Automobile modeling design method and system, electronic equipment and storage medium

Abstract

The invention belongs to the technical field of product design, and discloses an automobile modeling design method and system based on combination of a visual focus sequence and artificial intelligence, electronic equipment and a storage medium, and the method comprises the steps: capturing the visual focus sequence when a user browses an automobile modeling picture through employing an eye movement tracking technology, so as to extract the key morphological characteristics of an automobile; sorting and summarizing M types of perceptual vocabularies of automobiles by means of a large language model, and screening out N types of perceptual vocabularies with the highest weight from the M types of perceptual vocabularies through an entropy-BBWM-game theory hybrid analysis method; building a mapping relation between the automobile morphological characteristics and N types of perceptual vocabularies based on a KAT deep neural network prediction model, and selecting an optimal morphological characteristic combination according to the mapping relation; and based on the optimal morphological feature combination, generating N types of automobile models through an SDM model. According to the method, the problems of inaccurate user emotion quantization, low reliability of morphological feature combination prediction, insufficient multi-source data fusion efficiency and the like in automobile appearance design can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of product design, and particularly to an automotive styling design method, system, electronic device, and storage medium based on the combination of visual focus sequence and artificial intelligence. Background Art

[0002] Currently, the form design of new energy vehicles mainly relies on Kansei Engineering technology. By means of emotional mapping methods (such as the Kano model, entropy-TOPSIS, support vector regression, etc.), the emotional needs of users are transformed into design features. For example, predicting product trends by analyzing user online reviews, or constructing a form deconstruction table in combination with the visual focus sequence to identify key design elements. In recent years, artificial intelligence technologies (such as Transformer networks) have been introduced into the design field to process morphological data and sequence dependencies, and generate product form combination schemes. Existing research captures the user's visual focus sequence through eye-tracking technology, and combines machine learning models to optimize the design scheme, providing a data-driven technical path for automotive exterior design.

[0003] Traditional Kansei Engineering methods rely on weight calculations with strong subjectivity (such as AHP, KANO), resulting in low efficiency and poor generalization in emotional need analysis; the correlation modeling between the visual focus sequence and morphological features is still limited by the high complexity and low interpretability of the Transformer network, and it is prone to problems such as unstable prediction and overfitting. In addition, the deep integration of user emotional data, visual behavior data, and morphological features in the existing design process is insufficient, making it difficult to accurately capture dynamic aesthetic preferences, which restricts the personalization and sustainable optimization of design schemes. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide an automotive styling design method, system, electronic device, and storage medium based on visual focus sequence and improved artificial intelligence model. By comprehensively calculating emotional weights through entropy-BBWM-game theory, and combining the Kolmogorov-Arnold network (KAN) to enhance the robustness and interpretability of the Transformer, an efficient mapping from user emotional needs to morphological features is realized, and finally an automotive innovation design framework that takes into account aesthetic preferences and sustainability is formed. The aim is to solve problems such as inaccurate quantification of user emotions, low reliability of morphological feature combination prediction, and insufficient multi-source data fusion efficiency in automotive exterior design.

[0005] To solve the above technical problems, the present application is implemented as follows: In a first aspect, the embodiments of the present application provide a bionic design method for ceramic products, and the method includes: Use eye-tracking technology to capture the visual focus sequence of users when browsing car styling pictures, so as to extract the key morphological features of the car; Use a large language model to sort out and summarize M types of emotional vocabulary of the car, and through the entropy-BBWM-game theory hybrid analysis method, select the N types of emotional vocabulary with the highest weights from the M types of emotional vocabulary; Based on the deep neural network prediction model of KAT, construct the mapping relationship between the morphological features of the car and the N types of emotional vocabulary, and select the optimal morphological feature combination according to the mapping relationship; Based on the optimal morphological feature combination, generate N car styling through the SDM model.

[0006] As an optional implementation manner of the first aspect of the present application, the step of using eye-tracking technology to capture the visual focus sequence of users when browsing car styling pictures to extract the key morphological features of the car includes: tracking the eye movement trajectory of the user when observing the three views of the car based on the corneal reflection principle and recording the fixation point transfer path to determine the visual focus sequence; according to the visual focus sequence, perform priority sorting on the key morphological elements of the car to obtain the key morphological features of the car.

[0007] As an optional implementation manner of the first aspect of the present application, in the step of using a large language model to sort out and summarize M types of emotional vocabulary of the car and selecting the N types of emotional vocabulary with the highest weights from the M types of emotional vocabulary through the entropy-BBWM-game theory hybrid analysis method, the BBWM is a decision analysis method based on Bayesian statistics, and specifically includes: determining the evaluation index set of emotional vocabulary; determining the optimal index and the worst index from the evaluation index set based on historical experience data; constructing a first comparison matrix formed by the preference degree of the optimal index relative to all other indexes, and a second comparison matrix formed by the preference degree of all other indexes relative to the worst index; based on the Bayesian statistical method, model the first comparison matrix and the second comparison matrix through polynomial distribution and Dirichlet distribution, and use Markov chain Monte Carlo technology to calculate the posterior distribution to determine the weights of each evaluation index; combine the information entropy method to normalize the weights of each evaluation index and calculate the optimal combined weight.

[0008] As an alternative implementation of the first aspect of the present application, the steps of constructing a first comparison matrix formed by the preference degrees of the optimal index relative to all other indexes and a second comparison matrix formed by the preference degrees of all other indexes relative to the worst index include: using the numbers 1 to 9 to determine the preference degrees of the optimal index relative to all other indexes and to determine the preference degrees of all other indexes relative to the worst index, where 1 represents equally important and 9 represents extremely important, and the larger the number, the greater the degree of importance; through pairwise comparison, obtaining the first comparison matrix formed by the preference degrees of the optimal index relative to all other indexes and obtaining the second comparison matrix formed by the preference degrees of the worst index relative to all other indexes.

[0009] As an alternative implementation of the first aspect of the present application, the steps of constructing the mapping relationship between the automotive form features and the N types of emotional words based on the deep neural network prediction model of KAT and selecting the optimal form feature combination according to the mapping relationship include: constructing the automotive form features and the N types of emotional words into a data set; constructing a deep neural network prediction model of KAT based on the Kolmogorov-Arnold representation theorem as the theoretical basis; inputting the data set into the deep neural network prediction model of KAT for training to establish the mapping relationship between the automotive form features and the N types of emotional words; inputting the form feature data of the automotive to be measured into the trained deep neural network prediction model of KAT and outputting the predicted emotional value; obtaining the optimal form feature combination according to the predicted emotional value.

[0010] As an alternative implementation of the first aspect of the present application, the steps of constructing a deep neural network prediction model of KAT based on the Kolmogorov-Arnold representation theorem as the theoretical basis include: designing a network architecture with multiple KAN layers according to the Kolmogorov-Arnold representation theorem; stacking multiple KAN layers together in sequence and combining with the Transformer architecture to form a deep neural network prediction model of KAT; in the deep neural network prediction model of KAT, using the standard Transformer attention mechanism to capture global dependencies and replacing the multi-layer perceptron MLP in the traditional Transformer with a KAN layer.

[0011] As an alternative implementation of the first aspect of the present application, the steps of generating N car shapes through the SDM model based on the optimal morphological feature combination include: constructing a high-precision Class A surface model on the Rhino platform using NURBS surface modeling technology, focusing on optimizing the detailed feature structure of the car, and achieving parametric design through Grasshopper parameterization to obtain a modeling model; outputting a line drawing of the modeling model to be imported into the SDM model for generation and iteration, and finally generating N car shapes; through point cloud experimental evaluation, obtaining the emotional scores of the N car shapes, and further adjusting the optimized design of the car shapes according to the emotional scores.

[0012] In a second aspect, an embodiment of the present application provides a car shape design system, which includes: A morphological feature extraction module, configured to capture the visual focus sequence of a user when browsing car shape pictures by using eye tracking technology, so as to extract key car morphological features; A perceptual vocabulary screening module, configured to sort out and summarize M types of perceptual vocabulary of cars with the help of a large language model, and screen out the top N types of perceptual vocabulary with the highest weights from the M types of perceptual vocabulary through an entropy - BBWM - game theory hybrid analysis method; A feature combination selection module, configured to construct a mapping relationship between the car morphological features and the N types of perceptual vocabulary based on the deep neural network prediction model of KAT, and select the optimal morphological feature combination according to the mapping relationship; A car shape generation module, configured to generate N car shapes through the SDM model based on the optimal morphological feature combination.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0014] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0015] Compared with the prior art, the present invention proposes a car shape design method, which mainly has the following advantages: 1. When using eye tracking technology to capture the visual focus sequence, the eye movement trajectory of the user when observing the three - view drawings of the car is tracked based on the corneal reflection principle, and the fixation point transfer path is recorded to determine the visual focus sequence; according to the visual focus sequence, the key car morphological elements are sorted by priority to obtain the key car morphological features.

[0016] 2. During the process of screening emotional words, the BBWM method includes: determining the evaluation index set of emotional words; determining the optimal and worst indicators based on historical experience data; constructing a preference comparison matrix; using Bayesian statistical methods and Markov Chain Monte Carlo techniques to calculate the posterior distribution and determine the weights of each evaluation index; and combining the information entropy method to calculate the optimal combination weights.

[0017] 3. When constructing the KAT deep neural network model, the Kolmogorov-Arnold representation theorem is used as the theoretical basis to design a network architecture with multiple KAN layers, and a KAT model is formed by combining the Transformer architecture. The standard Transformer attention mechanism is used to capture global dependencies, and the multi-layer perceptron (MLP) in the traditional Transformer is replaced with a KAN layer.

[0018] 4. When generating car styling, the NURBS surface modeling technology is used to build a high-precision Class-A surface model on the Rhino platform, and parametric design is achieved through the Grasshopper parametric plug-in; the output line drawing is imported into the SDM model for generation and iteration, and finally N car styling are generated. The emotional scores are obtained through point cloud experiments, and the optimized design of the car styling is adjusted according to the scores. Brief Description of the Drawings

[0019] Figure 1 is a flowchart of a car styling design method provided by the first embodiment of the present invention; Figure 2 is a schematic structural diagram of a car styling design system provided by the third embodiment of the present invention. Detailed Description of the Embodiments

[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0021] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein. In addition, "and / or" in the specification and claims means at least one of the connected objects, and the character " / " generally means that the related objects before and after are in an "or" relationship.

[0022] In order to illustrate the technical solution described in this application, a specific embodiment is provided below for illustration.

[0023] Example 1 See also Figure 1 , which is a flow chart of a method for automobile styling design proposed in the first embodiment of the present application, and the steps of the proposed method are as follows.

[0024] S1. Use eye tracking technology to capture the user's visual focus sequence when browsing car styling pictures to extract the key morphological features of the car.

[0025] In some embodiments, the eye movement trajectory of the user when observing the three views of the car is tracked based on the principle of corneal reflection and the gaze point transfer path is recorded to determine the visual focus sequence; according to the visual focus sequence, the key morphological elements of the car are prioritized to obtain the key morphological features of the car.

[0026] It should be noted that the visual focus sequence refers to the order in which the subjects' eye gaze points stay when observing the three views of the car. It reveals the subjects' subconscious focus when facing complex views, and usually corresponds to the key attractive parts of the car's shape. The eye tracker tracks the user's eye movement trajectory based on the principle of corneal reflection and records the path of gaze point transfer to determine the visual focus sequence, thereby revealing the distribution law of consumers' psychological emphasis on car shape cognition. With the help of eye movement test data set analysis, the priority ranking of key morphological elements of car styling can be accurately extracted, providing data support for car designers to build a visual preference decision model, so that they can systematically grasp consumers' aesthetic orientation and achieve efficient matching of styling features and market demand through scientific design paths.

[0027] In this step, the eye tracker based on the principle of corneal reflection tracks the eye movement trajectory of the user when observing the three views of the car and records the path of the gaze point transfer, which can accurately determine the visual focus sequence, thereby revealing the subconscious focus of the subjects when facing complex views. These focus points usually correspond to the key attractive parts of the car's form. The extraction of the visual focus sequence provides a scientific basis for analyzing the distribution law of consumers' psychological emphasis on the cognition of car form. Through the analysis of the eye movement test data set, the priority ranking of the key morphological elements of the car's styling can be accurately extracted to help designers systematically grasp the aesthetic orientation of consumers.

[0028] S2. With the help of a large language model, M categories of sensory vocabulary for automobiles are sorted out and summarized. Through the entropy-BBWM-game theory hybrid analysis method, N categories of sensory vocabulary with the highest weights are selected from the M categories of sensory vocabulary.

[0029] For example, before carrying out the calculation work related to perceptual vocabulary, the first task is to widely collect and reasonably classify perceptual vocabulary. Based on this, this study uses ChatGPT, a powerful language model, to conduct a comprehensive search and sorting of the field to which the automobile belongs, and carefully classifies the massive perceptual vocabulary collected into M categories. In order to accurately screen out the N perceptual words with the highest weight, the study adopts a hybrid analysis method of BBWM (for integrating expert subjective preferences), entropy weight method (for objective quantitative data discreteness), and game theory (for dynamic balance of subjective and objective weights) to accurately screen out the most influential N perceptual words, providing a new methodology for sustainable design of automotive styling.

[0030] Among them, it should be emphasized that BBWM is a decision analysis method based on Bayesian statistics. The specific calculation steps include: determining a set of evaluation indicators for perceptual vocabulary; determining the best indicator and the worst indicator from the evaluation indicator set based on historical experience data; constructing a first comparison matrix formed by the preference degree of the best indicator relative to all other indicators, and a second comparison matrix formed by the preference degree of all other indicators relative to the worst indicator; based on the Bayesian statistical method, the first comparison matrix and the second comparison matrix are modeled through multinomial distribution and Dirichlet distribution, and the Markov chain Monte Carlo technology is used to calculate the posterior distribution and determine the weight of each evaluation indicator; combined with the information entropy method, the weight of each evaluation indicator is normalized to calculate the optimal combination weight.

[0031] Furthermore, the numbers 1 to 9 are used to determine the preference of the best indicator relative to all other indicators, and the preference of all other indicators relative to the worst indicator. 1 represents equal importance, 9 represents extreme importance, and the larger the number, the greater its importance. Through pairwise comparison, the first comparison matrix formed by the preference of the best indicator relative to all other indicators and the second comparison matrix formed by the preference of the worst indicator relative to all other indicators are obtained.

[0032] Exemplarily, the specific steps of the decision analysis method based on Bayesian statistics include: Step 1: Determine the set of evaluation indicators .

[0033] Step 2: In this step, multiple evaluators determine the optimal indicator based on experience from the set of indicators determined in the first step. and the worst indicator , the optimal indicator is the most important indicator or the most ideal indicator, while the worst indicator is the least important or the least ideal indicator among other criteria.

[0034] Step 3: Construct comparison criteria. Use the numbers 1 to 9 to determine the preference degree of the optimal indicator relative to all other indicators. 1 represents equally important, 9 represents extremely important, and the larger the number, the greater the importance. Through pairwise comparison, obtain the comparison matrix relative to the optimal indicator .

[0035] Step 4: Determine the preference degree of all other criteria relative to the worst criterion. Through pairwise comparison, obtain the comparison matrix relative to the worst indicator .

[0036] Step 5: On the basis of previously determining the optimal and worst indicators and constructing the comparison matrix, further use complex mathematical methods to calculate the weight of each evaluation indicator. After determining and as the model input and output respectively, a polynomial distribution needs to be added. The polynomial probability distribution of the worst indicator . Similarly, for the optimal indicator use polynomial distribution for modeling, but the probability distribution is opposite to that of the worst indicator . In summary, the weight determination method has been transformed into probability distribution estimation, and a Bayesian hierarchical model needs to be established for solution. Suppose there are N (n = 1, 2... N) evaluation experts, then the Nth evaluation expert determines the optimal-worst comparison matrix according to the evaluation indicators …, as and , and the set of optimal-worst comparison matrices of N evaluation experts is: and . Suppose represents the comprehensive weight determined by all evaluation experts, which is calculated from the set of weights of each index determined by each evaluation expert , and it is calculated through the joint probability distribution and . According to the above formula, calculate the probability of each random variable through the following probability rules to prove the clear conditional independence between variables.

[0037] First, considering the independence between all different variables, apply the Bayesian method to the joint probability formula. Among them, the distribution of each parameter is calculated based on three parameters: the probability chain rule, the conditional independence of different variables, and the preference of each decision maker for the standard. Use polynomial distribution to model and . Then, for a given , it can be expected that any is near it. Therefore, re-parameterize the mean and concentration parameters of the Dirichlet distribution. Then, given of the model is the mean of the distribution, and the degree of closeness between the two is controlled by a non - negative index. Therefore, in the above formula, represents the weight vector associated with each reviewer must be at nearby. In addition, the concentration parameter also needs to be modeled using a gamma distribution that satisfies the non - negativity constraint. Among them, a and b refer to the shape parameters of the gamma distribution. Finally, use the parameter of the non - informative Dirichlet distribution to calculate the prior distribution on. Considering that the model does not have a closed - form solution, the Markov Chain Monte Carlo (MCMC) technique is used to calculate the posterior distribution. For MCMC sampling, Just Another Gibbs Sampler (JAGS) in the probabilistic language is used to sample and calculate its posterior. The model finally outputs the posterior distribution of the optimal weight for each reviewer and the optimal aggregated weight for all reviewers

[0038] The tool of information entropy can be used to calculate the weights of various indicators, providing a basis for multi - indicator comprehensive evaluation. The specific steps are as follows

[0039] First, perform data normalization in the first step: normalize the data of positive indicators, negative indicators, and moderate - type indicators to avoid the impact of dimensions on data processing. In the second step, calculate the proportion of each indicator. In the third step, calculate the information entropy. In the fourth step, calculate the information entropy redundancy and then calculate the indicator weights. In the fifth step, construct a combined weight vector and then perform differential processing on the coefficients. Finally, obtain the weight coefficients

[0040] Among them, when game theory is used to determine the weight by integrating subjective and objective analysis, it mainly realizes the organic integration of subjective weight and objective weight and obtains the optimal combined weight through steps such as constructing a combined weight vector, introducing the idea of game theory, and normalizing the coefficients

[0041] In this step, first, by using large language models such as ChatGPT to comprehensively search and sort out the emotional vocabulary in the automotive field, 12 categories of emotional vocabulary can be efficiently and systematically summarized, providing a rich and structured data basis for subsequent analysis. Secondly, through a hybrid analysis method of entropy weight method, BBWM and game theory, the dynamic balance of subjective and objective weights is achieved. The entropy weight method objectively quantifies the data dispersion, BBWM integrates the subjective preferences of experts, and game theory further optimizes the weight allocation, thus accurately screening out the most influential 3 core emotional vocabulary. This process not only improves the scientificity and reliability of emotional vocabulary screening, but also provides a clear emotional orientation for automotive styling design, enabling designers to more accurately grasp user needs and enhance the market adaptability and user satisfaction of the design. In addition, BBWM, a decision-making analysis method based on Bayesian statistics, ensures the accuracy and stability of weight calculation through polynomial distribution, Dirichlet distribution and Markov chain Monte Carlo technology, providing a solid theoretical support for multi-index comprehensive evaluation.

[0042] S3. Construct a mapping relationship between the automotive form features and N categories of emotional vocabulary based on the KAT-based deep neural network prediction model, and select the optimal form feature combination according to the mapping relationship.

[0043] In some embodiments, the automotive form features and N categories of emotional vocabulary are constructed into a data set; based on the Kolmogorov-Arnold representation theorem as the theoretical basis, a KAT-based deep neural network prediction model is constructed; the data set is input into the KAT-based deep neural network prediction model for training to establish a mapping relationship between the automotive form features and N categories of emotional vocabulary; in the trained KAT-based deep neural network prediction model, the data of the automotive form features to be measured is input, and the predicted emotional value is output; the optimal form feature combination is obtained according to the predicted emotional value.

[0044] Furthermore, according to the Kolmogorov-Arnold representation theorem, a network architecture with multiple KAN layers is designed; multiple KAN layers are stacked together in sequence and combined with the Transformer architecture to form a KAT-based deep neural network prediction model; in the KAT-based deep neural network prediction model, the standard Transformer attention mechanism is used to capture global dependencies, and the multi-layer perceptron MLP in the traditional Transformer is replaced by the KAN layer.

[0045] As described in the example of step S2, after combining the entropy weight method - BBWM - game theory, the 3 emotional vocabulary with the highest weights are selected from the representatives of 12 categories of emotional vocabulary.

[0046] Afterwards, exemplarily, a morphological analysis table is made in combination with the results after the eye movement experiment, and a dataset containing 127 car morphologies and emotional values is constructed. Then, through the dataset, KAT is used to establish the mapping relationship between NEV-SUV and user emotions, so as to predict the best combination of morphological features corresponding to the highest emotional value.

[0047] In this step, large language models such as ChatGPT are used to comprehensively search and sort out the emotional vocabulary in the automotive field, and 12 categories of emotional vocabulary can be efficiently and systematically summarized, providing a rich and structured data basis for subsequent analysis. Secondly, through the hybrid analysis method of entropy weight method, BBWM and game theory, the dynamic balance of subjective and objective weights is achieved. The entropy weight method objectively quantifies the dispersion of data, BBWM integrates the subjective preferences of experts, and game theory further optimizes the weight allocation, so as to accurately screen out the three most influential core emotional vocabulary. This process not only improves the scientificity and reliability of the screening of emotional vocabulary, but also provides a clear emotional orientation for automotive styling design, enabling designers to better grasp user needs and improve the market adaptability and user satisfaction of the design. In addition, BBWM is a decision analysis method based on Bayesian statistics. Through polynomial distribution, Dirichlet distribution and Markov chain Monte Carlo technology, it ensures the accuracy and stability of weight calculation, providing a solid theoretical support for multi-index comprehensive evaluation.

[0048] S4. Based on the optimal combination of morphological features, N car shapes are generated through the SDM model.

[0049] In some embodiments, the NURBS surface modeling technology is used to construct a high-precision Class A surface model on the Rhino platform, focusing on optimizing the detailed feature structure of the car, and parametric design is realized through Grasshopper parameterization to obtain the modeling model; the line drawing of the modeling model is output and imported into the SDM model for generation and iteration, and finally N car shapes are generated; through the point cloud experiment evaluation, the emotional scores of the N car shapes are obtained, and the optimized design of the car shapes is further adjusted according to the emotional scores.

[0050] Specifically, N car shapes are generated using the SDM design and a high score is obtained in the actual point cloud experiment evaluation, so that the abstract emotional intention can be expressed as a concrete physical shape. After completing the KAT prediction of the best combination of morphological features corresponding to the highest emotional value, find the combination with the highest evaluation value for each emotional vocabulary. Then, actual creation is carried out with the help of SDM, and finally a score should be obtained in the point cloud experiment evaluation session to achieve the accurate transformation of the abstract emotional intention into a concrete physical shape. The specific implementation process of the point cloud test is as follows.

[0051] First, in the PyCharm development environment, import the 3D model (obj format) of the NEV-SUV with the help of the Open3D library. First, ensure that the Open3D library is installed. If not, it can be installed via the command pip install open3d. Then, import the Open3D library and perform visualization through the draw_geometries function.

[0052] Secondly, accurately calculate the point cloud density of the entire 3D model. A distance-based method can be adopted, and the average nearest neighbor distance of each point in the point cloud is calculated to estimate the density of the point cloud distribution. In specific implementation, use Open3D to obtain the nearest neighbor distance of each point, and then calculate its average value, which is the point cloud density.

[0053] Finally, in the 3D model, use a specific screening algorithm to select the cross-section, and discard other faces except the cross-section through data filtering technology, simplifying the complex three-dimensional model into a two-dimensional plane for easy analysis. To obtain the most representative data, further use the area maximization algorithm to screen out the cross-section with the largest area, thereby accurately calculating the final height. After batch processing the heights of the NEV-SUV models with good market sales using this process, verify whether the height of the resulting model in this study is within a reasonable range.

[0054] In this step, based on the optimal combination of morphological features, generate N car shapes through the SDM model, and use NURBS surface modeling technology and Grasshopper parametric design to construct a high-precision Class A surface model, which has significant technical effects: First, the SDM model combined with parametric design realizes the rapid generation and iteration of car shapes, significantly improving the design efficiency, while ensuring the diversity and innovation of the shapes. Second, evaluate the emotional scores of the N generated car shapes through point cloud experiment evaluation, transform the abstract user emotional intention into a concrete physical shape expression, and achieve the precise matching of emotion and design, enhancing the market adaptability and user satisfaction of the design. In addition, use the Open3D library to calculate the point cloud density and screen the cross-section of the 3D model, further optimizing the model details and ensuring the scientificity and rationality of the design results.

[0055] Example 2 Taking the automotive styling design of new energy SUV (NEV-SUV) as an example, the method of the present invention is further described in combination with the accompanying drawings and the design process of the product effect drawing, so as to verify the effectiveness of the method proposed by the present invention. The steps of the proposed method are as follows.

[0056] First, the research team delved into mainstream automotive information websites such as "Dongchedi" and "Autohome" to widely collect full-view information of new energy SUVs from the front, rear, and specific perspectives. Due to the wide range of data sources and different imaging conditions, professional software was used to process the collected pictures, removing background interference, unifying the picture sizes, and correcting the distortion caused by the shooting angles. The research team used the Remove website to batch process three views of 127 cars, ensuring that the pictures had the background removed and the sizes unified. Such meticulous preprocessing provided high-quality and consistent image data for the subsequent eye movement experiments. After this process, a dataset covering the full views of new energy SUVs from the front, rear, and specific perspectives was successfully constructed, laying a foundation for the subsequent precise ranking research on the automotive form based on the eye tracker.

[0057] Next, in order to evaluate the users' visual perception, an eye tracking experiment was designed. The experiment first selected 6 pictures from the 127 samples that significantly affected the users' emotions and unified the resolution and dimensions of these pictures. 10 participants with NEV-SUV driving experience and 5 industrial design researchers were recruited to ensure the avoidance of interference from text prompts. The TobiiPro Glass 3 eye tracker and a 21-inch monitor were used to collect eye movement data, and heat maps and eye movement trajectories were analyzed and generated through the Tobii Pro software. During the experiment, the participants fixated on the center of the screen, and one picture was shown every 15 seconds, with a total of 6 pictures shown twice each, and finally the overlaid heat maps and eye movement trajectories were exported. The experimental results showed that the participants mainly focused on the front hood and wheels of the NEV-SUV.

[0058] Before constructing the emotional mapping model of NEV-SUV, the research team systematically collected emotional words using ChatGPT. Through the design of multi-dimensional prompt words, a large number of positive emotional words describing the appearance and driving experience of NEV-SUV were collected. Then these words were classified into 12 categories such as "dynamic", "fashionable", "elegant", "luxurious", "exquisite", "unique", "classic", "cool", "simple", "low-key", "tech", and "powerful". This process ensured a comprehensive coverage of the emotional dimensions and provided a rich vocabulary basis for the subsequent emotional analysis.

[0059] Subsequently, the research team selected seven experts to subjectively assign weights to each index according to the Bayesian optimal weight method (BBWM). The experts respectively selected the optimal index and the worst index based on their judgments, and constructed the BO (Best-to-Others) matrix and OW (Others-to-Worst) matrix according to these preference relationships. This method effectively dealt with the preference inconsistency of the experts' opinions and ensured the scientificity and objectivity of the weight assignment results.

[0060] On this basis, the research team designed a questionnaire, combined with the collected NEV-SUV samples and perceptual vocabulary, and invited 120 subjects with a design background to score. A seven-level Likert scale was used for scoring. Finally, the collected mean values were combined with the morphological structure table to construct a perceptual evaluation matrix table, providing a basis for subsequent analysis of the relationship between morphological features and emotions.

[0061] To establish the mapping relationship between user emotions and the morphology of NEV-SUVs, the research team constructed a deep neural network prediction model based on KAT. First, the morphological features of NEV-SUVs were normalized to ensure the consistency of the dimension of feature data and avoid training bias. Then, a KAT model based on the improved Transformer architecture was constructed, using the KAN layer to replace the MLP layer and retaining the multi-head attention mechanism to adapt to complex emotion and morphology data analysis. During the training process, the dataset was divided into a training set (70%) and a test set (30%). The Adam optimizer and the mean squared error loss function were used, and the model was iteratively trained until the accuracy reached the standard, thus successfully establishing the mapping relationship between user emotions and the morphology of NEV-SUVs.

[0062] To further analyze the relationship between morphological features and user emotions, the research team used the exhaustive method to calculate all possible combinations of morphological features. Finally, 117,649 different combinations of morphological features were calculated, and the three emotional dimensions of "fashionable", "dynamic" and "luxurious" were predicted respectively. Through analysis, it was concluded that under specific combinations of morphological features, the emotional experiences of "fashionable", "dynamic" and "luxurious" of users can be maximally stimulated, providing data support for design decisions.

[0063] Finally, based on the KE theoretical framework, the research team carried out innovative styling design practice in the field of new energy SUVs. Through the entropy-BBWM-game theory method, three core perceptual vocabulary of "fashionable", "dynamic" and "luxurious" were selected as the design guidance, and the corresponding emotional evaluation system was constructed. In the digital modeling stage, the NURBS surface modeling technology was used to construct a high-precision Class A surface model on the Rhino 7.0 platform. The light guide structure of the through-type taillight was optimized, and the gradient algorithm of the grille unit was realized through the Grasshopper parametric plug-in. In the rendering link, the model line drawing was imported into the SDM model for generation and iterative optimization, and finally the innovative styling design of the new energy SUV was completed.

[0064] Embodiment 3 Please refer to Figure 2 , which shows a schematic structural diagram of an automobile styling design system proposed in the third embodiment of the present application. The system includes: The morphological feature extraction module 100 is used to capture the visual focus sequence of the user when browsing car styling pictures by using eye tracking technology, so as to extract the key morphological features of the car; The emotional vocabulary screening module 200 is used to sort and summarize M emotional vocabulary of cars with the help of a large language model, and screen out the top N emotional vocabulary with the highest weights from the M emotional vocabulary through an entropy - BBWM - game theory hybrid analysis method; The feature combination selection module 300 is used to construct a mapping relationship between the car morphological features and the N emotional vocabulary based on the deep neural network prediction model of KAT, and select the optimal morphological feature combination according to the mapping relationship; The car styling generation module 400 is used to generate N car stylings through the SDM model based on the optimal morphological feature combination.

[0065] A car styling design system in an embodiment of the present application can be a device, or a component, an integrated circuit, or a chip in a terminal. The device can be a mobile electronic device or a non - mobile electronic device. Exemplarily, the mobile electronic device can be a mobile phone, a tablet computer, a notebook computer, a handheld computer, an in - vehicle electronic device, a wearable device, an ultra - mobile personal computer (UMPC), a netbook, or a personal digital assistant (PDA), etc., and the non - mobile electronic device can be a server, a Network Attached Storage (NAS), a personal computer (PC), etc. The embodiments of the present application do not make specific limitations.

[0066] A car styling design system in an embodiment of the present application can be a device with an operating system. The operating system can be an Android operating system, an iOS operating system, or other possible operating systems. The embodiments of the present application do not make specific limitations.

[0067] A car styling design system provided by the embodiments of the present application can implement Figure 1 each process implemented by a car styling design method in the method embodiment. To avoid repetition, it will not be elaborated here.

[0068] Optionally, an embodiment of the present application further provides an electronic device, including a processor, a memory, a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, it implements each process of the above - mentioned car styling design method embodiment and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0069] An embodiment of the present application further provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, it implements each process of the above embodiment of a method for automotive styling design and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0070] Among them, the processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs, etc.

[0071] It should be noted that in this article, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article, or device. Without more limitations, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, article, or device including that element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the reverse order according to the functions involved. For example, the described method may be executed in a different order than described, and various steps may be added, omitted, or combined. Additionally, the features described with reference to certain examples may be combined in other examples.

[0072] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment method can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases, the former is a better implementation method. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc) and includes several instructions for causing a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of the present application.

[0073] The embodiments of the present application have been described above in conjunction with the accompanying drawings. However, the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative rather than restrictive. Under the inspiration of the present application, those of ordinary skill in the art can also make many forms without departing from the purpose of the present application and the scope protected by the claims, and all of them fall within the protection scope of the present application.

Claims

1. An automobile styling design method, characterized in that Including: Using eye-tracking technology to capture the visual focus sequence of users when browsing car styling pictures, so as to extract the key morphological features of the car; With the help of a large language model, organize and summarize M types of emotional vocabulary of the car, and through the entropy-BBWM-game theory hybrid analysis method, screen out the N types of emotional vocabulary with the highest weights from the M types of emotional vocabulary; Based on the deep neural network prediction model of KAT, construct the mapping relationship between the car morphological features and the N types of emotional vocabulary, and select the optimal morphological feature combination according to the mapping relationship; Based on the optimal morphological feature combination, generate N car styling through the SDM model.

2. The automotive styling design method according to claim 1, characterized in that The steps of using eye-tracking technology to capture the visual focus sequence of users when browsing car styling pictures to extract the key morphological features of the car include: Based on the corneal reflection principle, track the eye movement trajectory of the user when observing the three views of the car and record the fixation point transfer path to determine the visual focus sequence; According to the visual focus sequence, rank the priority of the key morphological elements of the car to obtain the key morphological features of the car.

3. A method for automotive styling design according to claim 1, characterized in that, In the step of using the large language model to organize and summarize M types of emotional vocabulary of the car, and screening out the N types of emotional vocabulary with the highest weights from the M types of emotional vocabulary through the entropy-BBWM-game theory hybrid analysis method, the BBWM is a decision analysis method based on Bayesian statistics, specifically including: Determine the evaluation index set of emotional vocabulary; Based on historical experience data, determine the optimal index and the worst index from the evaluation index set; Construct the first comparison matrix formed by the preference degree of the optimal index relative to all other indexes, and the second comparison matrix formed by the preference degree of all other indexes relative to the worst index; Based on the Bayesian statistical method, model the first comparison matrix and the second comparison matrix through multinomial distribution and Dirichlet distribution, and use the Markov chain Monte Carlo technique to calculate the posterior distribution to determine the weights of each evaluation index; Combined with the information entropy method, normalize the weights of each evaluation index and calculate the optimal combination weight.

4. A method for automotive styling design according to claim 3, characterized in that, The steps of constructing the first comparison matrix formed by the preference degree of the optimal index relative to all other indexes, and the second comparison matrix formed by the preference degree of all other indexes relative to the worst index include: Use the numbers 1-9 to determine the preference degree of the optimal index relative to all other indexes, and determine the preference degree of all other indexes relative to the worst index. 1 represents equally important, 9 represents extremely important, and the larger the number, the greater the importance; Through pairwise comparison, obtain the first comparison matrix formed by the preference degree of the optimal index relative to all other indexes, and obtain the second comparison matrix formed by the preference degree of the worst index relative to all other indexes.

5. A method for automotive styling design according to claim 1, characterized in that, The steps of constructing the mapping relationship between the car morphological features and the N types of emotional vocabulary based on the deep neural network prediction model of KAT, and selecting the optimal morphological feature combination according to the mapping relationship include: Construct the car morphological features and the N types of emotional vocabulary into a data set; Based on the Kolmogorov - Arnold representation theorem as the theoretical basis, construct a deep neural network prediction model of KAT; Input the dataset into the deep neural network prediction model of KAT for training to establish the mapping relationship between the automotive form features and the N types of perceptual vocabulary; In the trained deep neural network prediction model of KAT, input the data of the automotive form features to be measured and output the predicted perceptual values; Obtain the optimal form feature combination according to the predicted perceptual values.

6. A method for automotive styling design according to claim 5, characterized in that, The steps of constructing a deep neural network prediction model of KAT based on the Kolmogorov - Arnold representation theorem as the theoretical basis include: According to the Kolmogorov - Arnold representation theorem, design the network architecture of multiple KAN layers; Stack multiple KAN layers together in sequence and combine with the Transformer architecture to form a deep neural network prediction model of KAT; In the deep neural network prediction model of KAT, use the standard Transformer attention mechanism to capture global dependencies, and replace the multi - layer perceptron MLP in the traditional Transformer with a KAN layer.

7. A method for automotive styling design according to claim 1, characterized in that, The steps of generating N automotive shapes through the SDM model based on the optimal form feature combination include: Apply NURBS surface modeling technology on the Rhino platform to construct a high - precision Class - A surface model, focus on optimizing the automotive detail feature structure, and achieve parametric design through Grasshopper parameterization to obtain the modeling model; Output the line drawing of the modeling model to import it into the SDM model for generation and iteration, and finally generate N automotive shapes; Through point - cloud experiment evaluation, obtain the emotional scores of the N automotive shapes, and further adjust the optimized design of the automotive shapes according to the emotional scores.

8. An automotive styling design system, characterized in that, The system includes: A form feature extraction module, which is used to capture the visual focus sequence of users when browsing automotive shape pictures by using eye - tracking technology to extract the key form features of the vehicle; A perceptual vocabulary screening module, which is used to organize and summarize the M types of perceptual vocabulary of the vehicle with the help of a large - language model, and screen out the N types of perceptual vocabulary with the highest weights from the M types of perceptual vocabulary through the entropy - BBWM - game - theory hybrid analysis method; A feature combination selection module, which is used to construct the mapping relationship between the automotive form features and the N types of perceptual vocabulary based on the deep neural network prediction model of KAT, and select the optimal form feature combination according to the mapping relationship; An automotive shape generation module, which is used to generate N automotive shapes through the SDM model based on the optimal form feature combination.

9. An electronic device, characterized in that, It includes a processor, a memory, and a program or instruction stored on the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of an automotive shape design method as described in any one of claims 1 - 7 are implemented.

10. A readable storage medium, characterized in that, The program or instruction is stored on the readable storage medium. When the program or instruction is executed by the processor, the steps of an automotive shape design method as described in any one of claims 1 - 7 are implemented.

Images