Electric bicycle shape design method, system, electronic device and storage medium

Abstract

This application belongs to the field of vehicle industrial design and discloses a method, system, electronic device, and storage medium for electric bicycle form design. The method includes: mining user sentiment vocabulary through online comments, constructing a sentiment lexicon by combining an improved word frequency-inverse document frequency algorithm with D-S evidence theory, and selecting key emotional words; constructing a convolutional long short-term memory neural network model optimized by the snake swarm algorithm to realize a mapping model between customer sentiment and product form characteristics; conducting a comprehensive subjective and objective evaluation of the design scheme by combining eye-tracking experiments and subjective evaluations, thereby selecting the optimal design scheme; selecting the optimal scheme and inputting it into a generative AI platform for multi-angle visual rendering, supplemented by ergonomic modeling and aerodynamic simulation, to verify its structural feasibility and performance. This method provides systematic technical support for enhancing the emotional value and optimizing the design of industrial products.

Description

Technical Field

[0001] This invention relates to the field of sustainable vehicle industrial design technology driven by big data, and in particular to an electric bicycle form design method, system, electronic device and storage medium that integrates sensory engineering, deep learning and multi-source evaluation. Background Technology

[0002] Driven by the trends of green transportation and sustainable urban development, micro-mobility products have become practical tools for solving short-distance commutes. Among them, electric-assist bicycles have become an important choice for short-distance travel due to their environmental friendliness, efficiency, and health benefits. Compared with traditional bicycles, electric-assist bicycles, aided by batteries, not only improve the comfort and efficiency of the riding experience but also alleviate urban traffic congestion and carbon emission pressures.

[0003] With continuous optimization of production technology, the market for e-bikes is becoming increasingly homogenized. Mainstream brands are finding it difficult to differentiate their products in terms of range, speed, and weight, and consumers are shifting their focus towards the emotional satisfaction and personalized expression offered by the products. This trend necessitates that product design not only meet functional requirements but also accurately respond to users' emotional needs to influence their purchasing decisions. However, during the product design process, companies often struggle to listen to customers and accurately capture their true needs, leading to new product sales failures. Therefore, conducting scientific surveys of user needs can improve R&D efficiency, shorten product development cycles, and increase the success rate of product launches.

[0004] Existing technologies for capturing users' emotional needs largely rely on traditional questionnaires and interviews, which suffer from high subjectivity, limited sample size, and low efficiency. In the design generation stage, they depend on the designer's experience and lack data-driven intelligent generation and optimization mechanisms. In the solution evaluation stage, subjective evaluation and objective engineering verification are often disconnected. Therefore, there is an urgent need for a method for designing the form of electric-assist bicycles that can systematically integrate users' emotional needs, intelligently generate design solutions, and conduct multi-dimensional scientific verification. Summary of the Invention

[0005] Based on this, the present invention aims to provide a systematic, sensory-driven electric bicycle form design method, system, electronic device, and storage medium, which aims to solve the problems in the existing electric-assist bicycle form design process, such as inaccurate capture of user sensory needs, reliance on subjective experience in design scheme generation, and lack of comprehensiveness and scientific rigor in scheme evaluation.

[0006] In a first aspect, the present invention provides a method for designing the form of an electric bicycle, comprising the following steps:

[0007] Based on online user review data, a method combining word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory fusion was used to extract and determine key emotional words to represent users' emotional needs.

[0008] A convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is constructed and trained. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word.

[0009] Combining subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, the optimal morphological design combination is comprehensively evaluated and ranked using a priority ordering method to select the optimal design scheme.

[0010] The optimal design scheme was rendered from multiple angles, and ergonomic and computational aerodynamic simulations were performed to verify its structural feasibility and performance.

[0011] As an optional implementation of the first aspect of this application, the step of extracting and determining key emotional words for representing users' emotional needs includes: collecting and preprocessing user comment texts to form a valid comment text set; using a term frequency-inverse document frequency algorithm to calculate the terms in the text set to obtain initial weights; dividing the comment text into three semantic structure regions—title, first sentence, and body—and assigning each region a preset weighting coefficient to it, correcting the initial weights to obtain corrected weights; using the normalized initial weights and the corrected weights as two independent sources of evidence, fusing them using the Dempster synthesis rule of DS evidence theory to generate a final comprehensive score, and sorting them from high to low according to the comprehensive score to select the key emotional words.

[0012] As an optional implementation of the first aspect of this application, in the step of using the convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm to establish a nonlinear mapping relationship between the key perceptual vocabulary and the electric bicycle morphological features composed of multiple morphological components: the electric bicycle morphological features are composed of multiple morphological components including at least handlebars, seat, frame, mudguards, pedals, wheels and sprocket assembly, and the different forms of each morphological component are encoded as preset integers.

[0013] As an optional implementation of the first aspect of this application, the step of constructing and training a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm includes: performing convolution and pooling operations on the sequence composed of the morphological component encodings using a one-dimensional convolutional neural network layer to extract local spatial features; inputting the local spatial features into a two-layer long short-term memory network to capture the sequence dependencies between the morphological component combinations; the snake swarm algorithm uses the mean squared error of the convolutional neural network-long short-term memory network model on the validation set as the fitness function, and optimizes the number of hidden layer units of the long short-term memory network by iteratively updating the positions of the snake head, snake body and snake tail.

[0014] As an optional implementation of the first aspect of this application, in the step of comprehensively evaluating and ranking the optimal morphological design combination by combining subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, in order to select the optimal design scheme: the objective physiological data consists of at least seven eye-tracking indicators obtained through eye-tracking experiments, including total fixation duration, average fixation duration, number of fixations, first fixation duration, total access duration, average access duration, and number of accesses; the comprehensive evaluation assigns a weight of 50% to the weighted average of the subjective questionnaire evaluation data and the objective physiological data, and calculates the final score using the priority order method.

[0015] As an optional implementation of the first aspect of this application, the steps of performing multi-angle visual rendering of the optimal design scheme and performing ergonomic simulation and computational aerodynamic simulation on it include: the multi-angle visual rendering includes: using a sketch-aware rendering platform to convert the line drawing of the optimal design scheme into a basic rendering image, then inputting the basic rendering image into a 3D generation platform to obtain multi-angle images, and finally inputting the multi-angle images into a stable diffusion model for refined rendering; the ergonomic simulation is performed using CATIA software, and the RULA score is used to evaluate and optimize the comfort of the riding posture; the aerodynamic simulation is performed using Fluent software, and the aerodynamic performance is evaluated by calculating the aerodynamic drag coefficient and analyzing the velocity field, pressure field, and trajectory diagram.

[0016] As an optional implementation of the first aspect of this application, the optimization of the ergonomic simulation includes: adjusting the geometric parameters of the optimal design scheme, including the top tube length, handlebar height, seat height, and seat tube angle, until the RULA score is reduced to a preset reasonable range.

[0017] Secondly, embodiments of this application provide an electric bicycle form design system, including:

[0018] The emotional vocabulary extraction module is configured to extract and determine key emotional words that represent users' emotional needs based on online user comment data. It uses a method that combines word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory fusion.

[0019] The morphology mapping generation module is configured to construct and train a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word.

[0020] The design scheme evaluation and screening module is configured to combine subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, and use a priority order method to comprehensively evaluate and rank the optimal morphological design combination in order to screen out the optimal design scheme.

[0021] The scheme rendering and verification module is configured to perform multi-angle visual rendering of the optimal design scheme and perform ergonomic simulation and computational aerodynamic simulation to verify its structural feasibility and performance.

[0022] Thirdly, embodiments of this application provide an electronic device, which includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of the method described in the first aspect.

[0023] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. More precise demand capture: By integrating the improved TF-IDF algorithm and DS evidence theory, core emotional words are objectively and efficiently extracted from massive user comments, avoiding the subjective bias and limitations of traditional research methods, and making design input closer to the real market demand.

[0026] 2. Intelligent Design Generation: Employing a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm, it can automatically learn and establish complex nonlinear relationships between emotional imagery and specific product form characteristics, realizing intelligent generation from emotional needs to design solutions, greatly improving design efficiency and innovation.

[0027] 3. Scientific evaluation system: Combining subjective questionnaires and objective eye-tracking experiments, a comprehensive evaluation system integrating subjective and objective methods was constructed. The priority ranking method was used for quantitative sorting, making the scheme selection process more comprehensive and reliable, and effectively making up for the shortcomings of a single evaluation method.

[0028] 4. Integrated verification process: By combining generative AI visual rendering with professional CATIA ergonomics and Fluent aerodynamics simulation, an integrated verification closed loop from aesthetic expression to engineering feasibility is achieved, ensuring that the final solution has both emotional value, user comfort and excellent physical performance. Attached Figure Description

[0029] Figure 1 This is a flowchart of an electric bicycle form design method according to an embodiment of the present invention;

[0030] Figure 2 This is a flowchart of the convolutional neural network-long short-term memory network model (SO-CNN-LSTM) optimized by the snake swarm algorithm proposed in this invention;

[0031] Figure 3 This is a schematic diagram of the structure of an electric bicycle form design system provided in an embodiment of the present invention.

[0032] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0033] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0034] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0035] Example 1

[0036] Please see Figure 1 This is a flowchart illustrating a method for designing the form of an electric bicycle according to an embodiment of the present invention. The method may include the following steps:

[0037] S1: Based on online user review data, a method combining word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory is used to extract and determine key emotional words to represent users' emotional needs.

[0038] For example, step S1 specifically includes the following steps S11 to S14:

[0039] S11. Constructing an emotional lexicon

[0040] We crawled user reviews of e-bikes from major global e-commerce platforms between 2015 and 2025, collecting a total of 16,024 original texts. We then used the Jieba word segmentation tool for Chinese word segmentation and part-of-speech tagging, and further performed preprocessing steps such as stop word filtering and low-frequency word removal to ultimately form a valid set of review texts.

[0041] S12, Calculating term frequency-inverse document frequency modeling

[0042] An improved term frequency-inverse document frequency (IF-IVF) model is used to weight the original corpus and construct a basic candidate set of intuitive words. Specifically, the IF-IVF weights consist of the following two parts:

[0043] TF is used to measure the frequency of a particular term in a specific text.

[0044]

[0045] in, Indicating word frequency Number of times it appears in the document This represents the total number of occurrences of all words in the document. This refers to the entire corpus, that is, the collection of all collected comment texts; This refers to a specific comment in the corpus, and .

[0046] IDF is used to measure the discriminative power of a term across the entire corpus.

[0047]

[0048] in, The total number of comments for the text collection. Indicates word frequency The number of documents is calculated by adding 1 to avoid a denominator of 0.

[0049] Finally, the term frequency-inverse document frequency score of the term. It can be represented as:

[0050]

[0051] In the actual calculation, the collected user comments of electric bicycles were first segmented into Chinese and labeled with parts of speech. After filtering out stop words, the word frequency-inverse document frequency score of each term was calculated. The terms were then sorted from high to low to select the first few representative preliminary candidate words, laying the foundation for subsequent structural position information correction and weight fusion.

[0052] S13. Structural Correction and Position-Aware Modeling Based on Word Position, Part-of-Speech, and Category Factors (Word Position, Part-of-Speech, and Category Factors)

[0053] This paper introduces word position, part-of-speech, and category factors to structurally modify the traditional word frequency-inverse document frequency score, thereby improving the contextual sensitivity of intuitive word extraction. Specifically, the word position, part-of-speech, and category factor method divides the comment text into three semantic structural regions: title, first sentence, and body. Different weighting coefficients are assigned to each region. , , And construct the following location-aware weighted model:

[0054]

[0055] in Words expressing feelings In the traditional comment title area Score, Words expressing feelings In the first sentence of the comment Score, Words expressing feelings In the main body of the comment Score.

[0056] in, In this embodiment, based on experience and verification, the setup is... , , This highlights the importance of the terms at the beginning of the structure.

[0057] S14, DS Evidence Theory Integration and Confidence Weighting

[0058] Let the candidate set of emotional words be... , nIndicating the number of candidate emotional words, each source of evidence... For each term Assign a basic probability assignment function It satisfies the following constraints:

[0059]

[0060] Where A represents any subset. This represents the empty set.

[0061] In this embodiment, the original word frequency-inverse document frequency score and the score corrected by word position, part-of-speech, and category factors are normalized and used as two independent sources of evidence. and Specifically, the normalized term weights are directly mapped to their confidence values ​​in each piece of evidence:

[0062]

[0063] in Words expressing feelings Tradition in the original text Score, This indicates the weighted average after incorporating a correction mechanism based on word position, part of speech, and category factors. Score.

[0064] Subsequently, according to Dempster's combination rule, the two basic probability assignments are fused. The combination formula is as follows:

[0065]

[0066] in, The degree of conflict between two pieces of evidence is defined as follows:

[0067]

[0068] Ultimately, based on the fusion Sort from highest to lowest, retaining the most confident emotional words as the core input for modeling and image generation in this embodiment.

[0069] S2: Construct and train a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word.

[0070] For example, step S2 specifically includes steps S21 to S22, which can be referred to Figure 2 This is a flowchart of the convolutional neural network-long short-term memory network model (SO-CNN-LSTM) optimized by the snake swarm algorithm proposed in this invention.

[0071] S21. The physical structure of 100 commercially available e-bikes (covering urban commuting and mountain bike types) was morphologically deconstructed, resulting in seven typical design components: handlebars, seat, frame, mudguards, pedals, wheels, and chainring. Each component was coded with integers from 1 to 7 based on its morphological differences, forming a unified e-bike morphological feature dataset. A standardized morphological analysis table was then constructed based on this. Simultaneously, to obtain the sensory label data required for model training, a questionnaire based on a 7-point Likert scale was designed, inviting 100 participants to independently rate each e-bike on four dimensions: "lightweight," "flexible," "refined," and "comfortable." Each user evaluated the appearance of each of the 100 e-bikes without brand identification. The average score of the sensory words was used as the final score for each sample bike, resulting in a sensory evaluation matrix.

[0072] S22, Constructing a mapping model

[0073] Step S22 can be further divided into the following steps S221 to S223:

[0074] S221, Snake Swarm Algorithm

[0075] In this embodiment, the snake swarm optimization algorithm is used to optimize key parameters in a long short-term memory neural network time series model:

[0076] (1) Individual initialization: Let the search space dimension be . Population size is Then the first The initial position vectors of each individual are:

[0077]

[0078] in and They represent the first The upper and lower boundaries of the dimension parameter.

[0079] (2) Fitness function design: The mean squared error (MSE) of the convolutional neural network-long short-term memory network on the validation set is used as the fitness evaluation function:

[0080]

[0081] in Rate based on genuine feelings. These are the model's predicted values.

[0082] (3) Snake head update (global guidance): The snake head performs global guidance based on the current optimal position, and its position update formula is:

[0083]

[0084] in Represents random variables Follow the interval Uniform distribution on Represents the uniform distribution function. This represents the position vector of the individual snake head at the current iteration step t. Step size factor It is the currently optimal individual globally.

[0085] (4) Snake Body Update (Exploration Enhancement): The body parts simulate the collaborative update of a snake swarm, and the update method is as follows:

[0086]

[0087] in Control the intensity of local disturbances. Represents the standard normal distribution. This represents the position vector of the individual snake body at the current iteration step t.

[0088] (5) Snake Tail Update (Local Convergence): The snake tail simulates the local search process, following the snake head to enhance convergence.

[0089]

[0090] in As a local convergence regulator, This represents the position vector of the snake tail individual at the current step length t.

[0091] S222, Convolutional Neural Network

[0092] (1) Convolution operation: Let the input feature sequence be , Indicates the last moment l The input, with a length of One-dimensional convolution kernel Perform local feature extraction. Convolution output at time step It is given by the following formula:

[0093]

[0094] in, This represents the convolution operation. For bias terms, It is a non-linear activation function. Indicates The data is centered on the convolution window.

[0095] (2) Pooling Layer: To reduce feature dimensionality and retain key information, the convolution output will be processed by max pooling. Let the pooling window size be... The result is expressed as follows:

[0096]

[0097] in, With a window width of half, pooling operations can effectively compress feature lengths, improve computational efficiency, and reduce overfitting.

[0098] Step S223, Long Short-Term Memory Network

[0099] Pooled feature sequences As a compressed representation of time series data, it is input into the Long Short-Term Memory (LSTM) network to capture dependencies over longer time periods. The computational steps of the LTM network are as follows:

[0100] (1) Calculation of candidate memory units:

[0101]

[0102] in, For candidate memory content, Enter the current time step. This is the hidden state from the previous moment. , This is the corresponding weight matrix. This is a bias term.

[0103] (2) Input gate and forget gate control:

[0104]

[0105]

[0106] in, and These represent the activation and output of the input gate and the forget gate, respectively, controlling the degree of current information being written and the degree of information being retained from the previous time step. For the sigmoid function, Indicates input The weight matrix between the input gate and the input gate Indicates input The weight matrix between the forget gate and the forget gate Indicates the previous hidden state The weight matrix between the input gate and the input gate Indicates the previous hidden state The weight matrix between the forget gate and the forget gate This represents the bias term of the input gate. This represents the bias term for the forget gate.

[0107] (3) Memory unit state update:

[0108]

[0109] This formula indicates that the current memory state is obtained by a weighted combination of the previous memory state and the current input information. Indicates the memory state from the previous moment. Passing through the Gate of Oblivion Weighted retention, compared with the currently entered information via input gate Weighted writing is combined into the common components.

[0110] (4) Output gate and hidden state update:

[0111]

[0112]

[0113] The final hidden state Based on the current state of the memory cell through After activation and control by the output gate, the output is used as the output of the Long Short-Term Memory network at the current time step. This represents the activation vector of the input gate. This represents the hidden state vector at the current time step. Indicates the current input The weight matrix between the input gate and the input gate Indicates the previous hidden state The weight matrix between the output gate and the output gate This represents the bias term of the output gate.

[0114] S3: Combining subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, the optimal morphological design combination is comprehensively evaluated and ranked using a priority ordering method to select the optimal design scheme.

[0115] For example, step S3 specifically includes the following steps S31 to S33:

[0116] S31, Subjective Evaluation

[0117] In terms of subjective evaluation, this embodiment invited 67 users to participate in the assessment, 23 of whom had design backgrounds. A 7-point Likert scale was used, with users independently scoring the top five selected e-bike designs based on four subjective terms: "lightweight," "flexible," "refined," and "comfortable." To avoid visual interference, e-bike designs without rendering were used for evaluation. The specific formula is as follows:

[0118]

[0119] but For the first Users rated the design. The total number of users who participated in the subjective evaluation. This represents the average subjective evaluation score of the design scheme.

[0120] S32, Objective Evaluation

[0121] A visual preference experiment was conducted using a Tobii Pro Glasses 3 eye tracker and a 27-inch computer monitor, with five graduate students from design backgrounds invited. Each observer spent approximately 6 minutes browsing the screen, and eye-tracking diagrams were prepared according to the screen size. Seven key indicators were recorded during the observation process: (1) total fixation time; (2) average fixation time; (3) number of fixations; (4) first fixation time; (5) total visit time; (6) average visit time; and (7) number of visits. After the experiment, the hotspot areas and observation trajectories of 20 electric-assisted bicycles were exported using Tobii Pro Lab software. Precise selection was performed based on the shape of each electric-assisted bicycle to export accurate attention data. All indicators were normalized using a minimum-maximum normalization process.

[0122]

[0123] but These are the raw values ​​of the eye-tracking metrics. This is the minimum value of this index among all design options. This represents the maximum value of this indicator across all design options. This is the normalized index value.

[0124] S33. Conduct a comprehensive evaluation using the priority ranking method.

[0125] In terms of weighting, subjective scores and objective eye-tracking metrics each account for 50%, ensuring a balanced representation of users' subjective perceptions and actual visual behavior.

[0126] The seven normalized eye-tracking metrics are integrated into a single objective composite score. Weighted average is used:

[0127]

[0128] in For the first Normalized values ​​of individual eye-tracking metrics For the first The weight of each indicator.

[0129] Calculate the overall evaluation score The optimal design scheme is selected based on a priority order method, and the specific calculation formula is as follows:

[0130]

[0131] S4: Perform multi-angle visual rendering of the optimal design scheme, and conduct ergonomic simulation and computational aerodynamic simulation to verify its structural feasibility and performance.

[0132] For example, step S4 specifically includes the following steps S41 to S43:

[0133] S41. Generate high-quality renderings

[0134] This invention uses the Vizcom platform for sketch-aware rendering, then inputs it as the initial image into a 3D generation platform to obtain multi-angle images. These multi-angle images are then input into a stable diffusion model for rendering optimization. The core idea is to progressively introduce Gaussian noise and then reconstruct the original image through a learning denoising network, thereby achieving conditionally guided high-fidelity image generation. The specific method is as follows:

[0135] (1) Forward process: In the diffusion stage, the original image Gaussian noise is gradually added to form a noisy image sequence. In its first The noise form of the step is:

[0136]

[0137] in, Indicates the cumulative retention factor. , For predefined noise scheduling coefficients, (˙) represents a Gaussian normal distribution, and I represents the unit covariance matrix, i.e., isotropic Gaussian noise with equal variance.

[0138] (2) Reverse process: By training a neural network model For any time Denoising is performed to reconstruct the original image. The inverse modeling probability distribution is:

[0139]

[0140] Among them, the mean With variance This determines the reconstruction performance of the model. The expression for calculating the mean is as follows:

[0141]

[0142] (3) Initial image prediction and reconstruction: based on the noisy image at the current time. With estimated noise An approximate estimate of the original image can be derived from this. Used for model reconstruction:

[0143]

[0144] (4) To control the stability of the denoising process, the variance term It can be set as a fixed constant or a learning parameter. A common practice is to set it as a scaled form of unit variance.

[0145]

[0146] in Usually based on It can be deduced that fine-tuning can also be done during training.

[0147] (5) Model optimization objective function

[0148] The stable diffusion model learns network parameters by minimizing the reconstruction error loss function, and its basic form is the mean square error of the predicted noise:

[0149]

[0150] in [˙] indicates the original data sample ,noise and time step The joint expectations.

[0151] S42, CATIA Human Body Simulation Verification

[0152] Using an adult male anatomy model (175cm tall, 70kg) as a baseline, a 3D solid model of the S13 solution was constructed using CATIA V5 R20, and complete human-machine assembly was performed. Virtual assembly and posture simulation were conducted on the S13 solution. A simulated urban commuting cycling scenario (no additional load, Load=0) was implemented, and cycling comfort was evaluated using RULA scores and pressure distribution.

[0153] In the initial simulation, the initial design suffered from an excessively short top tube (430mm), low handlebars (780mm), insufficient saddle height (757mm), and an overly steep seat tube angle (70°), resulting in a RULA score of 5, indicating that prolonged riding could easily lead to fatigue. After optimization, the top tube was increased to 510mm, the handlebars were raised to 790mm, the saddle height was adjusted to 770mm, and the seat tube angle was increased to 72°. The RULA score significantly decreased to 3, demonstrating that this configuration provides an ergonomically sound riding posture and good muscle load distribution.

[0154] S43, Aerodynamic Simulation Verification

[0155] To evaluate the aerodynamic performance of electric-assist bicycles during riding, the aerodynamic drag coefficient ( ) was calculated. C d The classic formula for ) is shown below:

[0156]

[0157] in, Indicates aerodynamic force, air density, Set the airflow speed (wind speed or average riding speed). The frontal area of ​​the vehicle.

[0158] Example 2

[0159] Please see Figure 3 The diagram shown is a structural schematic of an electric bicycle form design system according to the second embodiment of this application. The system includes the following key modules:

[0160] The emotional vocabulary extraction module 100 is configured to extract and determine key emotional words that represent users' emotional needs based on online user comment data, using a method that combines word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory fusion.

[0161] The morphology mapping generation module 200 is configured to construct and train a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word.

[0162] The design scheme evaluation and screening module 300 is configured to combine subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, and use a priority order method to comprehensively evaluate and rank the optimal morphological design combination in order to screen out the optimal design scheme.

[0163] The scheme rendering and verification module 400 is configured to perform multi-angle visual rendering of the optimal design scheme and perform ergonomic simulation and computational aerodynamic simulation on it to verify its structural feasibility and performance.

[0164] An electric bicycle form design system in this application embodiment can be a device, or a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. For example, mobile electronic devices can be mobile phones, tablets, laptops, PDAs, in-vehicle electronic devices, wearable devices, ultra-mobile personal computers (UMPCs), netbooks, or personal digital assistants (PDAs), etc., while non-mobile electronic devices can be servers, network-attached storage (NAS), personal computers (PCs), etc. This application embodiment does not impose specific limitations.

[0165] The electric bicycle form design system in this application embodiment can be a device with an operating system. This operating system can be Android, iOS, or other possible operating systems; this application embodiment does not specifically limit it.

[0166] The electric bicycle form design system provided in this application embodiment can achieve... Figure 1 The various processes of implementing an electric bicycle form design method in the method embodiment are not described in detail here to avoid repetition.

[0167] Optionally, embodiments of this application also provide an electronic device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the various processes of the above-described embodiment of an electric bicycle form design method and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0168] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described embodiment of an electric bicycle form design method and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0169] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0170] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0171] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this 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 disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0172] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for designing the form of an electric bicycle, characterized in that, Includes the following steps: Based on online user review data, a method combining word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory fusion was used to extract and determine key emotional words to represent users' emotional needs. A convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is constructed and trained. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word. The steps of constructing and training a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm include: performing convolution and pooling operations on the sequence of morphological component encodings using a one-dimensional convolutional neural network layer to extract local spatial features; inputting the local spatial features into a two-layer long short-term memory network to capture the sequence dependencies between the morphological component combinations; and using the mean squared error of the convolutional neural network-long short-term memory network model on the validation set as the fitness function, the snake swarm algorithm optimizes the number of hidden layer units in the long short-term memory network by iteratively updating the positions of the snake head, body, and tail. Combining subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, the optimal morphological design combination is comprehensively evaluated and ranked using a priority ordering method to select the optimal design scheme. The optimal design scheme was rendered from multiple angles, and ergonomic and computational aerodynamic simulations were performed to verify its structural feasibility and performance.

2. The method according to claim 1, characterized in that, The step of extracting and determining key emotional words to represent users' emotional needs includes: Collect and preprocess user comment texts to form a valid comment text set; The term frequency-inverse document frequency algorithm is used to calculate the initial weights of the terms in the text set; The comment text is divided into three semantic structure regions: title, first sentence, and body. Each region is assigned a preset weighting coefficient, and the initial weights are corrected to obtain the corrected weights. The normalized initial weights and the modified weights are used as two independent sources of evidence. They are then fused using the Dempster synthesis rule of the DS evidence theory to generate a final comprehensive score. The key emotional words are then selected by sorting them from high to low based on the comprehensive score.

3. The method according to claim 1 or 2, characterized in that, In the step of establishing the nonlinear mapping relationship between the key perceptual vocabulary and the morphological features of an electric bicycle composed of multiple morphological components using the convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm: The electric bicycle's morphological features consist of at least several morphological components, including handlebars, seat, frame, mudguards, pedals, wheels, and sprocket assembly. The different forms of each morphological component are encoded as preset integers.

4. The method according to claim 1, characterized in that, Combining subjective questionnaire evaluation data based on the Likert scale with objective physiological data based on eye-tracking experiments, a priority ranking method is used to comprehensively evaluate and rank the optimal morphological design combinations in order to select the optimal design scheme. The objective physiological data are at least seven eye-tracking indicators obtained through eye-tracking experiments, including total fixation duration, average fixation duration, number of fixations, first fixation duration, total access duration, average access duration, and number of accesses. The comprehensive evaluation assigns a weight of 50% to the weighted average of the subjective questionnaire evaluation data and the objective physiological data, and calculates the final score using a priority order method.

5. The method according to claim 1, characterized in that, The steps of performing multi-angle visual rendering of the optimal design scheme and conducting ergonomic simulation and computational aerodynamic simulation are as follows: The multi-angle visual rendering includes: using a sketch-aware rendering platform to convert the line drawing of the optimal design scheme into a basic rendering image, then inputting the basic rendering image into a 3D generation platform to obtain multi-angle images, and finally inputting the multi-angle images into a stable diffusion model for refined rendering. The ergonomic simulation was performed using CATIA software, and the RULA rating was used to evaluate and optimize the comfort of the riding posture. The aerodynamic simulation was performed using Fluent software, and aerodynamic performance was evaluated by calculating the aerodynamic drag coefficient and analyzing the velocity field, pressure field, and trajectory diagram.

6. The method according to claim 5, characterized in that, The optimization of the ergonomic simulation includes adjusting the geometric parameters of the optimal design, including the top tube length, handlebar height, seat height, and seat tube angle, until the RULA score is reduced to a preset reasonable range.

7. An electric bicycle form design system, characterized in that, include: The emotional vocabulary extraction module is configured to extract and determine key emotional words that represent users' emotional needs based on online user comment data. It uses a method that combines word frequency-inverse document frequency algorithm, word position, part of speech, category factors and weight correction and DS evidence theory fusion. The morphology mapping generation module is configured to construct and train a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to establish a nonlinear mapping relationship between the key sensory words and the morphological features of the electric bicycle composed of multiple morphological components. The convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm is used to predict and generate the corresponding optimal morphological design combination for each key sensory word. The steps of constructing and training a convolutional neural network-long short-term memory network model optimized by the snake swarm algorithm include: performing convolution and pooling operations on the sequence of morphological component encodings using a one-dimensional convolutional neural network layer to extract local spatial features; inputting the local spatial features into a two-layer long short-term memory network to capture the sequence dependencies between the morphological component combinations; and using the mean squared error of the convolutional neural network-long short-term memory network model on the validation set as the fitness function, the snake swarm algorithm optimizes the number of hidden layer units in the long short-term memory network by iteratively updating the positions of the snake head, body, and tail. The design scheme evaluation and screening module is configured to combine subjective questionnaire evaluation data based on Likert scales and objective physiological data based on eye-tracking experiments, and use a priority order method to comprehensively evaluate and rank the optimal morphological design combination in order to screen out the optimal design scheme. The scheme rendering and verification module is configured to perform multi-angle visual rendering of the optimal design scheme and perform ergonomic simulation and computational aerodynamic simulation to verify its structural feasibility and performance.

8. An electronic device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein when the program or instructions are executed by the processor, they implement the steps of an electric bicycle form design method as described in any one of claims 1-6.

9. A readable storage medium, characterized in that, The program or instructions are stored on the readable storage medium, and when the program or instructions are executed by the processor, they implement the steps of the electric bicycle form design method as described in any one of claims 1-6.

Images