Ship modeling proofreading design method and system based on finite element model

Abstract

The invention provides a ship modeling proofreading design method and system based on a finite element model, and the method comprises the steps: S100: analyzing the finite element model, determining the physical attributes of plates and profiles at different gravity center positions of a ship space, and forming a database; s200, forming a training data set of each category of attributes according to the physical attributes of the plates and the profiles; s300, based on the training data set of each category attribute, selecting a proper machine learning algorithm to carry out machine learning, and generating a prediction model; s400, extracting gravity centers and carding attributes of plates and profiles in the ship modeling software to form to-be-predicted data; s500, inputting to-be-predicted data on the basis of the prediction model, and predicting various types of attributes of the plates and the profiles to obtain a prediction result; and S600, comparing the prediction result with the physical attributes of the plates and the profiles extracted from the database, and outputting a proofreading result. According to the invention, automatic proofreading and predictable learning of ship structure modeling are realized, and the efficiency and accuracy of ship structure modeling are effectively improved.

Description

Technical Field

[0001] The present application belongs to the field of ship technology, and in particular, relates to a ship modeling and proofreading design method and system based on a finite element model. Background Art

[0002] With the development of large-scale ships, ship structures are becoming more and more complex, and the number of ship structural parts is also increasing. A ship can even be composed of hundreds of thousands of parts. It is necessary to confirm that the material and specifications of each structural part are correct to ensure the safety of the ship's navigation. At present, ship design usually first carries out basic design according to the requirements of the specifications, and constructs a basic design finite element model to verify the mechanical properties of the ship. Then, a two-dimensional drawing is drawn and the ship structure is modeled according to the two-dimensional drawing to meet the production design requirements. In this process, the materials and materials of different positions of the two-dimensional drawings and the three-dimensional model of the structure are manually compared. The manual proofreading of massive parts is cumbersome, the verification cycle is long, the work efficiency is low, and the accuracy cannot be guaranteed once. Repeated proofreading is required, which has a great impact on the ship construction cycle. Therefore, it is necessary to provide an efficient and accurate ship modeling proofreading design method. Summary of the invention

[0003] In view of the shortcomings of the prior art described above, the purpose of the present application is to provide a ship modeling and proofreading design method and system based on a finite element model, so as to solve the problem that the proofreading of ship structure models in the prior art is cumbersome, has a long cycle and has low accuracy.

[0004] In a first aspect, the present application provides a ship modeling and proofreading design method based on a finite element model, comprising at least the following steps:

[0005] S100: Analyze the finite element model to determine the physical properties of plates and profiles at different center of gravity positions in the ship space and form a database;

[0006] S200: forming a training data set of attributes of each category according to the physical attributes of the plate and profile;

[0007] S300: Based on the training data set of each category attribute, select a suitable machine learning algorithm to perform machine learning and generate a prediction model;

[0008] S400: Extract the center of gravity of plates and profiles in the ship modeling software and sort out the properties to form the data to be predicted;

[0009] S500: Based on the prediction model, the data to be predicted is input, and the attributes of each category of the plate and profile are predicted to obtain a prediction result;

[0010] S600: Compare the prediction result with the physical properties of the plates and profiles extracted from the database, and output the proofreading result.

[0011] In an optional implementation manner, the physical properties include: material properties and thickness properties of the plate, and material properties and specification properties of the profile.

[0012] In an optional implementation manner, step S100 includes:

[0013] S101: Select a structural finite element file, and obtain the corresponding plate material and thickness, profile material and specification in the finite element code through file analysis to form a material database;

[0014] S102: parsing the position code in the finite element file, obtaining the position coordinates, and forming a position code coordinate database;

[0015] S103: parsing the mesh composition code and material code in the finite element file, obtaining mesh coordinates, and forming a plate database;

[0016] S104: parsing the BEAM component code and material code in the finite element file, obtaining the line segment coordinates, and forming a profile database;

[0017] S105: Calculate the centroid based on the grid and line segment coordinates, and obtain the material and specification of the plate and profile at the centroid position of each spatial grid.

[0018] In an optional implementation, the method for calculating the centroid based on the grid and line segment coordinates is:

[0019] According to the definition of materials in the start area of ​​the finite element file, the materials are divided into at least plates, T-shaped bars, angle steels, and flat steels;

[0020] Screen out the plates, find the spatial coordinate definition of each code according to the composition code of each grid of the plates, calculate the centroid of several spatial coordinates, and store the specific material description and the corresponding centroid X, Y, and Z coordinate values ​​obtained by converting the plate material definition code into the database;

[0021] Filter out one of the materials of T-profile, angle steel, and flat steel, find the CBEAM material description code corresponding to the filtered material, the coordinate code of the starting point and the end point, convert it into a specific material description and specific coordinate value, calculate the centroid according to the coordinate value, and store the material description and the X, Y, and Z coordinate values ​​of the centroid in the database.

[0022] In an optional implementation manner, a data verification step is also included:

[0023] According to the length, width and height of the ship, check whether the centroid calculation is within the range;

[0024] According to the standard range of marine materials, check whether the materials of plates and profiles meet the requirements.

[0025] In an optional implementation manner, in step S200, according to the data stored in the database, the plates, T-profiles, angle steels, and flat steels are classified to generate a plate training set, a T-profile training set, an angle steel training set, and a flat steel training set respectively; wherein,

[0026] The attributes of the plate training set include the X, Y, and Z coordinate values ​​of the centroid, the thickness and material of the plate; the X, Y, and Z coordinate values ​​are processed as numerical formats, and the thickness and material of the plate are combined as classification processing;

[0027] The T-profile training set, angle steel training set, and flat steel training set include the X, Y, and Z coordinates of the centroid, the starting point coordinates, the end point coordinates, and the specifications and materials of the profiles. The X, Y, and Z coordinates of the centroid and the starting point coordinates and the end point coordinates are processed as numerical formats, and the specifications and materials of the T-profiles are combined and processed as categories.

[0028] In an optional implementation, in step S300, different algorithms are used for training according to different regions of the ship:

[0029] In the ship cargo hold area, the bagging tree algorithm of the ensemble classifier is selected for machine learning classification;

[0030] In the tail area of ​​the cabin, the weighted KNN algorithm of the nearest neighbor classifier is selected for machine learning classification;

[0031] In the upper construction area and bow area, the subspace discrimination algorithm of the integrated classifier is selected for machine learning.

[0032] In an optional implementation manner, in step S300,

[0033] For T-profiles, the weighted KNN algorithm is used in the ship cargo hold area, and the integrated bagging tree algorithm is used in other areas;

[0034] For angle steel and flat steel, the medium Gaussian SVM algorithm of support vector machine is used in the whole ship area.

[0035] In a second aspect, the present application provides a ship modeling and proofreading system based on a finite element model, comprising:

[0036] Finite element analysis module: used to read and analyze finite element files to form a database of materials, plates, and profiles;

[0037] Training data set generation module: used to generate plate and profile training data sets;

[0038] Plate machine learning training module: used to train data and generate plate prediction models;

[0039] Profile machine learning training module: used to train data and generate profile prediction models;

[0040] Structural model data extraction module: used to extract the center of gravity, material, and specifications of the plates and profiles in structural modeling, form a corresponding database, and generate data to be predicted;

[0041] Plate automatic proofreading module: used to input plate data to be predicted, generate prediction results, and automatically compare with plate data extracted from the database to generate comparison results;

[0042] Profile automatic proofreading module: used to input profile data to be predicted, generate prediction results, and automatically compare with profile data extracted from the database to generate comparison results.

[0043] Compared with the prior art, the technical solution provided by this application has the following beneficial effects:

[0044] The technical solution provided by this application is based on finite element model and machine learning to form an automatic proofreading design method and system for ship structure modeling. The method includes: based on the finite element model file, parsing the quadrilateral grid and triangular grid of the plate and the line segment of the profile, obtaining the material and specification and other properties of the plate and profile at different centroid positions of the ship space grid, and forming a training data set for each category of plate and profile attributes; selecting a suitable machine learning algorithm, carrying out machine learning, and generating a prediction model; extracting the material, specification and center of gravity of the plate and profile for structural modeling in the ship modeling software to form data to be predicted; based on the generated prediction model, inputting the data to be predicted for prediction and comparing it with the extracted material and specification to obtain the proofreading result of ship structure modeling. This application realizes the automatic proofreading of ship structure modeling and predictability learning, which provides a technical basis for the next stage of automatic design to automatically assign material and specification and other properties to the ship structure model, effectively improves the efficiency and accuracy of ship structure modeling, and provides a basis for future artificial intelligence ship design. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Shown is a step diagram of the ship modeling and proofreading design method based on the finite element model provided by this application;

[0046] Figure 2 Shown is a diagram of a ship modeling and proofreading system based on a finite element model provided by this application. DETAILED DESCRIPTION

[0047] The following describes the implementation methods of the present application through specific examples. Those skilled in the art can easily understand other advantages and principles of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific implementation methods. The details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application.

[0048] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the restrictive conditions for the implementation of this patent, so they have no substantial technical significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects and purposes that can be achieved by this patent, should still fall within the scope of the technical content disclosed by this patent. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and so on quoted in this specification are only for the convenience of description, and are not used to limit the scope of the implementation of this patent. The change or adjustment of their relative relationship should also be regarded as the scope of the implementation of this patent without substantial change in the technical content.

[0049] Embodiment 1:

[0050] This embodiment provides a ship modeling and proofreading design method based on a finite element model, which generates training data by analyzing the finite element model, and uses a machine learning algorithm to build a prediction model to achieve automatic proofreading of ship structure modeling. Figure 1 , the method comprises at least the following steps:

[0051] S100: Analyze the finite element model to determine the physical properties of plates and profiles at different center of gravity positions in the ship space and form a database. The physical properties include: plate material properties and thickness properties, profile material properties and specification properties

[0052] S101: Select a structural finite element file, and obtain the corresponding plate material and thickness, profile material and specification in the finite element code through file analysis to form a material database.

[0053] S102: parsing the position code in the finite element file, obtaining the position coordinates, and forming a position code coordinate database.

[0054] S103: parsing the mesh composition code and material code in the finite element file, obtaining mesh coordinates, and forming a plate database.

[0055] S104: parsing the BEAM component code and material code in the finite element file, obtaining the line segment coordinates, and forming a profile database.

[0056] Optionally, the above data parsing method is divided into plates and profiles according to the classification of the geometric description of the finite element file. The geometric grid of the plate is described by CQUAD4 and CTRIA3, and each grid is described by the codes of several spatial coordinate points. The geometric grid of the profile is usually described by CBEAM, including the starting point and the end point, and the starting point and the end point are both described by codes. Regarding the material category, there is a clear definition code in the starting area of ​​the finite element file. Regarding the attribute definition of the finite element mesh, it includes the composition code of the material, thickness, specification code, and mesh corner point or starting point and end point.

[0057] S105: Calculate the centroid based on the grid and line segment coordinates, and obtain the material and specification of the plate and profile at the centroid position of each spatial grid.

[0058] Specifically, the method for calculating the centroid based on the grid and line segment coordinates is:

[0059] According to the definition of materials in the start area of ​​the finite element file, the materials are divided into at least plates, T-shaped bars, angle steels, and flat steels;

[0060] Screen out the plates, find the spatial coordinate definition of each code according to the composition code of each grid of the plates, calculate the centroid of several spatial coordinates, and store the specific material description and the corresponding centroid X, Y, and Z coordinate values ​​obtained by converting the plate material definition code into the database;

[0061] Filter out one of the materials of T-profile, angle steel, and flat steel, find the CBEAM material description code corresponding to the filtered material, the coordinate code of the starting point and the end point, convert it into a specific material description and specific coordinate value, calculate the centroid according to the coordinate value, and store the material description and the X, Y, and Z coordinate values ​​of the centroid in the database.

[0062] Specifically, the data verification steps include: verifying whether the centroid calculation is within the range based on the length, width and height of the ship; and verifying whether the materials of the plates and profiles meet the requirements based on the standard range of marine materials.

[0063] S200: Forming training data sets of attributes of each category according to the physical attributes of the plates and profiles.

[0064] Specifically, according to the data stored in the database, the plates, T-profiles, angle steels, and flat steels are classified to generate a plate training set, a T-profile training set, an angle steel training set, and a flat steel training set respectively; wherein,

[0065] The attributes of the plate training set include the X, Y, and Z coordinate values ​​of the centroid, the thickness and material of the plate; the X, Y, and Z coordinate values ​​are processed as numerical formats, and the thickness and material of the plate are combined as classification processing;

[0066] The T-profile training set, angle steel training set, and flat steel training set include the X, Y, and Z coordinates of the centroid, the starting point coordinates, the end point coordinates, and the specifications and materials of the profiles. The X, Y, and Z coordinates of the centroid and the starting point coordinates and the end point coordinates are processed as numerical formats, and the specifications and materials of the T-profiles are combined and processed as categories.

[0067] S300: Based on the training data set of each category attribute, select a suitable machine learning algorithm to perform machine learning and generate a prediction model. This step uses a classification learner to perform machine learning. The machine learning algorithms include support vector machines, nearest neighbor classifiers, ensemble classifiers, neural network classifiers, logistic regression classifiers, etc. Specifically,

[0068] For plates, T-shaped bars, angle steels, and flat steels, the steps of machine training are similar. The difference lies in the selection of algorithms for different areas of the ship, and the accuracy is different. Different algorithms are used for training according to different areas of the ship: in the cargo hold area of ​​the ship, the bagging tree algorithm of the integrated classifier is selected for machine learning classification; in the rear area of ​​the engine room, the weighted KNN algorithm of the nearest neighbor classifier is selected for machine learning classification; in the upper construction area and the bow area, the subspace discrimination algorithm of the integrated classifier is selected for machine learning.

[0069] Furthermore, for T-profiles, the weighted KNN algorithm is used in the cargo hold area of ​​the ship, and the integrated bagging tree algorithm is used in other areas. If the integrated bagging tree algorithm is selected, the maximum number of splits and the number of learners can be defined. After machine learning, the verification confusion matrix and the accuracy verification rate are checked. Generally, close to 99% is considered to meet the training requirements. When the accuracy verification rate is low, the cross-validation folds of the training data set can be increased or decreased for training. The corresponding parameters can also be adjusted according to different algorithms. If the integrated bagging tree algorithm is selected, the number of learners can be increased, or the number of feature selections can be reduced. If the weighted KNN algorithm is selected, the number of neighboring points can be increased or decreased, and different distance metrics can be selected for adjustment. These methods can obtain different accuracy verification rates. For angle steel and flat steel, the medium Gaussian SVM algorithm of the support vector machine is used in the entire ship area.

[0070] Furthermore, check the scatter plot, check the incorrect intervals of the model prediction, and find the algorithm with the smallest incorrect intervals in different areas of the ship. The specific method is as follows: select the prediction variable, check the incorrect classification according to the color of the prediction model, judge the quantity and spatial distribution density of these incorrect classifications, and find the algorithm with the lowest spatial distribution density. After meeting the accuracy verification requirements, generate a training model.

[0071] S400: extracting the center of gravity of plates and profiles in the ship modeling software and sorting out the attributes to form the data to be predicted. Specifically, corresponding data sets to be predicted are established according to the plates and profiles, respectively, including one or more attribute data in the extracted data.

[0072] The extracted data of the plate includes the X, Y, and Z values ​​of the center of gravity of each plate part and the thickness and material properties of the part. When generating the data to be predicted, only the X, Y, and Z values ​​of the center of gravity are exported, and finally classified as empty values. The thickness and material properties of the parts are stored in the database as comparison values.

[0073] The extracted data of T-profiles include the X, Y, Z values ​​of the center of gravity of each T-profile part, the X, Y, Z coordinates of the starting point, the X, Y, Z coordinates of the end point, the thickness and material properties of the T-profile parts. When generating the data to be predicted, the X, Y, Z values ​​of the center of gravity, the X, Y, Z coordinates of the starting point, and the X, Y, Z coordinates of the end point are exported, and finally classified as null values. The thickness and material properties of the T-profile parts are stored in the database as comparison values.

[0074] S500: Based on the prediction model, the data to be predicted is input to predict the attributes of each category of the plate and profile to obtain the prediction result. For example, the center of gravity position of the plate is input to obtain the material and specification of the plate.

[0075] S600: Compare the predicted result with the physical properties of the plates and profiles extracted from the database, and output the proofreading result. For example, compare the predicted result with the material and thickness of the plates or profiles in the database, analyze the difference value, and submit the proofreading result.

[0076] Through the integration of finite element data and machine learning in the above steps, the finite element file is parsed to extract geometric and material properties and build a structured database; according to the structural characteristics of different areas of the ship (such as cargo holds and engine rooms), the optimal machine learning algorithm is selected according to different areas; the predicted results are automatically compared with the actual data, and a visual report is generated with dynamic proofreading and visual feedback functions. The above method significantly improves the proofreading efficiency and accuracy, and provides good technical support for intelligent ship manufacturing.

[0077] Embodiment 2:

[0078] This embodiment provides a ship modeling and proofreading system based on a finite element model, including a finite element analysis module 21, a training data set generation module 22, a plate machine learning training module 23, a profile machine learning training module 24, a structure model data extraction module 25, a plate automatic proofreading module 26, and a profile automatic proofreading module 27. The functions of each module and the relationship between them are described in detail below.

[0079] Finite element analysis module 21: used to read and analyze finite element files to form a database of ship structure design materials, plates, profiles, etc. based on the finite element model files.

[0080] The training data set generation module 22 generates the plate and profile training data sets based on the finite element analysis module 21 database and the rules of the plate and profile training sets.

[0081] Plate machine learning training module 23: used to carry out machine learning training according to the selected algorithm, import the plate data generated by the training data set generation module 22 and generate a plate prediction model.

[0082] The profile machine learning training module 24 is used to carry out machine learning training according to the selected algorithm, import the profile data generated by the training data set generation module 22, and generate a profile prediction model.

[0083] Structural model data extraction module 25: used to extract the center of gravity, material, and specifications of the plates and profiles used in structural modeling, form a corresponding database, and generate data to be predicted according to the input requirements of the plate machine learning training module 23 and the profile machine learning training module 24.

[0084] Plate automatic proofreading module 26: used to input the plate data to be predicted generated by the structural model data extraction module 25, generate prediction results based on the prediction model of the plate machine learning training module 23, and automatically compare with the plate data extracted from the database of the structural model data extraction module 25 to generate comparison results.

[0085] Profile automatic proofreading module 27: used to input the profile data to be predicted generated by the structural model data extraction module 25, generate prediction results based on the prediction model of the profile machine learning training module 24, and automatically compare with the plate data extracted from the database of the structural model data extraction module 25 to generate comparison results.

[0086] In summary, the ship modeling and proofreading design method and system based on the finite element model provided in this application have high industrial utilization value because they effectively overcome various shortcomings in the prior art.

[0087] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A ship modeling and proofreading design method based on a finite element model, characterized in that: At least the following steps are included: S100: Analyze the finite element model to determine the physical properties of plates and profiles at different center of gravity positions in the ship space and form a database; S200: forming a training data set of attributes of each category according to the physical attributes of the plate and profile; S300: Based on the training data set of each category attribute, select a suitable machine learning algorithm to perform machine learning and generate a prediction model; S400: Extract the center of gravity of plates and profiles in the ship modeling software and sort out the properties to form the data to be predicted; S500: Based on the prediction model, the data to be predicted is input to predict the attributes of each category of the plate and profile, and a prediction result is obtained; S600: Compare the prediction result with the physical properties of the plates and profiles extracted from the database, and output the proofreading result.

2. The ship modeling and proofreading design method based on finite element model according to claim 1 is characterized in that: The physical properties include: plate material properties and thickness properties, profile material properties and specification properties.

3. The ship modeling and proofreading design method based on finite element model according to claim 2 is characterized in that: Step S100 includes: S101: Select a structural finite element file, and obtain the corresponding plate material and thickness, profile material and specification in the finite element code through file analysis to form a material database; S102: parsing the position code in the finite element file, obtaining the position coordinates, and forming a position code coordinate database; S103: parsing the mesh composition code and material code in the finite element file, obtaining mesh coordinates, and forming a plate database; S104: parsing the BEAM component code and material code in the finite element file, obtaining the line segment coordinates, and forming a profile database; S105: Calculate the centroid based on the grid and line segment coordinates, and obtain the material and specification of the plate and profile at the centroid position of each spatial grid.

4. The ship modeling and proofreading design method based on finite element model according to claim 3 is characterized in that: The method for calculating the centroid based on the grid and line segment coordinates is: According to the definition of materials in the start area of ​​the finite element file, the materials are divided into at least plates, T-shaped bars, angle steels, and flat steels; Screen out the plates, find the spatial coordinate definition of each code according to the composition code of each grid of the plates, calculate the centroid of several spatial coordinates, and store the specific material description and the corresponding centroid X, Y, and Z coordinate values ​​obtained by converting the plate material definition code into the database; Filter out one of the materials of T-profile, angle steel, and flat steel, find the CBEAM material description code corresponding to the filtered material, the coordinate code of the starting point and the end point, convert it into a specific material description and specific coordinate value, calculate the centroid according to the coordinate value, and store the material description and the X, Y, and Z coordinate values ​​of the centroid in the database.

5. The ship modeling and proofreading design method based on finite element model according to claim 4 is characterized in that: It also includes data verification steps: According to the length, width and height of the ship, check whether the centroid calculation is within the range; According to the standard range of marine materials, check whether the materials of plates and profiles meet the requirements.

6. The ship modeling and proofreading design method based on finite element model according to claim 2 is characterized in that: In step S200, according to the data stored in the database, the data is classified into plate, T-profile, angle steel, and flat steel, and a plate training set, a T-profile training set, an angle steel training set, and a flat steel training set are generated respectively; wherein, The attributes of the plate training set include the X, Y, and Z coordinate values ​​of the centroid, the thickness and material of the plate; the X, Y, and Z coordinate values ​​are processed as numerical formats, and the thickness and material of the plate are combined as classification processing; The T-profile training set, angle steel training set, and flat steel training set include the X, Y, and Z coordinates of the centroid, the starting point coordinates, the end point coordinates, and the specifications and materials of the profiles. The X, Y, and Z coordinates of the centroid and the starting point coordinates and the end point coordinates are processed as numerical formats, and the specifications and materials of the T-profiles are combined and processed as categories.

7. The ship modeling and proofreading design method based on finite element model according to claim 2 is characterized in that: In step S300, different algorithms are used for training according to different areas of the ship: In the ship cargo hold area, the bagging tree algorithm of the ensemble classifier is selected for machine learning classification; In the tail area of ​​the cabin, the weighted KNN algorithm of the nearest neighbor classifier is selected for machine learning classification; In the upper construction area and bow area, the subspace discrimination algorithm of the integrated classifier is selected for machine learning.

8. The ship modeling and proofreading design method based on finite element model according to claim 7 is characterized in that: In step S300, for T-profiles, a weighted KNN algorithm is used in the ship cargo hold area, and an integrated bagging tree algorithm is used in other areas; for angle steels and flat steels, a medium Gaussian SVM algorithm of a support vector machine is used in the entire ship area.

9. A ship modeling and proofreading system based on a finite element model, characterized in that: include: Finite element analysis module: used to read and analyze finite element files to form a database of materials, plates, and profiles; Training data set generation module: used to generate plate and profile training data sets; Plate machine learning training module: used to train data and generate plate prediction models; Profile machine learning training module: used to train data and generate profile prediction models; Structural model data extraction module: used to extract the center of gravity, material, and specifications of the plates and profiles in structural modeling, form a corresponding database, and generate data to be predicted; Plate automatic proofreading module: used to input plate data to be predicted, generate prediction results, and automatically compare with plate data extracted from the database to generate comparison results; Profile automatic proofreading module: used to input profile data to be predicted, generate prediction results, and automatically compare with profile data extracted from the database to generate comparison results.

Images