Design method and system for novel krill boat special bulbous bow

Abstract

The invention relates to the technical field of ship design and manufacturing, in particular to a novel method and system for designing a bulbous bow special for a krill ship. According to the main technical scheme, the method comprises the steps of obtaining multi-modal data required by design; processing the text to generate a to-be-input condition matrix; processing the image information to generate a mask matrix to be input; processing the audio to generate a randomly generated pure noise matrix; obtaining a pre-trained ship bulbous bow design diffusion model; and inputting the condition matrix to be input, the mask matrix to be input and a randomly generated pure noise matrix into a ship bulbous bow design diffusion model to generate a ship bulbous bow design matrix. The system comprises a data acquisition module, a mask matrix generation module, a condition matrix generation module, a model training model and a design scheme generation module. The method has the advantages of being high in design efficiency and design reliability, short in research and development period and low in cost, ship manufacturing is more intelligent, efficient, energy-saving and environment-friendly, and the method is worthy of application and popularization.

Description

Technical Field

[0001] The present invention relates to the technical field of ship design and manufacturing, and in particular to a design method and system for a novel bulbous bow specially used for krill boats. Background Art

[0002] Antarctic krill fishing vessels are important equipment for Antarctic pelagic fishing. Choosing the right bow shape is crucial in ship design, as it directly impacts the vessel's performance and operating efficiency. The bulbous bow, a key feature of ship design, features a smooth, spherical or hemispherical bow. Compared to traditional pointed or flat-bottomed bows, the bulbous bow's front is smoother, lacks sharp edges, and offers a more streamlined shape. The bulbous bow's rounded shape reduces water separation and vortex formation, thereby reducing drag, minimizing the vessel's resistance in the water and the impact of waves, increasing stability, and improving navigation performance and economic efficiency. It is widely used in commercial and scientific research vessels.

[0003] The design of the bulbous bow is based on the principles of fluid mechanics and also needs to consider the ship's structure and hull stability. The shape and size of the bow should be coordinated with the rest of the hull to ensure the structural stability of the entire hull and the ship's maneuverability. At the same time, during the design process, the production cost and manufacturing process of the bulbous bow also need to be comprehensively considered to ensure the feasibility and economy of the final design. The application prospects of bulbous bow technology in ship engineering are broad, but it also faces some challenges. The design of the bulbous bow needs to comprehensively consider multiple factors such as the ship's navigation performance, structural strength and stability, requiring ship designers to have a high level of engineering technology.

[0004] In recent years, with the rapid development of machine learning and artificial intelligence algorithms, machine learning has been increasingly applied to various technical fields. One notable model category is diffusion models, which have gained attention for their ability to capture and simulate complex processes such as data generation and image synthesis. As an advanced generative model, diffusion models have become a key advancement in machine learning over the past few years. Multimodality refers to the use of information from multiple different forms or perceptual channels for expression, communication, and understanding, typically including multiple sensory input and output methods such as vision, hearing, text, and touch. In computer science, artificial intelligence, and machine learning, multimodal technology refers to the integration of data from different modalities, such as images, text, audio, and video, to enhance the model's understanding and reasoning capabilities. Whether the design of a ship's bulbous bow can incorporate machine learning-based multimodal fusion diffusion models to improve design efficiency and effectiveness is a question worthy of research. Currently, there are no reports on the application of multimodal fusion diffusion models to ship bulbous bow design. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to propose a new design method and system for a bulbous bow specially designed for krill boats, so as to improve design efficiency, enhance design reliability and reduce R&D costs.

[0006] To solve the above technical problems, the present invention proposes a novel design method for a bulbous bow specifically for krill boats, comprising the following steps: Obtain natural language text of design requirements, ship bulbous bow design drawings, and bulbous bow acoustic signals during ship navigation; Extract key text information from the natural language text, perform matrix processing on the information, and generate a condition matrix to be input; Extracting key image information from the ship bulbous bow design drawing, performing matrix processing on the information, and generating a mask matrix to be input; extracting key audio information from an acoustic signal of the bulbous bow of the vessel when the vessel is sailing, performing matrix processing on the information, and generating a randomly generated pure noise matrix; The pre-trained ship bulbous bow design diffusion model is obtained by training based on the conditional matrix and mask matrix sample data and the pre-calibrated ship bulbous bow design matrix label data; The condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix are input into a pre-trained ship bulbous bow design diffusion model to generate a ship bulbous bow design matrix.

[0007] The condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix are input into the pre-trained ship bulbous bow diffusion model to generate the ship bulbous bow design matrix, which specifically includes: The randomly generated pure noise matrix S0 and the conditional matrix C are input into the ship bulbous bow diffusion model to obtain the noise matrix N1. The noise matrix N1 is multiplied element by element with the mask matrix M, and then the pure noise matrix S0 is subtracted element by element with the above result to obtain the subsequent denoising matrix S1; Replace the pure noise matrix with the denoising matrix, repeat the above process N-2 times, and continuously calculate the latest denoising matrix; The denoising matrix S obtained in step N-1 is N-1 The condition matrix C is input into the ship bulbous bow diffusion model to obtain the noise matrix N N , the noise matrix N N Multiply element-by-element with the mask matrix M, and then convert the denoising matrix S N-1 Subtract element by element from the above result to obtain the ship bulbous bow design matrix.

[0008] Preferably, the natural language text includes the hard conditions of laws and regulations and the flexible design conditions required by the shipowner and the shipyard construction party.

[0009] The ship bulbous bow design drawings include bulbous bow lines, bulbous bow structural parts, bulbous bow welding parts, bulbous bow frame-core structure, and bulbous bow frame-support structure; The acoustic signal of the bulbous bow of the ship when sailing includes a vibration signal of the bulbous bow of the ship when sailing and an acoustic signal of the surrounding environment.

[0010] Furthermore, key image information is extracted from the ship bulbous bow design drawing, and the information is matrixed to generate a mask matrix to be input, specifically including: Preprocess the bulbous bow design drawings to obtain the coordinates, materials, construction process and other attributes corresponding to the key image information; initialize the single-layer condition matrix and mask matrix with all elements set to zero; For the spatial position information, geometric size information, and topological connectivity information in the key image information, the element positions on the single-layer condition matrix are located according to the coordinates, and the elements are assigned values according to the attributes to obtain several single-layer condition matrices; For the potential structure position information in the key image information, the element position on the mask matrix is located according to the coordinates, and the elements are assigned values according to the attributes to obtain the mask matrix.

[0011] Furthermore, key text information is extracted from the natural language text, and the information is matrixed to generate a condition matrix to be input, specifically including: Input the design requirements and operation instructions in natural language into the large language model enhanced with ship domain knowledge to obtain the attributes corresponding to the key text information; According to the pre-specified mapping relationship, the design requirement vector is obtained based on the attributes corresponding to the key text information; By copying and stacking, the size of the design requirement vector is expanded to be consistent with the single-layer condition matrix, and several single-layer condition matrices are obtained; The condition matrix is obtained by stacking all single-layer condition matrices corresponding to the key image information and the key text information.

[0012] Optimally, the pre-trained ship bulbous bow design diffusion model is obtained after training based on the condition matrix and mask matrix sample data and the pre-calibrated bulbous bow structure matrix label data, specifically including: For any iteration, based on the bulbous bow structure layout matrix label data S and the predefined noise level change law, the nth step is randomly selected and the real denoising matrix S corresponding to the noise level of the nth step is sampled. ′ n-1 and the true noise matrix N ′ n ; The real denoising matrix S ′ n-1The condition matrix C is input into the ship bulbous bow design diffusion model to obtain the predicted noise matrix N n By adjusting the parameters of the diffusion model for the bulbous bow design of ships, the prediction noise matrix N is minimized. n and the true noise matrix N ′ n Repeat the above iterative process until the predicted noise matrix N n and the true noise matrix N ′ n The difference cannot be reduced any further.

[0013] The present invention also provides a system based on the design method of the novel bulbous bow for krill boats, comprising: A data acquisition module is used to obtain the bulbous bow design drawings of the ship to be processed and the design requirement text in natural language form; a mask matrix generation module, configured to extract key image information from the ship's bulbous bow design drawing, perform matrix processing on the information, and generate a mask matrix to be input; A condition matrix generation module is used to extract key text information from the design requirement text, perform matrix processing on the information, and generate a condition matrix to be input; The model training model is trained based on the condition matrix and mask matrix sample data and the pre-calibrated bulbous bow structure layout matrix label data to obtain the pre-trained ship bulbous bow design diffusion model; The design scheme generation module is used to input the condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix into the pre-trained ship bulbous bow design diffusion model to generate the ship bulbous bow design matrix.

[0014] The design method and system of the novel bulbous bow for krill boats of the present invention have the following significant advantages: 1. Improve design efficiency Through automated and intelligent design processes, the time from concept to finished product can be greatly shortened. Designers no longer need to rely on traditional trial and error methods or limited experience, but can quickly generate multiple design options and evaluate their performance.

[0015] 2. Enhance design reliability Utilizing multimodal data and advanced machine learning algorithms, the impact of bulbous bows on ship performance can be simulated and predicted more accurately.

[0016] 3. Reduce R&D costs By reducing the design time and cost of the model, the overall R&D cost is reduced. At the same time, through virtual simulation testing and simulation calculations, potential problems can be identified at an early stage.

[0017] 4. Data-driven decision making The entire design process is built on high-quality data, ensuring that every decision is well-founded. Furthermore, as actual operational data accumulates, the model can further improve its predictive capabilities and design quality through continuous learning.

[0018] In summary, the present invention not only improves the speed and accuracy of design, but also makes shipbuilding more intelligent, efficient, energy-saving and environmentally friendly. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0020] Figure 1 It is an overall flow chart of the design method of the present invention.

[0021] Figure 2 This is a comparison chart of the ship sailing resistance simulation effects of the present invention and the prior art.

[0022] Figure 3 The figure is a comparison chart of the design cycle histograms of the present invention and the prior art. DETAILED DESCRIPTION

[0023] Diffusion models are generative machine learning models originally developed to address image denoising. They gradually add noise to data and then train the model to reverse this process, gradually generating target data from pure noise. This model effectively captures data distribution and generates high-quality samples. Multimodal fusion integrates data from different modes to provide richer and more comprehensive information. Multimodal data refers to data collected through text, images, audio, and other sources.

[0024] In the design of a bulbous bow, the present invention utilizes a multimodal fusion diffusion model to achieve the desired design objectives. Selecting appropriate and effective design data is a key task. The present invention integrates multimodal data, simultaneously taking into account text, images, and audio, to achieve the optimal bulbous bow design. The text considered includes information describing the ship's performance, design parameters, and material properties; the images include drawings such as the ship's exterior, internal structure, and construction drawings; and the audio includes the sounds of wind, waves, and currents surrounding the ship.

[0025] Combine Figure 1 As shown, the design method of the novel bulbous bow for krill boats of the present invention comprises the following steps: Step i: Data collection and preparation, including: obtaining natural language text of design requirements, ship bulbous bow design drawings and bulbous bow acoustic signals of ship navigation.

[0026] The natural language text of the design requirements includes the rigid requirements of laws and regulations, as well as the flexible design requirements required by shipowners and shipyard construction parties. The required bulbous bow design drawings include the bulbous bow lines, bulbous bow structural components, bulbous bow weldments, bulbous bow frame-core structure, and bulbous bow frame-support structure. The required bulbous bow acoustic signals during navigation include the vibration signals of the bulbous bow and the acoustic signals of the surrounding environment.

[0027] Step ii: Multimodal data analysis, including: ii_1. Extract key text information from the natural language text, perform matrix processing on the information, and generate a matrix of conditions to be input, specifically including: Input the design requirements and operation instructions in natural language into the large language model enhanced with ship domain knowledge to obtain the attributes corresponding to the key text information; According to the pre-specified mapping relationship, the design requirement vector is obtained based on the attributes corresponding to the key text information; By copying and stacking, the size of the design requirement vector is expanded to be consistent with the single-layer condition matrix, and several single-layer condition matrices are obtained; The condition matrix is obtained by stacking all single-layer condition matrices corresponding to the key image information and the key text information.

[0028] Step ii_2, extracting key image information from the ship bulbous bow design drawing, matrixing the information, and generating a mask matrix to be input, specifically comprising: Preprocess the bulbous bow design drawings to obtain the coordinates, materials, construction process and other attributes corresponding to the key image information; initialize the single-layer condition matrix and mask matrix with all elements set to zero; For the spatial position information, geometric size information, and topological connectivity information in the key image information, the element positions on the single-layer condition matrix are located according to the coordinates, and the elements are assigned values according to the attributes to obtain several single-layer condition matrices; For the potential structure position information in the key image information, the element position on the mask matrix is located according to the coordinates, and the elements are assigned values according to the attributes to obtain the mask matrix.

[0029] Step ii_3, extracting key audio information from the acoustic signal of the bulbous bow when the ship is sailing, and performing matrix processing on the information to generate a randomly generated pure noise matrix; Step iii: Model training and verification. Based on the conditional matrix and mask matrix sample data and the pre-calibrated bulbous bow structure layout matrix label data, a pre-trained ship bulbous bow design diffusion model is obtained, which specifically includes: For any iteration, based on the bulbous bow structure layout matrix label data S and the predefined noise level change law, the nth step is randomly selected and the real denoising matrix S corresponding to the noise level of the nth step is sampled. ′ n-1 and the true noise matrix N ′ n ; The real denoising matrix S ′ n-1 The condition matrix C is input into the ship bulbous bow design diffusion model to obtain the predicted noise matrix N n ; By adjusting the parameters of the diffusion model for the bulbous bow design of ships, the prediction noise matrix N is minimized. n and the true noise matrix N ′ n differences; Repeat the above iterative process until the predicted noise matrix N n and the true noise matrix N ′ n The difference cannot be reduced any further.

[0030] Step iv, design scheme generation. Input the condition matrix to be input, the mask matrix to be input, and the randomly generated pure noise matrix into the pre-trained ship bulbous bow design diffusion model to generate the ship bulbous bow design matrix, specifically including: The randomly generated pure noise matrix S0 and the conditional matrix C are input into the ship bulbous bow diffusion model to obtain the noise matrix N1. The noise matrix N1 is multiplied element by element with the mask matrix M, and then the pure noise matrix S0 is subtracted element by element with the above result to obtain the subsequent denoising matrix S1; Replace the pure noise matrix with the denoising matrix, repeat the above process N-2 times, and continuously calculate the latest denoising matrix; The denoising matrix S obtained in step N-1 is N-1 The condition matrix C is input into the ship bulbous bow diffusion model to obtain the noise matrix N N , the noise matrix N N Multiply element-by-element with the mask matrix M, and then convert the denoising matrix S N-1 Subtract element by element from the above result to obtain the ship bulbous bow design matrix.

[0031] Step v: Performance Evaluation and Iteration. The bulbous bow obtained by applying the bulbous bow design matrix is evaluated through actual or simulation verification. The training model parameters are improved or optimized based on the evaluation results. Furthermore, as actual operational data accumulates, the model can further improve its predictive capabilities and design quality through continuous learning, achieving performance iteration.

[0032] The present invention also proposes a system based on the design method of the novel bulbous bow for krill boats, comprising: A data acquisition module is used to obtain the bulbous bow design drawings of the ship to be processed and the design requirement text in natural language form; a mask matrix generation module, configured to extract key image information from the ship's bulbous bow design drawing, perform matrix processing on the information, and generate a mask matrix to be input; A condition matrix generation module is used to extract key text information from the design requirement text, perform matrix processing on the information, and generate a condition matrix to be input; A model training module is used to train the sample data based on the conditional matrix and the mask matrix and the label data of the pre-calibrated bulbous bow structure layout matrix to obtain a pre-trained ship bulbous bow design diffusion model; The design scheme generation module is used to input the condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix into the pre-trained ship bulbous bow design diffusion model to generate the ship bulbous bow design matrix.

[0033] like Figure 2 As shown, Figure 2 The upper part is a simulation diagram of the sailing resistance of a bulbous bow designed using traditional methods. Figure 2 The lower half is a simulation diagram of the sailing resistance of the bulbous bow of a ship designed with the present invention. It can be clearly seen that the wave-making resistance of the bulbous bow ship designed with the present invention is reduced by 5-20%.

[0034] Due to the reduced drag, based on current experience, it is estimated that fuel consumption can be reduced by 2%-10%. Therefore, fuel economy is improved.

[0035] like Figure 3 As shown, Figure 3 The vertical axis represents the time required for design (according to industry practice, ship design time is measured in days). The bar chart on the left represents the design cycle required to design a bulbous bow using traditional methods. Figure 3 The bar graph on the right shows the design cycle required for the bulbous bow of a ship using the present invention. The time from concept to finished product is reduced by 30%-50%, thereby improving design efficiency.

[0036] From the perspective of economic benefits, the use of this invention in the design of a bulbous bow of a ship benefits from the automated process and rapid iteration capabilities, which reduces testing and labor costs. Therefore, the overall R&D cost is expected to be reduced by 10%-30%, and the economic benefits are significantly improved.

[0037] Finally, it should be noted that the above specific implementation methods are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A new design method for a bulbous bow for krill boats, characterized in that: The following steps are involved: Obtain natural language text of design requirements, ship bulbous bow design drawings, and bulbous bow acoustic signals during ship navigation; Extract key text information from the natural language text, perform matrix processing on the information, and generate a condition matrix to be input; Extracting key image information from the ship bulbous bow design drawing, performing matrix processing on the information, and generating a mask matrix to be input; extracting key audio information from an acoustic signal of the bulbous bow of the vessel when the vessel is sailing, performing matrix processing on the information, and generating a randomly generated pure noise matrix; The pre-trained ship bulbous bow design diffusion model is obtained by training based on the conditional matrix and mask matrix sample data and the pre-calibrated ship bulbous bow design matrix label data; The condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix are input into a pre-trained ship bulbous bow design diffusion model to generate a ship bulbous bow design matrix.

2. The design method of the new bulbous bow for krill boats according to claim 1 is characterized in that: The step of inputting the condition matrix to be input, the mask matrix to be input, and the randomly generated pure noise matrix into the pre-trained ship bulbous bow diffusion model to generate the ship bulbous bow design matrix specifically includes: The randomly generated pure noise matrix S0 and conditional matrix C are input into the ship bulbous bow diffusion model to obtain the noise matrix N1; Multiply the noise matrix N1 by the mask matrix M element by element, and then subtract the pure noise matrix S0 from the above result element by element to obtain the denoising matrix S1 used in the subsequent process. Replace the pure noise matrix with the denoising matrix, repeat the above process N-2 times, and continuously calculate the latest denoising matrix; The denoising matrix S obtained in step N-1 is N-1 The condition matrix C is input into the ship bulbous bow diffusion model to obtain the noise matrix N N ; The noise matrix N N Multiply element-by-element with the mask matrix M, and then convert the denoising matrix S N-1 Subtract element by element from the above result to obtain the ship bulbous bow design matrix.

3. The design method of the new bulbous bow for krill boats according to claim 1 is characterized in that: The natural language text includes the rigid requirements of laws and regulations and the flexible design requirements required by the shipowner and the shipyard construction party; The ship bulbous bow design drawings include bulbous bow lines, bulbous bow structural parts, bulbous bow welding parts, bulbous bow frame-core structure, and bulbous bow frame-support structure; The acoustic signal of the bulbous bow of the ship when sailing includes a vibration signal of the bulbous bow of the ship when sailing and an acoustic signal of the surrounding environment.

4. The design method of the new bulbous bow for krill boats according to claim 1 is characterized in that: Extracting key image information from the ship's bulbous bow design drawing, performing matrix processing on the information, and generating a mask matrix to be input, specifically includes: Preprocess the bulbous bow design drawings to obtain the coordinates, material, and construction process attributes corresponding to the key image information; initialize a single-layer condition matrix and mask matrix with all elements set to zero; For the spatial position information, geometric size information, and topological connectivity information in the key image information, the element positions on the single-layer condition matrix are located according to the coordinates, and the elements are assigned values according to the attributes to obtain several single-layer condition matrices; For the potential structure position information in the key image information, the element position on the mask matrix is located according to the coordinates, and the elements are assigned values according to the attributes to obtain the mask matrix.

5. The design method of the new bulbous bow for krill boats according to claim 1 is characterized in that: Extracting key text information from the natural language text, performing matrix processing on the information, and generating a matrix of conditions to be input specifically include: Input the design requirements and operation instructions in natural language into the large language model enhanced with ship domain knowledge to obtain the attributes corresponding to the key text information; According to the pre-specified mapping relationship, the design requirement vector is obtained based on the attributes corresponding to the key text information; By copying and stacking, the size of the design requirement vector is expanded to be consistent with the single-layer condition matrix, and several single-layer condition matrices are obtained; The condition matrix is obtained by stacking all single-layer condition matrices corresponding to the key image information and the key text information.

6. The design method of the new bulbous bow for krill boats according to claim 1 is characterized in that: The pre-trained ship bulbous bow design diffusion model is obtained after training based on the condition matrix and mask matrix sample data and the pre-calibrated bulbous bow structure matrix label data, specifically including: For any iteration, based on the bulbous bow structure layout matrix label data S and the predefined noise level change law, the nth step is randomly selected and the real denoising matrix S corresponding to the noise level of the nth step is sampled. ′ n-1 and the true noise matrix N ′ n ; The real denoising matrix S ′ n-1 The condition matrix C is input into the ship bulbous bow design diffusion model to obtain the predicted noise matrix N n ; By adjusting the parameters of the diffusion model for the bulbous bow design of ships, the prediction noise matrix N is minimized. n and the true noise matrix N ′ n differences; Repeat the above iterative process until the predicted noise matrix N n and the true noise matrix N ′ n The difference cannot be reduced any further.

7. A system based on the design method of the novel bulbous bow for krill boats according to claim 1, characterized in that: include: A data acquisition module is used to obtain the bulbous bow design drawings of the ship to be processed and the design requirement text in natural language form; a mask matrix generation module, configured to extract key image information from the ship's bulbous bow design drawing, perform matrix processing on the information, and generate a mask matrix to be input; A condition matrix generation module is used to extract key text information from the design requirement text, perform matrix processing on the information, and generate a condition matrix to be input; The model training model is trained based on the condition matrix and mask matrix sample data and the pre-calibrated bulbous bow structure layout matrix label data to obtain the pre-trained ship bulbous bow design diffusion model; The design scheme generation module is used to input the condition matrix to be input, the mask matrix to be input and the randomly generated pure noise matrix into the pre-trained ship bulbous bow design diffusion model to generate the ship bulbous bow design matrix.

Images