Method and system for automatically labeling ship iron outfitting installation information based on AI

Abstract

The invention provides an AI-based ship iron outfitting installation information automatic labeling method and system. The method comprises the steps of constructing a historical model data set, constructing a historical annotation data set and constructing a knowledge base; integrating the historical model data set and the historical annotation data set to form a historical data set, performing data preprocessing on the historical data set to form AI training sample data, training AI models in combination with a knowledge base, and selecting an AI model with the highest prediction precision; collecting a model data set of a target ship, performing data preprocessing, inputting the data to the AI model with the highest prediction precision, and predicting annotation information of the target ship; and performing automatic labeling of the installation information based on the generative tool, the labeling information of the target ship and the three-dimensional design model. The method can overcome the limitation of manual marking, improve the marking efficiency and accuracy, reduce human errors, adapt to the development trend of the three-dimensional modeling technology and improve the automation level of ship design.

Description

Technical Field

[0001] The present invention relates to the technical field of ship design, and in particular to a method and system for automatically marking ship outfitting installation information based on AI. Background Art

[0002] The design and construction of ships is an extremely huge project. In actual operations, in order to achieve more efficient ship construction, detailed information on the installation and process of ship components is usually provided at the design end.

[0003] Taking installation information as an example, the common method is to use dimension annotation or text annotation to reflect the specific installation position of components or models on the ship, and construction personnel build the ship according to the construction drawings. However, the installation information annotation of the current iron rigging model is mainly based on manual annotation, which leads to low design efficiency and cannot meet the growing demand for ship construction.

[0004] Moreover, due to the influence of marking habits and differences in professional knowledge, the expression of installation information will be differentiated, which is extremely unfavorable for the installation and construction work of on-site construction personnel, easily leading to construction errors and reducing the quality and efficiency of ship construction. Summary of the invention

[0005] In view of the defects and deficiencies of the prior art, the present application provides a method and system for automatic labeling of ship outfitting installation information based on AI. The method for automatic labeling of ship outfitting installation information based on AI can reduce the amount of manual labeling, save a lot of time and energy, improve labeling efficiency and speed, improve labeling accuracy, and is conducive to the unification of labeling forms, the standardization of information expression of construction drawings, and the readability of construction drawings.

[0006] An embodiment of the present application provides a method for automatically marking ship outfitting installation information based on AI, comprising the following steps:

[0007] Construct historical model datasets, historical annotation datasets, and knowledge bases;

[0008] Integrate the historical model data set and the historical annotation data set to form a historical data set, and perform data preprocessing on the historical data set to form AI training sample data;

[0009] Based on the AI ​​training sample data and in combination with the knowledge base, the AI ​​model is trained, and the AI ​​model with the highest prediction accuracy is selected through iterative optimization;

[0010] Collecting a model data set of the target ship, performing data preprocessing on the model data set of the target ship, and inputting the data set into the AI ​​model with the highest prediction accuracy to predict the labeling information of the target ship;

[0011] Automatic annotation of installation information is performed based on the generative tool, the annotation information of the target ship, and the three-dimensional design model.

[0012] As an implementation manner, after the step of automatically marking the installation information based on the generative tool, the marking information of the target ship, and the three-dimensional design model, the method further includes:

[0013] Steps for correcting the automatically marked installation information;

[0014] And repeat the steps of building a historical model dataset, building a historical annotation dataset, building a knowledge base, and retraining the AI ​​model to form a new historical dataset and iteratively train the AI ​​model.

[0015] As an implementation method, constructing a historical model data set specifically includes:

[0016] The three-dimensional model data including the ship and its related components are collected to form a historical model data set.

[0017] As an implementation method, constructing a historical annotation dataset specifically includes:

[0018] Collect the annotation data of relevant 3D models in the ship to form a historical annotation dataset.

[0019] As an implementation method, building a knowledge base specifically includes:

[0020] Organize industry standards and senior experts’ experience information to form a knowledge base.

[0021] As an implementation manner, data preprocessing is performed on the historical data set, specifically including:

[0022] Complete missing feature data, correct or delete erroneous feature data, and perform standardization and normalization to improve the generalization ability of the AI ​​model.

[0023] As an implementation method, the AI ​​model with the highest prediction accuracy is selected through iterative optimization, specifically including:

[0024] Selecting an AI algorithm based on the data distribution of the historical data set;

[0025] In addition, the model weights are iteratively optimized by adjusting hyperparameters to improve the accuracy of the AI ​​model in identifying installation standard information.

[0026] As an implementation mode, training an AI model based on the AI ​​training sample data and in combination with the knowledge base includes:

[0027] Based on different training data, different AI models are trained according to ship type, region, and model type.

[0028] As an implementation manner, in the predicted target ship's annotation information, the target ship's annotation information includes:

[0029] The required annotation quantity, annotation type, annotation object, and annotation plane information of the target ship model.

[0030] Another embodiment of the present application provides a system for automatically marking ship outfitting installation information based on AI, including:

[0031] A historical model data set construction module is used to collect three-dimensional model data including the ship and its related components to construct a historical model data set;

[0032] A historical annotation data set construction module is used to collect annotation data of relevant three-dimensional models in the ship to construct a historical annotation data set;

[0033] The knowledge base construction module is used to build a knowledge base based on overall industry norms and senior expert experience information;

[0034] An AI training sample data forming module, used to integrate the historical model data set and the historical annotation data set to form a historical data set, and perform data preprocessing on the historical data set to form AI training sample data;

[0035] An AI model training module, used to train an AI model based on the AI ​​training sample data and in combination with the knowledge base, and select the AI ​​model with the highest prediction accuracy through iterative optimization;

[0036] The prediction module of the target ship's labeling information is used to collect the target ship's model data set and preprocess the target ship's model data set. The preprocessed data is input into the AI ​​model with the highest prediction accuracy to predict the target ship's labeling information.

[0037] The installation information automatic marking module is used to automatically mark the installation information based on the generative tool, the marking information of the target ship, and the three-dimensional design model.

[0038] As described above, the method and system for automatic labeling of ship outfitting installation information based on AI of the present application have the following beneficial effects:

[0039] Compared with the traditional manual marking method, the AI-based method for automatic marking of ship outfitting installation information in the present application uses AI technology to automatically mark ship outfitting installation information, which greatly reduces the labor intensity of staff and saves a lot of manpower costs; through the rapid processing and analysis of data by the AI ​​model, a large number of marking tasks can be completed in a short time, which significantly improves the efficiency of marking work and meets the high-efficiency requirements of modern ship design and construction; AI marking can continuously optimize the model through continuous learning and training, improve the accuracy of identifying installation standard information, and ensure the accuracy of marking information; AI technology can be used to unify the marking format, make the information expression of construction drawings more standardized, improve the readability of drawings, and help relevant personnel understand the content of drawings more clearly and accurately, and promote the smooth progress of ship outfitting installation work. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Shown is a flow chart of a method for automatic labeling of ship outfitting installation information based on AI in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and 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 invention.

[0042] It should be noted that the illustrations provided in this embodiment are only used to illustrate the basic concept of the present invention in a schematic manner, and therefore the illustrations only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0043] The design and construction of ships is a complex and time-consuming process that covers multiple stages from preliminary design to final construction. In this process, it is crucial to ensure the accurate installation of ship components, as this is directly related to the performance, safety and service life of the ship. In order to achieve this goal, the design side needs to provide detailed instructions on the installation information, process information and other construction information of ship components so that construction personnel can accurately understand and execute them.

[0044] In terms of installation information, the traditional approach is to specify the specific installation location of components or models on the ship on the construction drawings through dimensioning or text annotation. Although this annotation method can meet construction needs to a certain extent, it has many shortcomings. First, manual annotation is inefficient, especially when dealing with a large number of complex ship components, the annotation work becomes particularly cumbersome and time-consuming. This not only increases the design cost, but may also delay the construction progress of the ship.

[0045] Secondly, manual marking is greatly affected by differences in marking habits and professional knowledge. Different designers may use different marking methods and symbols, resulting in differences in the expression of installation information. This difference not only increases the difficulty of understanding and execution for construction personnel, but may also cause misunderstandings and errors, thus affecting the construction quality and safety of the ship.

[0046] In addition, with the continuous development of shipbuilding technology, 3D modeling technology has been widely used in ship design. However, the existing installation information annotation method of iron wing models is still mainly manual annotation, which fails to fully utilize the advantages of 3D modeling technology. This not only limits the improvement of design efficiency, but also hinders the improvement of the automation level of ship design.

[0047] In view of the above defects and shortcomings, the present application provides a method and system for automatically marking ship outfitting installation information based on AI, which will be described in detail through the following embodiments.

[0048] This embodiment provides a method for automatically marking ship outfitting installation information based on AI, such as Figure 1 As shown, the following steps are included:

[0049] S1: Build historical model dataset, build historical annotation dataset and build knowledge base;

[0050] S2: Integrate the historical model data set and the historical annotation data set to form a historical data set, perform data preprocessing on the historical data set, and form AI training sample data;

[0051] S3: Based on AI training sample data and combined with the knowledge base, train the AI ​​model and select the AI ​​model with the highest prediction accuracy through iterative optimization;

[0052] S4: Collect the model data set of the target ship, perform data preprocessing on the model data set of the target ship, and input it into the AI ​​model with the highest prediction accuracy to predict the labeling information of the target ship;

[0053] S5: Automatically annotate the installation information based on the generative tool, the annotation information of the target ship, and the three-dimensional design model.

[0054] Compared with the traditional manual marking method, the AI-based method for automatic marking of ship outfitting installation information provided in this embodiment greatly reduces the labor intensity of staff and saves a lot of manpower costs by using AI technology to automatically mark ship outfitting installation information; through the rapid processing and analysis of data by the AI ​​model, a large number of marking tasks can be completed in a short time, which significantly improves the efficiency of marking work and meets the high-efficiency requirements of modern ship design and construction; AI marking can continuously optimize the model through continuous learning and training, improve the accuracy of identifying installation standard information, and thus ensure the accuracy of standard information; AI technology can be used to unify the marking format, make the information expression of construction drawings more standardized, improve the readability of drawings, and help relevant personnel understand the content of drawings more clearly and accurately, and promote the smooth progress of ship outfitting installation work.

[0055] In an optional embodiment, the method for automatically marking ship outfitting installation information based on AI further includes the following steps:

[0056] S6: Correct the automatically marked installation information;

[0057] And repeat steps S1 to S3 to reconstruct a new historical model data set, a historical annotation data set and a knowledge base to form a new historical data set, and extract a new historical data set for AI model training.

[0058] Step S1 includes step S11 of constructing a historical model data set, step S12 of constructing a historical annotation data set, and step S13 of constructing a knowledge base.

[0059] Among them, the historical model dataset, historical annotation dataset, and historical knowledge base are formed based on the collection and organization of existing information.

[0060] Step S11 specifically includes: collecting three-dimensional model data including the ship and its related components to form a historical model data set.

[0061] Specifically, the historical model data set includes: ship data, ship design coordinate system information, 3D model data, model type, spatial range, and shape feature data. Ship data includes coordinate information of shipyard, ship width, midship, main deck, etc. 3D model data includes spatial position, center of gravity, etc.

[0062] Step S12 specifically includes: collecting the annotation data of the relevant three-dimensional model in the ship to form a historical annotation data set.

[0063] Specifically, the historical annotation data set includes: annotation type, annotation information, annotation object, spatial coordinates of the annotation object, annotation view plane, etc.

[0064] Step S13 specifically includes: collating industry standards and senior expert experience information to form a knowledge base.

[0065] Specifically, relevant industry standards and specifications are collected and organized, and combined with the experience of senior experts to form a structured knowledge base.

[0066] The data preprocessing of the historical data set in step S2 specifically includes: completing missing feature data, correcting or deleting erroneous feature data, standardizing and normalizing the data, etc., so as to improve the generalization ability of the AI ​​model.

[0067] Step S3 specifically includes: selecting a suitable AI algorithm according to the data distribution of the historical data set, inputting the historical data set preprocessed in step S2 into the AI ​​algorithm, and performing AI training in combination with the information in the historical knowledge base; and iteratively optimizing the model weights by adjusting the hyperparameters to improve the accuracy of the model in identifying installation standard information and improve the prediction results of the AI ​​model.

[0068] The AI ​​algorithm may be, for example, a convolutional neural network (CNN), a recurrent neural network (RNN), etc.

[0069] Step S3 also includes: based on different training data, different AI models can be trained according to ship type, region, and model type.

[0070] In step S4, a model data set of a target ship is collected, which specifically includes: collecting three-dimensional model data of the target ship and installation components to form a model data set of the target ship. The target ship refers to a newly designed ship.

[0071] The model data set of the target ship includes: ship data, ship design coordinate system information, 3D model data, model type, spatial range, shape feature data, etc. Ship data can be, for example, coordinate information such as shipyard, ship width, midship, main deck, etc.; 3D model data can be, for example, spatial position, center of gravity, etc.

[0072] In step S4, the model data set of the target ship is preprocessed, specifically including: completing missing feature data, correcting or deleting erroneous feature data, standardizing and normalizing the data, etc., so as to improve the generalization ability of the AI ​​model.

[0073] The process of predicting the labeling information of the target ship in step S4 includes: selecting the AI ​​model with the highest prediction accuracy in step S3, inputting the model data set of the target ship preprocessed in step S4 into the AI ​​model with the highest prediction accuracy, and predicting the labeling information of the target ship.

[0074] The annotation information of the target ship predicted in step S4 includes: the annotation quantity, annotation type, annotation object and annotation plane information required for the target ship model.

[0075] Step S5 specifically includes automatically labeling the installation information in sequence based on the labeling information of the target ship predicted by the AI ​​model and the three-dimensional design model based on the generation tool (such as design software CAD).

[0076] In step S6, the automatically marked installation information is corrected, specifically including: manually confirming and correcting the predicted marking information of the target ship.

[0077] This embodiment also provides a system for automatically marking ship outfitting installation information based on AI, including:

[0078] A historical model data set construction module is used to collect three-dimensional model data including the ship and its related components to construct a historical model data set;

[0079] A historical annotation data set construction module is used to collect annotation data of relevant three-dimensional models in the ship to construct a historical annotation data set;

[0080] The knowledge base construction module is used to build a knowledge base based on overall industry norms and senior expert experience information;

[0081] An AI training sample data forming module is used to integrate the historical model data set and the historical annotation data set to form a historical data set, and perform data preprocessing on the historical data set to form AI training sample data;

[0082] An AI model training module, used to train an AI model based on AI training sample data and in combination with the knowledge base, and select an AI model with the highest prediction accuracy through iterative optimization;

[0083] The prediction module of the target ship's annotation information is used to collect the target ship's model data set and preprocess the target ship's model data set. The preprocessed data is input into the AI ​​model with the highest prediction accuracy to predict the target ship's annotation information.

[0084] The installation information automatic annotation module is used to automatically annotate the installation information based on the generative tool, the predicted annotation information of the target ship, and the three-dimensional design model.

[0085] The AI-based automatic labeling system for ship outfitting installation information provided in this embodiment can overcome the limitations of manual labeling, improve labeling efficiency and accuracy, reduce human errors, and at the same time adapt to the development trend of three-dimensional modeling technology and improve the level of automation in ship design.

[0086] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A method for automatic labeling of ship outfitting installation information based on AI, characterized in that: The following steps are involved: Construct historical model datasets, historical annotation datasets, and knowledge bases; Integrate the historical model data set and the historical annotation data set to form a historical data set, and perform data preprocessing on the historical data set to form AI training sample data; Based on the AI ​​training sample data and in combination with the knowledge base, the AI ​​model is trained, and the AI ​​model with the highest prediction accuracy is selected through iterative optimization; Collecting a model data set of the target ship, performing data preprocessing on the model data set of the target ship, and inputting the data set into the AI ​​model with the highest prediction accuracy to predict the labeling information of the target ship; Automatic annotation of installation information is performed based on the generative tool, the annotation information of the target ship, and the three-dimensional design model.

2. The method according to claim 1, characterized in that After the step of automatically marking the installation information based on the generative tool, the marking information of the target ship, and the three-dimensional design model, the method further includes: Steps for correcting the automatically marked installation information; And repeat the steps of building a historical model dataset, building a historical annotation dataset, building a knowledge base, and retraining the AI ​​model to form a new historical dataset and iteratively train the AI ​​model.

3. The method according to claim 1, characterized in that: Construct a historical model dataset, including: The three-dimensional model data including the ship and its related components are collected to form a historical model data set.

4. The method according to claim 1, characterized in that Construct a historical annotation dataset, including: Collect the annotation data of relevant 3D models in the ship to form a historical annotation dataset.

5. The method according to claim 1, characterized in that Build a knowledge base, including: Organize industry standards and senior experts’ experience information to form a knowledge base.

6. The method according to claim 1, characterized in that The historical data set is subjected to data preprocessing, specifically including: Complete missing feature data, correct or delete erroneous feature data, and perform standardization and normalization to improve the generalization ability of the AI ​​model.

7. The method according to claim 1, characterized in that The AI ​​model with the highest prediction accuracy is selected through iterative optimization, including: Selecting an AI algorithm based on the data distribution of the historical data set; Also, by adjusting the hyperparameters, iteratively optimize the model weights to improve the accuracy of the AI ​​model in identifying installation standard information. Rate.

8. The method according to claim 1, characterized in that Based on the AI ​​training sample data and in combination with the knowledge base, training the AI ​​model includes: Based on different training data, different AI models are trained according to ship type, region, and model type.

9. The method according to claim 1, characterized in that: The target ship's annotation information is predicted to include: The required annotation quantity, annotation type, annotation object, and annotation plane information of the target ship model.

10. An AI-based system for automatically marking ship outfitting installation information, characterized in that: include: A historical model data set construction module is used to collect three-dimensional model data including the ship and its related components to construct a historical model data set; A historical annotation data set construction module is used to collect annotation data of relevant three-dimensional models in the ship to construct a historical annotation data set; The knowledge base construction module is used to build a knowledge base based on overall industry norms and senior expert experience information; An AI training sample data forming module, used to integrate the historical model data set and the historical annotation data set to form a historical data set, and perform data preprocessing on the historical data set to form AI training sample data; An AI model training module, used to train an AI model based on the AI ​​training sample data and in combination with the knowledge base, and select the AI ​​model with the highest prediction accuracy through iterative optimization; The prediction module of the target ship's annotation information is used to collect the target ship's model data set and pre-process the target ship's model data set. The pre-processed data is input into the AI ​​model with the highest prediction accuracy. Measure the marking information of the target ship; The installation information automatic marking module is used to automatically mark the installation information based on the generative tool, the marking information of the target ship, and the three-dimensional design model.

Images