A model construction method and device based on ship type optimization parameter abstract feature representation

Abstract

The application provides a model construction method and device based on ship type optimization parameter abstract feature representation, and relates to the technical field of three-dimensional modeling. The method comprises the following steps: encoding and extracting features of a first dimension ship type optimization parameter to obtain a second dimension abstract feature representation, wherein the second dimension is lower than the first dimension; taking the abstract feature representation as an optimization variable of a ship type optimization problem, and constructing a mathematical model of ship related performance based on the optimization variable; using the mathematical model to replace a numerical simulation tool to participate in a ship type optimization process until an optimization result in the form of abstract feature representation is obtained; reversely mapping and decoding the optimization result into actual ship type optimization parameters; and generating a corresponding optimized ship type through parameterization according to the ship type optimization parameters. The application can improve the modeling efficiency in a high-dimensional parameter space.

Description

Technical Field

[0001] This application relates to the technical field of 3D modeling, specifically to a model construction method and apparatus based on the abstract feature representation of ship hull optimization parameters. Background Technology

[0002] In existing technologies, ship hull optimization typically employs a numerical simulation-based optimization process. This involves directly using high-dimensional hull optimization parameters as input variables and calculating and evaluating ship performance through computational fluid dynamics simulations, structural analysis, or multi-objective evaluation models. However, with the increasing dimensionality of optimization parameters, the number of samples required to cover the parameter space rises exponentially. This often results in thousands to tens of thousands of highly complex simulation calculations per optimization cycle, leading to enormous cumulative simulation time costs, long optimization periods, and severe computational resource consumption. Therefore, existing technologies suffer from significant bottlenecks in modeling efficiency within high-dimensional parameter spaces. There is an urgent need to introduce techniques with dimensionality reduction capabilities to construct low-dimensional yet efficient performance prediction models, thereby overcoming the limitations of traditional optimization systems in large-scale design applications. Summary of the Invention

[0003] This application provides a model construction method and apparatus based on the abstract feature representation of ship hull optimization parameters, which can improve the efficiency of modeling in high-dimensional parameter spaces.

[0004] The first aspect of this application provides a model construction method based on the abstract feature representation of ship morphology optimization parameters, the method comprising:

[0005] The ship morphology optimization parameters of the first dimension are encoded and features are extracted to obtain the abstract feature representation of the second dimension, wherein the second dimension is lower than the first dimension;

[0006] The abstract features are represented as optimization variables in the ship type optimization problem, and a mathematical model of ship-related performance is constructed based on the optimization variables.

[0007] The mathematical model is used to replace numerical simulation tools in the ship type optimization process until the optimization result is obtained in terms of abstract features;

[0008] The optimization results are then back-mapped and decoded into actual ship type optimization parameters.

[0009] Based on the ship type optimization parameters, the corresponding optimized ship type is generated through parameterization.

[0010] Based on the above technical solutions, preferably, the step of encoding and extracting features from the first-dimensional ship type optimization parameters to obtain the abstract feature representation of the second dimension specifically includes:

[0011] The structure of the encoding network is defined, and the data samples in the first-dimensional parameter space are encoded through nonlinear mapping and nonlinear activation functions. The structural expression of the encoding network is as follows:

[0012] H = g(WX + b) m )

[0013] Where H is the abstract feature representation, H∈R^(k×n), R is the real number field, k is the second dimension, n is the number of samples, g(·) is the nonlinear activation function, W is the weight matrix, and b m Let X be the bias vector, and let X be the original input corresponding to the data sample in the first dimension parameter space, X∈R^(d×n), k is less than d, and d is the first dimension.

[0014] Based on the above technical solutions, preferably, after encoding and extracting features from the first-dimensional ship type optimization parameters to obtain the abstract feature representation of the second dimension, the method further includes:

[0015] A loss function is introduced to measure the error between the abstract feature representation and the original input. This loss function includes a squared error loss function and a cross-entropy loss function. The expression for the squared error loss function is:

[0016]

[0017] The expression for the cross-entropy loss function is:

[0018]

[0019] in, x is the error value. i For the original input, This is a representation of the abstract feature;

[0020] During training, by minimizing the loss function, the weight matrix and the bias vector are optimized so that the resulting abstract feature representation is close to the essential features of the ship hull optimization parameters in the first dimension in terms of information expression.

[0021] Based on the above technical solutions, preferably, the step of back-mapping and decoding the optimization result into actual ship type optimization parameters specifically includes:

[0022] The inverse mapping from the optimization result to the ship type optimization parameters is completed by linear mapping, and the actual ship type optimization parameters are obtained.

[0023] The reverse mapping decoding process is optimized by training the output weight matrix contained in the linear mapping, with the goal of minimizing the difference between the optimization result and the ship type optimization parameters.

[0024] Based on the above technical solutions, preferably, the step of representing the abstract features as optimization variables in the ship type optimization problem, and constructing a mathematical model of ship-related performance based on the optimization variables, specifically includes:

[0025] Using the abstract feature representation as the optimization variable, a supervised learning model is constructed with the abstract feature representation as input and ship performance indicators as output.

[0026] The supervised learning model is trained by minimizing the prediction error, forming a mathematical model that can be used for rapid performance evaluation.

[0027] Based on the above technical solutions, preferably, the step of using the mathematical model to replace numerical simulation tools in the ship morphology optimization process until an optimization result represented by abstract features is obtained, specifically includes:

[0028] The optimization process employs the mathematical model to replace numerical simulation tools, with the optimizer using abstract feature representations as input variables. Based on the mathematical model, the optimizer continuously calls the objective function value to complete multiple rounds of evaluation and gradient updates of the objective performance until the optimal abstract feature solution that meets the preset conditions is obtained, thus yielding the optimization result.

[0029] Based on the above technical solutions, preferably, the step of generating the corresponding optimized ship type through parameterization according to the ship type optimization parameters specifically includes:

[0030] Obtain a complete set of ship type optimization parameters;

[0031] A parametric modeling framework is introduced, and the set of ship type optimization parameters is used as the input of the geometric generation function. Interpolation technology is used to construct a continuous and differentiable hull shape surface in the longitudinal section, transverse section and waterline direction.

[0032] During the modeling process, a two-way traceability relationship between parameters and geometric structures is established simultaneously.

[0033] By adopting a local modular combination strategy, the hull is divided into bow, main body and stern, which are modeled separately and then spliced ​​together under unified control to generate a three-dimensional optimized hull model.

[0034] A second aspect of this application provides a model building apparatus based on the abstract feature representation of ship hull optimization parameters. The apparatus includes an acquisition module, a processing module, and an output module, wherein:

[0035] The acquisition module is used to encode and extract features from the first dimension of ship type optimization parameters to obtain an abstract feature representation of the second dimension, wherein the second dimension is lower than the first dimension;

[0036] The processing module is used to represent the abstract features as optimization variables for the ship type optimization problem, and to construct a mathematical model of ship-related performance based on the optimization variables.

[0037] The processing module is used to replace numerical simulation tools with the mathematical model in the ship type optimization process until the optimization result is obtained in terms of abstract features.

[0038] The processing module is used to reverse map and decode the optimization results into actual ship type optimization parameters;

[0039] The output module is used to generate the corresponding optimized ship type through parameterization based on the ship type optimization parameters.

[0040] Based on the above technical solution, preferably, the processing module is used to set the structure of the encoding network and to encode the data samples in the first-dimensional parameter space through nonlinear mapping and nonlinear activation functions. The structural expression of the encoding network is as follows:

[0041] H = g(WX + b) m )

[0042] Wherein, H represents the abstract feature. k is the second dimension, n is the number of samples, g(·) is the non-linear activation function, W is the weight matrix, and b m Let X be the bias vector, and let X be the original input corresponding to the data sample in the first dimension parameter space, X∈R^(d×n), k is less than d, and d is the first dimension.

[0043] Based on the above technical solution, preferably, the processing module is used to introduce a loss function to measure the error between the abstract feature representation and the original input. The loss function includes a squared error loss function and a cross-entropy loss function, wherein the expression for the squared error loss function is:

[0044]

[0045] The expression for the cross-entropy loss function is:

[0046]

[0047] in, x is the error value. i For the original input, This is a representation of the abstract feature;

[0048] The processing module is used to optimize the weight matrix and the bias vector by minimizing the loss function during training, so that the resulting abstract feature representation is close to the essential features of the ship hull optimization parameters in the first dimension in terms of information expression.

[0049] Based on the above technical solutions, preferably, the processing module is used to complete the reverse mapping from the optimization result to the ship type optimization parameters through linear mapping to obtain the actual ship type optimization parameters;

[0050] The processing module is used to optimize the inverse mapping decoding process by training the output weight matrix contained in the linear mapping, with the goal of minimizing the difference between the optimization result and the ship type optimization parameters.

[0051] Based on the above technical solutions, preferably, the processing module is used to construct a supervised learning model with the abstract feature representation as the optimization variable and the ship performance index as the output;

[0052] The processing module is used to train the supervised learning model by minimizing the prediction error, thereby forming a mathematical model that can be used for rapid performance evaluation.

[0053] Based on the above technical solutions, preferably, the processing module is used to replace numerical simulation tools with the mathematical model in the optimization process. The optimizer uses abstract feature representation as input variables, continuously calls the objective function value based on the mathematical model, completes multiple rounds of evaluation and gradient update of the target performance, until the optimal abstract feature solution that meets the preset conditions is obtained, and the optimization result is obtained.

[0054] Based on the above technical solutions, preferably, the acquisition module is used to obtain a complete set of ship type optimization parameters;

[0055] The processing module is used to introduce a parametric modeling framework, take the set of ship type optimization parameters as input to the geometric generation function, and use interpolation technology to construct a continuous and differentiable hull shape surface in the longitudinal section, transverse section and waterline direction.

[0056] The processing module is used to simultaneously establish a bidirectional traceability relationship between parameters and geometric structures during the modeling process;

[0057] The processing module is used to divide the hull into bow, body and stern by adopting a local modular combination strategy, model them separately and then splice them together under unified control to generate a three-dimensional optimized hull model.

[0058] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0059] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0060] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0061] 1. This application introduces a nonlinear coding mechanism into the high-dimensional ship hull optimization parameter space to extract low-dimensional abstract feature representations, and uses these abstract features as optimization variables to construct a supervised learning mathematical model. This model replaces numerical simulation tools in the performance prediction and optimization process, significantly reducing the dimensionality of input variables and the complexity of modeling. Compared with traditional methods that directly model in the original high-dimensional space, this approach maintains the integrity of the information structure while reducing the number of modeling samples required, accelerating model training convergence, and improving prediction computation efficiency, thereby significantly improving the modeling efficiency in the high-dimensional parameter space.

[0062] 2. This application introduces a loss function to quantitatively evaluate the information reconstruction error between the abstract feature representation and the original input, and optimizes the weight matrix and bias vector in the encoding network with the goal of minimizing this error. This ensures that the extracted abstract feature representation retains the essential structural information of the original ship hull optimization parameters to the maximum extent while achieving dimensionality compression. This mechanism effectively improves the representational power and reversibility of abstract features, providing a high-fidelity feature foundation for subsequent performance prediction model construction and accurate decoding of optimization solutions, thereby enhancing the stability, generalization, and reconstruction accuracy of the entire modeling system.

[0063] 3. This application constructs a reverse decoding mechanism based on linear mapping to accurately reconstruct the optimization results from low-dimensional abstract feature representations into high-dimensional ship hull optimization parameters. By training the output weight matrix, it minimizes the difference between the decoded results and the actual ship hull optimization parameters, achieving optimal approximation in the decoding process. This ensures the reconstructibility and design feasibility of the optimization results, effectively improving the accuracy and stability of the mapping from feature space to engineering parameter space. It establishes a reliable bridge from optimization calculation to the generation of actual ship hull parameters, thereby guaranteeing the direct application capability of the model optimization results in actual ship hull design. Attached Figure Description

[0064] Figure 1This is a flowchart illustrating a model construction method based on the abstract feature representation of ship morphology optimization parameters disclosed in an embodiment of this application;

[0065] Figure 2 This is a schematic diagram of the process of extracting and reconstructing the abstract feature representation of ship type optimization parameters disclosed in an embodiment of this application.

[0066] Figure 3 This is a schematic diagram of a model building device based on the abstract feature representation of ship morphology optimization parameters disclosed in an embodiment of this application;

[0067] Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application.

[0068] Explanation of reference numerals in the attached drawings: 301, acquisition module; 302, processing module; 303, output module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. Detailed Implementation

[0069] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0070] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0071] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0072] This embodiment discloses a model construction method based on the abstract feature representation of ship morphology optimization parameters, referring to... Figure 1 This includes the following steps S110-S150:

[0073] S110 encodes and extracts features from the ship shape optimization parameters in the first dimension to obtain an abstract feature representation in the second dimension.

[0074] A model construction method based on the abstract feature representation of ship shape optimization parameters disclosed in an embodiment of the present application is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablets, wearable devices, and PCs (Personal Computers). It can also be a background server running a model construction method based on the abstract feature representation of ship shape optimization parameters. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0075] First, set the structure form of the encoding network, and clarify that the input is the original input sample matrix X ∈ R d×n , where R is the real number field, d is the dimension of the ship shape optimization parameters in the first dimension, and n is the number of samples. The set encoding network consists of a weight matrix W ∈ R k×d and a bias vector b m ∈ R k×1 to perform a linear transformation and a non-linear activation process on each input sample, and output an abstract feature representation H ∈ R k×n , where k < d is the dimension of the abstract feature.

[0076] This non-linear mapping process uses the following expression:

[0077] H = g(WX + b m )

[0078] where g(·) is a non-linear activation function. Common function forms include the Sigmoid function, the Tanh function, or the ReLU function, etc., which are used to introduce non-linearity to enhance the model's fitting ability for complex structures. In a specific implementation, first randomly initialize the weight matrix W and the bias vector b m , then perform matrix multiplication WX on each input sample X one by one, add the bias term b m , and then apply the activation function g(·) to perform element-wise non-linear transformation on the result to generate the corresponding abstract feature representation H.

[0079] The goal of this encoding network is to compress high-dimensional inputs into low-dimensional representations while minimizing information loss, preserving the main trends and key feature structures among the input parameters as the basis for subsequent optimization modeling and predictive analysis. During training, joint optimization is performed using a decoding network and a reconstruction error loss function to ensure that the extracted abstract feature representations have good representativeness and reversibility. This effectively reduces the dimensionality complexity of the ship hull optimization parameter space while maintaining the original structural information, thus achieving the dual goals of parameter compression and improved optimization efficiency.

[0080] Furthermore, firstly, after encoding and extracting the first-dimensional ship type optimization parameters and obtaining the abstract feature representation H = g(WX + b) m Following this, a corresponding decoding network is constructed to reconstruct the original input. This involves mapping the abstract feature representation H back to the high-dimensional parameter space through a symmetric structure, generating approximate reconstructed values. Used to measure information loss.

[0081] To quantitatively evaluate the ability of abstract feature representations to preserve the original input, a loss function is introduced. This is used to calculate the reconstruction error for each group of samples. In actual training, one of the following two typical loss functions can be selected based on the data type and distribution characteristics: one is the squared error loss function, which is suitable for continuous variable inputs, and its expression is:

[0082]

[0083] Where, x i This represents the i-th original input sample. The first loss function measures the reconstruction accuracy in the Euclidean sense for the corresponding reconstructed samples. The second is the cross-entropy loss function, suitable for scenarios where the input data has a probability distribution or discrete binary features; its expression is:

[0084]

[0085] This loss function measures the deviation between two probability distributions from the perspective of information entropy. During training, minimizing this loss function is used as the optimization objective, and the weight matrix W and bias vector b in the encoding network are continuously updated through the backpropagation algorithm. m This optimizes the network's parameter structure. In each iteration, the loss function is recalculated based on the reconstructed value of the current network output, and the gradient is corrected and updated until the loss function converges to the minimum range, indicating that the abstract feature representation can fully express the essential feature information of the original input parameters.

[0086] Finally, the encoding network structure obtained by minimizing the loss function can stably output a low-dimensional abstract feature representation H, and ensure that the abstract feature has high fidelity to the first-dimensional ship shape optimization parameters during the mapping and restoration process, providing a usable data foundation and mathematical guarantee for subsequent optimization modeling and reverse decoding based on abstract features.

[0087] S120 uses abstract features as optimization variables in the ship type optimization problem, and constructs a mathematical model of ship-related performance based on these optimization variables.

[0088] In one possible implementation, abstract feature representations are used as optimization variables in the ship type optimization problem. Based on these optimization variables, a mathematical model of ship-related performance is constructed. Specifically, this includes: using abstract feature representations as optimization variables, constructing a supervised learning model with abstract feature representations as inputs and ship performance indicators as outputs; and training the supervised learning model by minimizing prediction errors to form a mathematical model that can be used for rapid performance evaluation.

[0089] Specifically, the first step is to use the abstract feature representation as a new optimization variable for performance modeling. This step encodes the abstract feature representation H∈R of the network output. k×n As input variables to the supervised learning model, the corresponding ship performance index Y∈R is also used. m×n As the target output of the supervised learning model, k represents the abstract feature dimension, m represents the performance index dimension, and n represents the number of samples. This input-output pairing method constructs a supervised learning dataset with abstract feature representations as input and ship performance indices as output, used to establish a mapping relationship between the abstract feature space and the performance response space.

[0090] Subsequently, a supervised learning model is constructed to learn this mapping relationship. The model structure can be a predictive model with non-linear expressive capabilities, such as a multilayer perceptron, support vector regression, radial basis function neural network, or extreme learning machine. Using the input feature H as the independent variable and the performance index Y as the dependent variable, the model is trained to approximate the function f:R. k →R m In model building, technical parameters such as network topology, activation function, initialization strategy, and regularization methods need to be set to enable the model to have good generalization ability and adapt to the nonlinear structural features implied by abstract features.

[0091] During the model training phase, model parameters are optimized by minimizing the error function between the predicted output and the true performance metric. The loss function is often chosen in the form of mean squared error, and its expression is as follows:

[0092]

[0093] Where θ is the parameter set of the supervised learning model, f(H) i;θ) represents the model's prediction result for the i-th input sample, Y i The corresponding target performance index is defined. Parameters are iteratively adjusted using optimization algorithms such as backpropagation or least squares to converge the loss function to its minimum, ultimately forming a mathematical model that can quickly respond to input abstract features and output high-precision performance prediction results.

[0094] Once trained, the supervised learning model is capable of predicting ship performance in a low-dimensional abstract feature space. It can be directly embedded into the ship optimization process as an objective function or constraint for quick invocation, significantly replacing the high-cost computation of traditional numerical simulation tools in performance evaluation, thereby achieving an order-of-magnitude improvement in optimization efficiency.

[0095] S130 uses mathematical models instead of numerical simulation tools in the ship type optimization process until the optimization results are obtained in terms of abstract features.

[0096] In one possible implementation, a mathematical model is used to replace numerical simulation tools in the ship hull optimization process until an optimization result represented by abstract features is obtained. Specifically, the mathematical model replaces numerical simulation tools in the optimization solution process. The optimizer uses the abstract feature representation as input variables, continuously calls the objective function value based on the mathematical model, completes multiple rounds of evaluation and gradient update of the target performance, until the optimal abstract feature solution that meets the preset conditions is obtained, and the optimization result is obtained.

[0097] Specifically, first, the optimizer structure is defined, and an optimization algorithm suitable for global search in continuous space is selected, such as particle swarm optimization, genetic algorithm, differential evolution algorithm, or gradient-enhanced optimization algorithm, and H∈R is represented by abstract features. k As input variables to the optimizer, k is the abstract feature dimension, which represents the structure of compressed global information output by the encoding network.

[0098] During the optimizer initialization phase, a set of initial sample points with abstract features are generated as the starting point for optimization. Subsequently, based on the supervised learning model that has been trained, each sample point H is... i Input to mathematical model f(H) i The corresponding ship performance indicators are predicted, and the objective function value is defined according to the optimization objective. If the objective is to minimize the total ship resistance or maximize propulsion efficiency, the objective function can be set as follows:

[0099]

[0100] Where f(H) is the performance index output by the mathematical model. This refers to the optimization objective mapping applied to this metric. In each optimization iteration, the optimizer performs operations such as sample selection, updating, crossover, or gradient correction based on the objective function value of the current abstract feature variable, generating a new generation of candidate solutions for the abstract feature.

[0101] This optimization process continuously calls the mathematical model to complete multiple rounds of objective function evaluation. Compared with the traditional method of using numerical simulation tools to evaluate each ship type design step by step, its model computational overhead is significantly reduced, allowing thousands or even tens of thousands of performance evaluations to be completed in a short time. It also supports multi-objective collaborative optimization or the solution of constrained optimization problems. After each iteration, the optimizer performs a convergence determination on the current optimal objective function value and the fluctuation range of variables. When the convergence conditions are met, such as the change in the objective function being less than a preset threshold, the maximum number of iterations being reached, or the fluctuation in the solution space tending to stabilize, the iteration process terminates.

[0102] The final output abstract feature representation is the optimal abstract feature solution, representing the best performance solution obtainable under the current mathematical model prediction results. This abstract feature solution will be passed as the optimization result to the subsequent decoding module to be restored to the actual ship morphology optimization parameters. The entire process completely replaces the simulation function with a mathematical model, constructing an efficient optimization framework with abstract features as input and a supervised learning model as the computational core. This achieves a solution path that balances accuracy and efficiency for complex multidimensional ship morphology optimization problems in a low-dimensional feature space.

[0103] S140, the optimization results are back-mapped and decoded into actual ship type optimization parameters.

[0104] In one possible implementation, the optimization result is back-mapped and decoded into actual ship type optimization parameters. Specifically, this includes: completing the back-mapping from the optimization result to the ship type optimization parameters through linear mapping to obtain the actual ship type optimization parameters; and optimizing the back-mapping decoding process by training the output weight matrix contained in the linear mapping, with the goal of minimizing the difference between the optimization result and the ship type optimization parameters.

[0105] Specifically, firstly, the decoding structure is defined as a set of linear mapping relationships, that is, the final optimization result is an abstract feature representation matrix H∈R. k×n The goal is to reconstruct the original ship form optimization parameter matrix. Where k is the dimension of the abstract features, d is the dimension of the first dimension of the ship type optimization parameters, and n is the number of samples. The linear mapping expression is defined as:

[0106]

[0107] Where, β∈R k×dThe output weight matrix controls the direction and magnitude of the mapping from abstract features to high-dimensional parameters. To ensure high reconstruction accuracy of this linear mapping, the output weight matrix β needs to be trained and optimized. The training objective is to optimize the reconstruction parameters. To approximate the original ship hull optimization parameters X as closely as possible, least squares error is introduced as the loss function:

[0108]

[0109] The optimal solution for the output weight matrix is ​​obtained by finding the minimum value of the loss function. When matrix H is invertible or can be generalizedly inverted, the weight matrix is ​​calculated using the Moore-Penrose generalized inverse method, and the solution is:

[0110]

[0111] in To abstract the feature representation matrix into its generalized inverse, singular value decomposition can be used for stable solution. To further enhance the generalization ability and decoding robustness of the inverse mapping, a loss function in the regularized form of the extreme learning machine autoencoder is introduced:

[0112]

[0113] Where C is the regularization coefficient, used to balance the optimization objective between the norm of the output weights and the reconstruction error. In actual solutions, depending on the relationship between the number of samples and the feature dimensions, the following closed-form solutions are adopted:

[0114]

[0115] Once this weight matrix is ​​obtained, it can be used to decode any optimized abstract feature representation. The corresponding actual ship type optimization parameters are calculated to ensure that the decoding results are engineering-feasible in terms of structural consistency and parameter accuracy. The final output actual ship type optimization parameters can be used as input to the ship type parametric modeling module to drive subsequent hull 3D modeling and performance verification, realizing an effective transition from feature space solution to physical design space solution. This implementation process is characterized by its simple structure, stable convergence, and high computational efficiency, and is a key technical path to achieve the reversibility and accuracy control of the decoding process.

[0116] S150 generates the corresponding optimized ship type through parameterization based on the ship type optimization parameters.

[0117] In one possible implementation, based on the ship hull optimization parameters, a corresponding optimized ship hull is generated parametrically. Specifically, this includes: obtaining a complete set of ship hull optimization parameters; introducing a parametric modeling framework, using the set of ship hull optimization parameters as input to a geometric generation function, and employing interpolation techniques to construct continuous and differentiable hull shape surfaces in the longitudinal section, transverse section, and waterline direction; simultaneously establishing a bidirectional traceability relationship between parameters and geometric structures during the modeling process; and adopting a local modular combination strategy to divide the hull into bow, hull, and stern, modeling them separately, and then splicing them together under unified control to generate a three-dimensional optimized ship hull model.

[0118] Specifically, firstly, after completing the abstract feature decoding, a complete set of ship hull optimization parameters is obtained. This set of parameters is a numerical vector of dimension d, containing core variables that control the shape of the hull, such as principal dimension parameters, coordinates of hull line control points, waterline curvature coefficient, volume coefficient, bilge line position, and bow and stern morphology parameters. This set of parameters serves as the input basis for subsequent ship hull geometric modeling, has clear structural semantics and geometric constraint meaning, and is the basic data carrier for realizing the transition from optimization space to three-dimensional ship hull space.

[0119] Subsequently, a parametric modeling framework is introduced to achieve the transformation from optimization parameters to geometric surfaces. Specifically, geometric generating functions are used to interpolate and model the longitudinal section, transverse section, and waterline direction. The optimization parameters are used as function inputs to control the node positions, curvature changes, and boundary shapes, generating the set of surfaces required in the 3D hull structure. Commonly used interpolation techniques include B-spline curves, NURBS surfaces, or the Bezier method. Combined with C1 or C2 continuity constraints, smooth transitions of the surfaces are achieved, ensuring that the hull surface is continuous and differentiable, and meets the requirements of streamlining and manufacturability, thereby establishing a parameter-driven spatial configuration modeling mechanism.

[0120] During the modeling process, a bidirectional traceability relationship between parameters and geometric structures is established simultaneously. By recording the geometric influence area and control variable path corresponding to each parameter dimension, a parameter-to-geometry mapping table and a geometry-to-parameter reverse lookup channel are constructed to support subsequent sensitivity analysis, variable adjustment, and rapid correction operations. This traceability mechanism enables transparent structural control, facilitating the precise identification of affected structural areas in performance feedback, optimization iteration, or local modification scenarios, thereby improving the control efficiency and interpretability of ship modeling results.

[0121] Finally, to enhance modeling flexibility and structural control accuracy, a local modular combination strategy is introduced. The hull is divided into structural units such as the bow, hull body, and stern, and local parameter control models are constructed for each unit. Each module independently performs parameter analysis, geometric construction, and surface splicing. Continuity constraints are applied at the boundaries to ensure smooth integration of the overall structure. By uniformly controlling the connection points and transition curvature between modules, the modeling freedom and local optimization capabilities can be improved while ensuring modeling integrity. The final 3D optimized hull model assembled from all modules can be output in standard CAD format for numerical verification, rapid prototyping, or engineering deployment, achieving efficient conversion from parametric results to a physical hull model.

[0122] Reference Figure 2 The diagram illustrates a process for extracting and reconstructing abstract features of ship morphology optimization parameters according to an embodiment of this application. The process includes three functional areas: the original input data input area, the feature extraction layer-by-layer area, the feature reconstruction layer-by-layer area, and the output reconstruction area. The leftmost part represents the original input data X∈R. d×n , representing the ship hull optimization parameter matrix of dimension d, containing n samples. This input is passed through the input layer to the multi-layer feature extraction structure on the right, forming the encoding network.

[0123] The feature extraction layer consists of multiple nonlinear hidden layers. Starting from the first layer, the number of nodes in each layer gradually decreases, eventually converging to the minimum abstract feature representation of dimension p. This compression process is driven by a series of weight matrices and nonlinear activation functions, extracting feature representations layer by layer from the original high-dimensional space to the low-dimensional abstract space. The intermediate layers, labeled K1, K2, and K3 from left to right, represent the multi-level compression process of abstract features, and the output is the core abstract feature representation H∈R. p×n , where p <d。

[0124] The feature reconstruction region, layer by layer, is a decoding network with a structure symmetrical to the encoding network. The number of nodes gradually increases, simulating the inverse mapping process from a low-dimensional abstract feature space to a high-dimensional parameter space. This reconstruction process relies on a set of independent weight matrices and activation functions, aiming to recover approximate reconstructed values ​​from the abstract feature representation that are as close as possible to the original parameters. Finally, the output reconstruction region on the right side outputs reconstructed data of dimension d. Compare with the original input X.

[0125] The entire network aims to minimize the input X and output during the training phase. The objective is to use the loss function between the two, often employing the squared error loss function or cross-entropy loss function. The backpropagation algorithm optimizes all encoding and decoding weight matrices, ensuring that the abstract feature representation possesses both information compression capability and good information fidelity. This structure is used in ship hull optimization to encode dimensionality optimization parameters and support subsequent back-mapping decoding. It is a core neural network framework for realizing an integrated process of feature extraction, approximate modeling, optimization decoding, and 3D modeling.

[0126] This embodiment also discloses a model construction device based on the abstract feature representation of ship morphology optimization parameters, referring to... Figure 3 It includes an acquisition module 301, a processing module 302, and an output module 303, wherein:

[0127] The acquisition module 301 is used to encode and extract features from the ship type optimization parameters of the first dimension to obtain an abstract feature representation of the second dimension, wherein the second dimension is lower than the first dimension.

[0128] The processing module 302 is used to represent abstract features as optimization variables in the ship type optimization problem, and to construct a mathematical model of ship-related performance based on the optimization variables.

[0129] The processing module 302 is used to replace numerical simulation tools with mathematical models in the ship type optimization process until the optimization results are obtained in terms of abstract features.

[0130] The processing module 302 is used to reverse map and decode the optimization results into actual ship type optimization parameters.

[0131] Output module 303 is used to generate the corresponding optimized ship type through parameterization based on the ship type optimization parameters.

[0132] In one possible implementation, the processing module 302 is used to define the structure of the encoding network and to encode the data samples in the first-dimensional parameter space through nonlinear mapping and nonlinear activation functions. The structural expression of the encoding network is as follows:

[0133] H = g(WX + b) m )

[0134] Where H represents the abstract feature. k is the second dimension, n is the number of samples, g(·) is the non-linear activation function, W is the weight matrix, and b m Let X be the bias vector, and let X be the original input corresponding to the data sample in the first-dimensional parameter space, X∈R^(d×n), k is less than d, and d is the first dimension.

[0135] In one possible implementation, processing module 302 is used to introduce a loss function to measure the error between the abstract feature representation and the original input. The loss function includes a squared error loss function and a cross-entropy loss function, wherein the expression for the squared error loss function is:

[0136]

[0137] The expression for the cross-entropy loss function is:

[0138]

[0139] in, x is the error value. i For the original input, It represents abstract features.

[0140] The processing module 302 is used to optimize the weight matrix and bias vector by minimizing the loss function during the training process, so that the obtained abstract feature representation is close to the essential features of the first dimension of ship morphology optimization parameters in terms of information expression.

[0141] In one possible implementation, the processing module 302 is used to complete the reverse mapping from the optimization result to the ship type optimization parameters through linear mapping to obtain the actual ship type optimization parameters.

[0142] Processing module 302 is used to optimize the reverse mapping decoding process by training the output weight matrix contained in the linear mapping, with the goal of minimizing the difference between the optimization result and the ship type optimization parameters.

[0143] In one possible implementation, the processing module 302 is used to construct a supervised learning model with abstract feature representations as inputs and ship performance indicators as outputs, using abstract feature representations as optimization variables.

[0144] The processing module 302 is used to train the supervised learning model by minimizing the prediction error, forming a mathematical model that can be used for rapid performance evaluation.

[0145] In one possible implementation, the processing module 302 is used to employ a mathematical model instead of numerical simulation tools in the optimization process. The optimizer uses abstract feature representation as input variables, continuously calls the objective function value based on the mathematical model, completes multiple rounds of evaluation and gradient update of the target performance, until the optimal abstract feature solution that meets the preset conditions is obtained, and the optimization result is obtained.

[0146] In one possible implementation, the acquisition module 301 is used to obtain a complete set of ship type optimization parameters.

[0147] The processing module 302 is used to introduce a parametric modeling framework, take the set of ship shape optimization parameters as input to the geometric generation function, and use interpolation technology to construct a continuous and differentiable hull shape surface in the longitudinal section, transverse section and waterline direction.

[0148] The processing module 302 is used to simultaneously establish a two-way traceability relationship between parameters and geometric structures during the modeling process.

[0149] The processing module 302 is used to divide the hull into bow, body and stern by adopting a local modular combination strategy, model them separately and then splice them together under unified control to generate a three-dimensional optimized hull model.

[0150] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0151] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, user interface 403, network interface 404, and at least one memory 405.

[0152] The communication bus 402 is used to enable communication between these components.

[0153] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0154] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0155] The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 405, and by calling data stored in memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 401.

[0156] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. As a computer storage medium, the memory 405 may include an operating system, a network communication module, a user interface 403 module, and an application program based on a model construction method for abstract feature representation of ship hull optimization parameters.

[0157] exist Figure 4In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 that is a model construction method based on the abstract feature representation of ship shape optimization parameters. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

[0158] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0159] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0160] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0163] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory 405 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0164] This application also discloses a computer-readable storage medium storing instructions. When executed by one or more processors 401, these instructions cause an electronic device to perform one or more of the methods described in the above embodiments.

[0165] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A model construction method based on ship type optimization parameter abstract feature representation, characterized in that, The method comprises: encoding and extracting features of the ship type optimization parameters in the first dimension to obtain abstract feature representations in the second dimension, wherein the second dimension is lower than the first dimension, and specifically comprising: setting the structure of the encoding network, completing the encoding of the data samples in the parameter space in the first dimension through nonlinear mapping and nonlinear activation function, and the structure expression of the encoding network is as follows: ； wherein H is the abstract feature representation, R is a real number field, k is the second dimension, n is the number of samples, g(·) is a nonlinear activation function, W is a weight matrix, b m is a bias vector, X is the original input corresponding to the data sample in the first dimension parameter space, k is less than d, and d is the first dimension. introducing a loss function to measure the error between the abstract feature representations and the original input, and the loss function comprises a square error loss function and a cross-entropy loss function, wherein the expression of the square error loss function is: ； the expression of the cross-entropy loss function is: ； wherein, is an error value, is the original input, is the abstract feature representation; In the training process, the weight matrix and the bias vector are optimized by minimizing the loss function, so that the abstract feature representations are close to the essential features of the ship type optimization parameters in the first dimension in information expression; the abstract feature representations are used as optimization variables of the ship type optimization problem, and a mathematical model of ship related performance is constructed based on the optimization variables; the mathematical model is used to replace the numerical simulation tool to participate in the ship type optimization process until the optimization result in the abstract feature representation is obtained; the optimization result is reversely mapped and decoded into the actual ship type optimization parameter, specifically comprising: the linear mapping is used to complete the reverse mapping from the optimization result to the ship type optimization parameter to obtain the actual ship type optimization parameter; the output weight matrix contained in the linear mapping is trained to minimize the difference between the optimization result and the ship type optimization parameter, and the reverse mapping decoding process is optimized; according to the ship type optimization parameter, a corresponding optimized ship type is generated by parameterization.

2. The model construction method based on ship type optimization parameter abstract feature representation according to claim 1, characterized in that, the abstract feature representations are used as optimization variables of the ship type optimization problem, and a mathematical model of ship related performance is constructed based on the optimization variables, specifically comprising: the abstract feature representations are used as the optimization variables to construct a supervised learning model with the abstract feature representations as the input and the ship performance indicators as the output; the supervised learning model is trained by minimizing the prediction error to form a mathematical model for rapid performance evaluation.

3. The model construction method based on ship type optimization parameter abstract feature representation according to claim 1, characterized in that, the mathematical model is used to replace the numerical simulation tool to participate in the ship type optimization process until the optimization result in the abstract feature representation is obtained, specifically comprising: the mathematical model is used to replace the numerical simulation tool to participate in the optimization solving process, and the optimizer uses the abstract feature representation as the input variable, continuously calls the objective function value based on the mathematical model, completes multiple rounds of evaluation and gradient update of the target performance, and obtains the optimal abstract feature solution that meets the preset condition to obtain the optimization result.

4. The model construction method based on ship type optimization parameter abstract feature representation according to claim 1, characterized in that, the abstract feature representations are used as optimization variables of the ship type optimization problem, and a mathematical model of ship related performance is constructed based on the optimization variables, specifically comprising: a complete set of ship type optimization parameters is obtained; a parameterization modeling framework is introduced, the ship type optimization parameter set is used as the input of the geometric generation function, and the interpolation technology is used to construct the continuous and derivable ship hull surface in the longitudinal section, transverse section and waterline direction; during the modeling process, the bidirectional tracing relationship between the parameters and the geometric structure is established simultaneously; The hull is divided into bow, main body and stern by local modular combination strategy, and the three parts are modeled and spliced to generate a three-dimensional optimized ship model.

5. A model construction apparatus based on a ship type optimization parameter abstract feature representation, characterized by, The device is used for performing the model construction method based on the abstract feature representation of ship optimization parameters, and the device comprises an acquisition module (301), a processing module (302) and an output module (303). The acquisition module (301) is used for encoding and extracting features of the ship optimization parameters in the first dimension to obtain the abstract feature representation in the second dimension, wherein the second dimension is lower than the first dimension. The processing module (302) is used for taking the abstract feature representation as an optimization variable of the ship optimization problem, and constructing a mathematical model of the related performance of the ship based on the optimization variable. The processing module (302) is used for replacing the numerical simulation tool with the mathematical model to participate in the ship optimization process until the optimization result in the abstract feature representation is obtained. The processing module (302) is used for reversely mapping and decoding the optimization result into the actual ship optimization parameters. The output module (303) is used for generating the corresponding optimized ship type through parameterization according to the ship optimization parameters.

6. An electronic device, comprising: The electronic device comprises a processor (401), a communication bus (402), a user interface (403), a network interface (404) and a memory (405), the memory (405) is used for storing instructions, the user interface (403) and the network interface (404) are used for communicating with other devices, the communication bus (402) is used for connecting and communicating between components in the electronic device, and the processor (401) is used for executing the instructions stored in the memory (405) to enable the electronic device to perform the method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1-4 is performed.

Images