need to be found between them, and this process requires calculating the similarity measure between each feature point in point cloud b and all feature points in point cloud c.

[0096] S2.2: For each point p 1 ′,..., p′ n} in the preprocessed pipe network point cloud data P′ = {p i ′, determine a 3×3 neighborhood centered on p i ′, and use the point feature histogram to calculate the geometric relationship of the points within the neighborhood of p i ′ to obtain the feature descriptor of the point p i ′;

[0097] S2.3: Obtain the pipe network point cloud data and Use to calculate the similarity measure between all the feature points in two point clouds and where d(p b,j , p c,l ) represents the Euclidean distance, p b,j represents the feature point in point cloud , p c,l represents the feature point in point cloud , p b,jr represents the coordinate value of feature point p b,j in the r - dimension, p c,lr represents the coordinate value of feature point p c,l in the r - dimension, r represents the dimension, and r = 1, 2, 3 correspond to the x, y, z coordinates;

[0098] S2.4: According to the similarity measure result, obtain the preliminary feature point pairs, randomly select m preliminary feature point pairs, and calculate a rigid body transformation model based on the m preliminary feature point pairs and in combination with the feature descriptors;

[0099] Furthermore, the specific steps of S2.4 include:

[0100] (1) According to S2.2, obtain the feature descriptor of each feature point and screen out the preliminary feature point pairs;

[0101] (2) Randomly select m feature point pairs from the screened - out preliminary feature point pairs as the input for subsequent calculations;

[0102] (3) Use the selected m preliminary feature point pairs and in combination with the feature descriptors to calculate the parameters of the rigid body transformation model. The rigid body transformation formula is A′ = R×A + t, where R represents the rotation matrix, t represents the translation vector, A represents the original coordinates, and A′ represents the transformed coordinates;

[0103] (4) Obtain the optimal rotation matrix R and translation vector t by solving the least - squares problem. The least - squares is the prior art content in this field and is not the creative solution of this application, so it will not be elaborated here;

[0104] (5) Use the remaining feature point pairs to verify the accuracy of the calculated rigid body transformation model, and optimize and adjust the model according to the verification result.

[0105] S2.5: Calculate the transformation error of the remaining feature point pairs according to the rigid body transformation model, and compare the transformation error with a preset error threshold. If it is less than the error threshold, regard this feature point as an inlier and form an inlier set;

[0106] S2.6: Through iterative repetition, find the pair of points with the largest number of inliers as the final feature point pair, and based on the final feature point pair, preliminarily align the point cloud;

[0107] It should be understood that a feature point pair is composed of two feature points from two different point clouds (such as point clouds and ). During the feature matching process of point cloud registration, the goal is to find the corresponding point of each feature point in point cloud in point cloud . Each pair of such corresponding points is a feature point pair; when using the Random Sample Consensus (RANSAC) method for feature point matching, an inlier refers to a feature point pair that conforms to a certain hypothesized rigid body transformation model. This rigid body transformation model includes rotation and translation, which describes the possible spatial position relationship between two point clouds; at the same time, in actual point cloud data, due to the geometric shape of the pipe network having a certain regularity and continuity, for the same object, such as a pipe, the corresponding parts in point clouds obtained from different scans should follow similar rigid body transformation laws. For example, the position relationship of a pipe in two scans is described by an overall rotation and translation. Therefore, when a reasonable rigid body transformation model is calculated through several randomly selected feature point pairs, there will be multiple other feature point pairs that conform to this transformation model, such as those with an error less than the threshold. These point pairs represent the truly corresponding parts in point clouds and , and they constitute the inlier set. By repeatedly applying the RANSAC method through multiple iterations, selecting the pair of points with the largest number of inliers as the final feature point pair is to find the matching point pair that can most accurately describe the transformation relationship between the two point clouds, thereby achieving more accurate point cloud registration.

[0108] S2.7: Select control points from the position coordinate data in step S1. For each control point, substitute its coordinates into the formula to obtain the error equation, where respectively represent the x - coordinate value, y - coordinate value, and z - coordinate value in the actual geographic coordinate system, k represents the scale parameter, respectively represent the rotation parameters on the x, y, and z coordinates, Δx i ′, Δy i ′, Δz i ′ respectively represent the translation parameters on the x, y, and z coordinates;

[0109] S2.8: Combine the error equations of all control points into a matrix form to obtain an error equation matrix, and use the least squares method to solve the error equation matrix to obtain the estimated values of the seven parameters. The least squares solution formula is the prior art content in the field and is not the creative solution of this application, so it will not be elaborated here;

[0110] S2.9: Substitute the calculated seven parameters into the formula in S2.7 again, perform coordinate transformation on all points in the preliminarily aligned pipe network point cloud model to align it with the actual geographic coordinate system, and obtain the aligned turbine hall pipe network point cloud model.

[0111] The specific steps of step S3 include:

[0112] S3.1: Obtain the aligned turbine hall pipe network point cloud model, analyze the distribution characteristics of the point cloud data therein, and combine with the geometric shape of the pipeline to select the starting end and branch point positions of the pipeline as seed points;

[0113] S3.2: Set the normal vector angle threshold according to the characteristics of the point cloud data and the geometric accuracy requirements of the pipe network.

[0114] S3.3: Starting from the seed points in S3.1, traverse the point cloud data in the aligned turbine hall pipe network point cloud model according to the normal vector angle threshold set in S3.2. For each seed point, check the angle between the normal vector of its adjacent points and the normal vector of the seed point;

[0115] S3.4: If the angle between the normal vector of the adjacent point and the normal vector of the seed point is less than the normal vector angle threshold, then include the adjacent point in the same region as the seed point, and continue to check the adjacent points of the newly included points until there are no new adjacent points whose normal vector angles with the seed point are less than the normal vector angle threshold, and obtain the pipe network feature data;

[0116] S3.5: According to the attribute data in the design drawings, construct three-dimensional template models of different pipe fittings, and use a shape feature-based matching method to match the extracted pipe network feature data with the three-dimensional template models of the pipe fittings. If the matching is successful, mark the attribute information in the three-dimensional template model of the pipe fitting on the corresponding pipe network features. The shape feature-based matching method is the prior art content in the field and is not the creative solution of this application, so it will not be elaborated here;

[0117] S3.6: Establish the topological structure relationship of the pipe network according to the marked pipe network features and their connection relationships.

[0118] Embodiment 2

[0119] Please refer to Figure 3 , another embodiment provided by the present invention: A rapid three-dimensional modeling system for the pipe network in the turbine hall of a thermal power plant, including:

[0120] Data acquisition module, data processing module, feature extraction and annotation module, 3D modeling module, verification and correction module;

[0121] The data acquisition module is used to collect the pipeline network data in the turbine building;

[0122] The data processing module is used to denoise, splice and fuse the collected data to form a point cloud model of the pipeline network in the turbine building and align it with the actual geographical coordinate system;

[0123] The feature extraction and annotation module is used to extract pipeline network features from the point cloud model of the pipeline network, perform annotation and classification, and establish the topological structure relationship of the pipeline network;

[0124] The 3D modeling module is used to perform 3D visualization modeling on the pipeline network using 3D modeling software to generate an intuitive and interactive 3D model of the pipeline network;

[0125] The verification and correction module is used to compare and verify the constructed 3D model of the pipeline network with the actual on-site situation, and correct and improve the 3D model according to the verification results.

[0126] The data acquisition module includes: a laser scanning unit and a total station measurement unit;

[0127] The laser scanning unit is used to perform an all-round scan of the pipeline network in the turbine building using laser scanning technology to obtain point cloud data of the pipeline network, and the point cloud data of the pipeline network can reflect the general geometric shape and spatial position information of the pipeline network;

[0128] The total station measurement unit is used to measure the core parts of the pipeline network to obtain position coordinate data, and the position coordinate data is used to supplement the possible local accuracy deficiency problems in laser scanning and provide key control point information for subsequent data registration and coordinate system alignment. Among them, the core parts of the pipeline network include but are not limited to key connection points, valves, and branch starting points.

[0129] The data processing module includes: a filtering and denoising unit, a point cloud registration unit, and a coordinate transformation unit;

[0130] The filtering and denoising unit is used to perform denoising processing on the collected point cloud data of the pipeline network using a filtering algorithm to remove noise points generated by environmental interference, such as light reflection and equipment vibration, improve the purity and quality of the data, so as to more accurately extract pipeline network features and construct models in the subsequent process;

[0131] The point cloud registration unit is used to splice and fuse the point cloud data obtained from multiple scans by using the point cloud registration algorithm. By finding the feature correspondence relationships between different scan data, they are integrated into a complete point cloud model of the turbine hall pipe network, ensuring the integrity and continuity of the data and avoiding problems such as model missing or incoherence caused by scan range limitations;

[0132] The coordinate transformation unit is used to align the pipe network point cloud model with the actual geographical coordinate system by using the coordinate transformation algorithm, so that the model is accurately matched with the actual situation in terms of spatial position, ensuring the accuracy of the pipe network model in the geographical space.

[0133] The feature extraction and annotation module includes: a feature extraction unit, an annotation and classification unit, and a topology construction unit;

[0134] The feature extraction unit is used to automatically extract the pipe network features from the pipe network point cloud model by using the region growing algorithm according to the geometric shape of the pipeline and the distribution characteristics of the point cloud data.

[0135] The annotation and classification unit is used to combine the attribute data in the design drawings and use the template matching algorithm to annotate and classify the extracted pipe network features;

[0136] Among them, mainly by setting appropriate growth criteria, the point cloud regions with similar features are identified as different pipe network elements such as pipelines and pipe fittings, so as to extract key geometric features such as the pipeline center line and pipe diameter information, providing accurate shape parameters for 3D modeling.

[0137] The topology construction unit is used to establish the topological structure relationship of the pipe network according to the extracted pipe network features and annotation information.

[0138] Among them, mainly the extracted pipe network features are compared with the information such as the standard pipe fitting shapes and sizes in the design drawings. The specific type of each pipe fitting, such as elbow, tee, and valve, is determined through template matching, and the corresponding attribute labels are assigned. At the same time, the topological structure relationship of the pipe network is established, clarifying the connection sequence and logical relationship between each pipe network element, so that the model not only has geometric shape information but also complete engineering attribute information.

[0139] The verification and correction module includes: a comparison and verification unit and a correction and improvement unit;

[0140] The comparison and verification unit is used to compare the 3D pipe network model with the actual situation on site to check the accuracy and integrity of the model;

[0141] Among them, mainly by on-site field measurement of the dimensions, positions and other information of some key parts, and comparing and analyzing with the corresponding data in the 3D model to check whether there are deviations or errors in the model; at the same time, inviting engineers with actual operation and maintenance experience to review the integrity and rationality of the model, evaluating from the perspective of engineering practice whether the model accurately reflects the actual situation of the pipe network in the turbine building, and finding out the difference points and potential problems between the model and the actual situation.

[0142] The correction and improvement unit is used to correct and improve the 3D model according to the results of comparison and verification, ensure the accuracy and practicability of the model, improve the accuracy and reliability of the model, and meet the actual application requirements.

[0143] Among them, for dimensional deviations, adjust the parameters of the corresponding pipes and pipe fittings in the model; for topological structure errors, re-comb and modify the connection relationships between the pipe network elements; for the missing or unclear parts in the model, feedback to the data acquisition module or the feature extraction and annotation module to supplement data acquisition or re-extract features. After multiple iterations of verification and correction, until the final accurate and complete rapid 3D model of the pipe network in the turbine building of the thermal power plant is obtained.

[0144] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make changes, modifications, substitutions and variations to the above embodiments without departing from the purpose and scope of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant, characterized in that: include: Step S1: Use laser scanning technology to perform an all-round scan of the pipe network in the turbine room to obtain pipe network point cloud data, and use a total station to measure the core parts to obtain position coordinate data; Step S2: pre-processing the collected pipe network point cloud data, and using the point cloud registration algorithm to stitch and fuse the point cloud data obtained by multiple scans to form a steam turbine room pipe network point cloud model, and combining the position coordinate data, using the coordinate transformation algorithm to align the steam turbine room pipe network point cloud model with the actual geographic coordinate system; Step S3: according to the geometric shape of the pipeline and the distribution characteristics of the point cloud data, the regional growing algorithm is used to automatically extract the pipeline network features from the aligned turbine room pipeline network point cloud model. Combined with the attribute data in the design drawing, the template matching algorithm is used to annotate and classify the extracted pipeline network features, and the topological structure relationship of the pipeline network is established. The pipeline network features include the pipeline centerline, pipe diameter information, pipe fitting type, and the position and posture of the pipe fitting in space; Step S4: Performing three-dimensional visualization modeling of the pipe network using three-dimensional modeling software according to the extracted pipe network features and topological structure; Step S5: Compare and verify the constructed three-dimensional model of the pipe network, and modify and improve the three-dimensional model according to the verification result to obtain the final rapid three-dimensional model of the pipe network of the steam turbine room of the thermal power plant.

2. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant according to claim 1, characterized in that: The specific steps of step S2 include: S2.1: Obtaining pipe network point cloud data And p i =(x i ,y i ,z i ),i∈[1,n], and use The collected pipe network point cloud data are preprocessed to obtain the preprocessed pipe network point cloud data P′={p′1,...,p′ n }, and p′ i =(x′ i ,y′ i ,z′ i ), where p n represents the nth pipe network point cloud data point, p′ n represents the nth preprocessed pipe network point cloud data point, x i ,y i 、z i Respectively represent the x, y, and z coordinate values ​​of the i-th pipe network point cloud data point, x′ i , y′ i , z′ i Respectively represent the x, y, and z coordinate values ​​of the i-th preprocessed pipe network point cloud data point, G(x j -x i ,y j -y i ,z j -z i ) represents the three-dimensional Gaussian weight function, G(x j -x i ,y j -y i ) represents the two-dimensional Gaussian weight function, N(i) represents the network point cloud data point p i The set of points within the surrounding filter window; S2.2: Preprocess the pipe network point cloud data P′={p′1,...,p′ n Each point p′ in i , with p′ i Determine a 3×3 neighborhood for the center and calculate p′ using the point feature histogram i The geometric relationship of points in the neighborhood to obtain point p′ i feature descriptor.

3. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant according to claim 2, characterized in that: The specific steps of step S2 also include: S2.3: Obtaining pipe network point cloud data and use Calculate two point clouds and The similarity measure between all feature points in , where d(p b,j ,p c,l ) represents the Euclidean distance, p b,j Representing point clouds The characteristic points in c,l Representing point clouds The characteristic points in b,jr Represents the feature point p b,j The coordinate value in the r dimension, p c,lr Represents the feature point p c,l The coordinate value in the r dimension, r represents the dimension, and r = 1, 2, 3 corresponds to the x, y, z coordinates; S2.4: According to the similarity measurement result, preliminary feature point pairs are obtained, m preliminary feature point pairs are randomly selected, and a rigid body transformation model is calculated based on the m preliminary feature point pairs in combination with feature descriptors.

4. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant as claimed in claim 3, characterized in that: The specific steps of step S2 also include: S2.5: Calculate the transformation error of the remaining feature point pairs according to the rigid body transformation model, and compare the transformation error with a preset error threshold. If the error is less than the error threshold, the feature point is regarded as an inlier point to form an inlier point set. S2.6: Through iterative repetition, find the set of point pairs with the largest number of inliers as the final feature point pairs, and preliminarily align the point cloud based on the final feature point pairs.

5. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant according to claim 4, characterized in that: The specific steps of step S2 also include: S2.7: Select control points from the position coordinate data in step S1. For each control point, substitute its coordinates into the formula The error equation is obtained, where They represent the x-coordinate value, y-coordinate value, and z-coordinate value in the actual geographic coordinate system, respectively; k represents the scale parameter. Respectively represent the rotation parameters on the x, y, and z coordinates, Δx′ i , Δy′ i , Δz′ i Represent the translation parameters in x, y, and z coordinates respectively; S2.8: Combining the error equations of all control points into a matrix form to obtain an error equation matrix, and using the least squares method to solve the error equation matrix to obtain the estimated values ​​of the seven parameters; S2.9: Substitute the calculated seven parameters into the formula in S2.7 again, perform coordinate transformation on all points in the initially aligned pipe network point cloud model to align them with the actual geographic coordinate system, and obtain the aligned turbine room pipe network point cloud model.

6. A method for rapid three-dimensional modeling of a pipeline network in a steam engine room of a thermal power plant according to claim 5, characterized in that: The specific steps of step S3 include: S3.1: Obtain the aligned point cloud model of the turbine room pipe network, analyze the distribution characteristics of the point cloud data, and select the starting end and branch point of the pipeline as seed points based on the geometric shape of the pipeline; S3.2: Set the normal vector angle threshold according to the characteristics of the point cloud data and the geometric accuracy requirements of the pipe network; S3.3: Starting from the seed point in S3.1, the point cloud data in the aligned turbine room pipe network point cloud model is traversed according to the normal vector angle threshold set in S3.

2. For each seed point, the angle between the normal vector of its adjacent point and the normal vector of the seed point is checked.

7. A method for rapid three-dimensional modeling of a pipe network in a steam engine room of a thermal power plant according to claim 6, characterized in that: The specific steps of step S3 also include: S3.4: If the angle between the normal vector of the adjacent point and the normal vector of the seed point is less than the normal vector angle threshold, the adjacent point is included in the same area as the seed point, and the adjacent points of the newly included point are continuously checked until there is no new adjacent point whose normal vector is less than the normal vector angle threshold with the normal vector of the seed point, and the pipe network characteristic data is obtained; S3.5: According to the attribute data in the design drawings, construct three-dimensional template models of different pipe fittings, and use a shape feature-based matching method to match the extracted pipe network feature data with the three-dimensional template model of the pipe fitting. If the match is successful, the attribute information in the three-dimensional template model of the pipe fitting is marked on the corresponding pipe network feature; S3.6: Establish the topological structure relationship of the pipeline network based on the marked pipeline network characteristics and the connection relationship between them.

8. A rapid three-dimensional modeling system for a thermal power plant steam engine room pipe network, which is used to implement a rapid three-dimensional modeling method for a thermal power plant steam engine room pipe network as claimed in any one of claims 1 to 7, characterized in that: include: Data acquisition module, data processing module, feature extraction and annotation module, 3D modeling module, verification and correction module; The data acquisition module is used to collect the pipe network data in the turbine room; The data processing module is used to remove noise, splice and fuse the collected data to form a point cloud model of the turbine room pipe network and align it with the actual geographic coordinate system; The feature extraction and annotation module is used to extract pipe network features from the pipe network point cloud model, annotate and classify them, and establish the topological structure relationship of the pipe network; The three-dimensional modeling module is used to perform three-dimensional visual modeling of the pipe network using three-dimensional modeling software to generate an intuitive and interactive three-dimensional model of the pipe network; The verification and correction module is used to compare and verify the constructed three-dimensional model of the pipe network with the actual situation on site, and to correct and improve the three-dimensional model according to the verification result.

9. A rapid three-dimensional modeling system for a thermal power plant steam engine room pipe network as claimed in claim 8, characterized in that: The data processing module includes: a filtering and denoising unit, a point cloud registration unit, and a coordinate transformation unit; The filtering and denoising unit is used to perform denoising processing on the collected pipe network point cloud data using a filtering algorithm; The point cloud registration unit is used to stitch and fuse the point cloud data obtained by multiple scans using a point cloud registration algorithm, and to form a point cloud model of the turbine room pipe network by finding the feature correspondence between different scan data; The coordinate transformation unit is used to align the turbine room pipe network point cloud model with the actual geographic coordinate system using a coordinate transformation algorithm, so that the turbine room pipe network point cloud model accurately matches the actual situation in terms of spatial position.

10. A rapid three-dimensional modeling system for a thermal power plant steam engine room pipe network as claimed in claim 9, characterized in that: The verification and correction module includes: a comparison and verification unit and a correction and improvement unit; The comparison and verification unit is used to compare the three-dimensional model of the pipe network with the actual situation on site to check the accuracy and completeness of the model; The correction and improvement unit is used to correct and improve the three-dimensional model according to the comparison and verification results.

Images