Monorail crane shape design optimization method and system based on generative artificial intelligence

Abstract

The invention discloses a monorail crane shape design optimization method and system based on generative artificial intelligence, and relates to the field of intelligent human aided design. The problems that in an existing monorail crane appearance design, concept creativity and engineering feasibility are disjointed, the design period is long, and the iteration cost is high are solved. The method comprises the following steps: constructing a monorail crane multi-dimensional semantic association network; generating a conceptual design drawing based on a coal mine knowledge pre-trained vertical large language model and the semantic network; performing multi-dimensional comparison on the concept design drawing, the user demand text and the existing product image, wherein the multi-dimensional comparison comprises semantic consistency comparison and engineering feasibility comparison; based on a comparison result, iteratively optimizing parameters for driving image generation by adopting a particle swarm algorithm until the concept design drawing meets a comprehensive evaluation index; and finally, outputting an optimization scheme through the subjective and objective evaluation modules. The method is also suitable for the application fields of common environment mine monorail crane appearances, extreme environment mine monorail crane appearances and the like.

Description

Technical Field

[0001] This invention relates to the field of intelligent artificial intelligence-assisted design, specifically to a method and system for optimizing the shape design of a monorail crane based on generative artificial intelligence. Background Technology

[0002] The exterior design of heavy industrial equipment such as monorail cranes needs to meet stringent underground safety regulations, ergonomics, and functional reliability requirements while also considering certain aesthetic attributes. In traditional design processes, conceptual ideas and engineering feasibility verification are often disconnected, resulting in long design cycles and high iteration costs. How to quickly transform designers' abstract ideas into feasible engineering solutions that meet all constraints has long been a pain point in this field.

[0003] Ergonomics is an interdisciplinary field that studies the coordination of human-machine-environment systems. It aims to improve efficiency, safety, and comfort by optimizing tools, work methods, and environmental design. Through a three-tiered architecture of physiological parameter deconstruction, cognitive model construction, and system performance optimization, ergonomics is elevated from static adaptation to dynamic collaboration.

[0004] Generative artificial intelligence (AI) is an important branch of artificial intelligence, a technology that autonomously generates new content based on algorithms, models, and rules. It can learn the inherent patterns in large-scale data to generate new content that is logical, coherent, and not directly present in the training data.

[0005] In recent years, generative artificial intelligence technology has demonstrated powerful capabilities in image generation, but the directly generated results often lack engineering rationality. On the other hand, optimization algorithms such as particle swarm optimization are highly effective in multi-objective optimization problems. Chinese invention patent CN120296880A discloses the optimization of the emotional imagery of consumer products such as automobiles, but fails to solve the complex problem unique to the mining equipment field, which requires balancing creativity and rigid engineering constraints.

[0006] In traditional monorail crane design, the disconnect between conceptual ideas and engineering feasibility verification leads to repeated modifications and high iteration costs. While existing generative AI technologies can generate creative images, they lack engineering rationality and cannot meet the rigid constraints of mining equipment. Traditional design processes require designers to manually draw concept diagrams and engineers to verify feasibility step by step, resulting in high cross-stage communication costs and long design cycles. Existing AI generation technologies lack industry specificity, requiring extensive manual modifications to adapt the generated results to engineering needs, without substantially improving efficiency. Furthermore, general AI design technologies do not consider the specific constraints of mining equipment. Traditional designs require manual adaptation to different extreme environments, resulting in low adaptation efficiency and the potential to overlook key constraints. Traditional evaluations rely heavily on engineers' objective feasibility verification or users' subjective satisfaction ratings, leading to a single evaluation dimension. Existing AI design technologies lack a systematic evaluation mechanism, making it difficult to quantify the comprehensive adaptability of design solutions. Therefore, there is an urgent need for a new design method that combines the creativity of generative AI with engineering optimization capabilities. Summary of the Invention

[0007] This invention aims to overcome the problems of disconnect between conceptual design and engineering feasibility, long design cycles, and high iteration costs in the existing monorail crane design. To address these issues, this invention proposes a method and system for optimizing monorail crane shape design based on generative artificial intelligence. This invention achieves its solutions to the aforementioned technical problems through the following technical solutions: Option 1: This invention proposes a method for optimizing the shape design of a monorail crane based on generative artificial intelligence. The method includes the following steps: S1. Based on knowledge in the coal mining field, train a vertical large language model to obtain a database of parameters and keywords for different types of monorail crane shapes; S2. Use the vertical large language model described in S1 to obtain user demand text and scenario data, and construct a multi-dimensional semantic association network and comprehensive evaluation index system for monorail cranes. S3. Based on the multi-dimensional semantic association network and comprehensive evaluation index system of the monorail crane, the conceptual design drawing of the monorail crane shape is generated from the vertical large model; S4. The monorail crane outline concept design drawing described in S3 is compared with the user requirement text and the existing monorail crane product images in the database in multiple dimensions, and the quantitative comparison results are output. The multi-dimensional comparison includes semantic consistency comparison and engineering feasibility comparison. Based on the quantitative comparison results, the parameters of the driving image generation model are iteratively adjusted using an optimization algorithm to generate an optimized concept design drawing. S5. First, conduct an objective comprehensive evaluation of the optimized conceptual design drawing in S4. After the evaluation, the user conducts a subjective satisfaction evaluation of the optimized conceptual design drawing. If the comprehensive evaluation is passed, the final monorail crane shape design scheme is output. If the objective comprehensive evaluation does not meet the requirements, the association network and evaluation indicators are reconstructed and returned to S4. If the subjective comprehensive evaluation does not meet the requirements, subjective modification opinions are collected, the conceptual design drawing is modified, and returned to S4.

[0008] Furthermore, a preferred embodiment is provided, wherein S2 specifically includes: S2.1 Extract keywords from the user's request text, and retrieve the top N divergent information with the highest relevance in the semantic association network, starting from the keywords; S2.2 Based on the divergent information selected by the user, a design path is constructed in the semantic network, and the path information is transformed into a design feature vector that drives the image generation model. S2.3 Input the design feature vector into the image generation model to generate a conceptual design drawing of the monorail crane.

[0009] Furthermore, a preferred embodiment is provided, wherein the parameters of the image generation model described in S4 include the power source, structural form, sensor arrangement of the vehicle body, and cockpit space layout, ergonomic design, operating device and instrument panel design of the exterior of the driver's cab in the exterior image.

[0010] Furthermore, a preferred embodiment is provided in which the semantic consistency comparison described in S4 is implemented using the CLIP model. That is, the CLIP model is used to extract the text feature vector of the user requirement text and the image feature vector of the concept design diagram respectively; the cosine similarity between the two feature vectors is calculated as the score of semantic consistency comparison.

[0011] Furthermore, a preferred embodiment is provided, wherein the engineering feasibility comparison described in S4 is implemented using a pre-trained deep convolutional neural network, that is, the pre-trained deep convolutional neural network is used to extract the depth feature vectors of the conceptual design drawing and the existing monorail crane product image respectively; the cosine similarity between the two depth feature vectors is calculated as the score for engineering feasibility comparison.

[0012] Furthermore, a preferred implementation is provided, wherein the objective comprehensive evaluation described in S5 includes an engineering feasibility score, based on the similarity between graph-to-graph comparison and the product database; an ergonomics score, based on simulation data of the digital model in the cockpit; and an environmental matching score, based on the matching analysis of the design drawings and the mine scene using an AI model.

[0013] Furthermore, a preferred embodiment is provided, wherein the subjective satisfaction assessment in S5 includes the assessment of the perceived safety score and the perceived functionality score; the assessment of both the perceived safety score and the perceived functionality score is obtained by the user giving subjective scores.

[0014] Option 2: A monorail crane shape design optimization system based on generative artificial intelligence, the system being implemented based on the method described in Option 1, the system comprising: The information extraction module is used to train a vertical large language model based on knowledge in the coal mining field, and to obtain a database of various parameters and keywords of different types of monorail cranes. The information dissemination module is used to obtain user demand text and scenario data using the vertical large language model described in the information extraction module, and to construct a multi-dimensional semantic association network and a comprehensive evaluation index system for monorail cranes. The image generation module is used to generate a conceptual design drawing of the monorail crane from a large vertical model, based on the monorail crane's multi-dimensional semantic association network and comprehensive evaluation index system. The comparison module is used to perform multi-dimensional comparison between the monorail crane outline concept design drawing described in the design feature conversion module and the user requirement text, as well as the monorail crane product images already in the database, and output the quantitative comparison results. The multi-dimensional comparison includes semantic consistency comparison and engineering feasibility comparison. Based on the quantitative comparison results, the parameters of the driving image generation model are iteratively adjusted using an optimization algorithm to generate an optimized concept design drawing. The comprehensive evaluation module is used to first conduct an objective comprehensive evaluation of the optimized conceptual design drawings in the comparison module. After passing the evaluation, users conduct a subjective satisfaction evaluation of the optimized conceptual design drawings. If the comprehensive evaluation is passed, the final monorail crane shape design scheme is output. If the objective comprehensive evaluation does not meet the requirements, the association network and evaluation indicators are reconstructed and the project is returned to the comparison module. If the subjective comprehensive evaluation does not meet the requirements, subjective modification opinions are collected, the conceptual design drawings are modified, and the project is returned to the comparison module.

[0015] Option 3: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method described in Option 1.

[0016] Option 4: A computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the method described in Option 1.

[0017] The advantages of this invention are: The core of the monorail crane shape design optimization method and system based on generative artificial intelligence described in this invention lies in constructing a closed-loop optimization system based on generative artificial intelligence, which deeply integrates subjective creativity with objective constraints. This includes: acquiring user needs and scenario data; constructing a multi-dimensional semantic association network for monorail cranes; generating a conceptual design diagram based on a vertical large language model pre-trained with coal mine knowledge and the semantic network; comparing the conceptual design diagram with user requirement text and existing product images in multiple dimensions, including semantic consistency comparison and engineering feasibility comparison; based on the comparison results, using a particle swarm optimization algorithm to iteratively optimize the parameters driving image generation until the conceptual design diagram meets the comprehensive evaluation indicators; and finally outputting the optimized solution through subjective and objective evaluation modules. The comprehensive evaluation indicators include subjective evaluation modules and objective evaluation modules. This invention deeply integrates conceptual creativity with engineering feasibility through an optimization loop, providing designers with efficient and reliable design references.

[0018] This invention deeply binds subjective creativity with objective constraints through a closed-loop process of "generation-comparison-optimization-evaluation." The optimization phase employs a particle swarm optimization algorithm to iteratively adjust image generation parameters, ensuring that the design retains its core creativity while meeting constraints, significantly addressing the industry pain point of the disconnect between creativity and feasibility. Furthermore, this invention automates feasibility verification and parameter adjustment through multi-dimensional comparison and algorithm optimization, eliminating the need for manual verification and greatly reducing the number of iterations. The vertical large-scale language model described in this invention is trained based on knowledge from the coal mining field. The information database contains parameters, keywords, and extreme environment adaptation requirements for different types of monorail cranes, resulting in conceptual design drawings that are inherently compatible with underground safety regulations and environmental adaptation needs. A comprehensive evaluation system combining objective and subjective perspectives is constructed to ensure the comprehensiveness of the design scheme.

[0019] This invention is also applicable to the design of monorail cranes in both ordinary and extreme environments, including high-gas, high-temperature and high-humidity, and dusty mines. Extreme environment mines require different specialized sensors and equipment accessories for different environments. For example, in high-gas mines, a gas concentration sensor is installed on the crane body; in high-temperature and high-humidity mines, a temperature and humidity sensor is installed, and moisture-proof devices are installed on specific components. Attached Figure Description

[0020] Figure 1 This is a flowchart of the monorail crane shape design optimization method based on generative artificial intelligence as described in Implementation Method 1.

[0021] Figure 2 This is a flowchart of the vertical large language model for knowledge training in the coal mining field as described in Implementation Method 1.

[0022] Figure 3This is a flowchart illustrating the construction of a multi-dimensional semantic association network and comprehensive evaluation index system for monorail cranes as described in Implementation Method 1. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0024] Implementation Method 1, see [link] Figures 1 to 3 This embodiment describes a method and system for optimizing the shape design of a monorail crane based on generative artificial intelligence. The method specifically includes the following steps: S1. Training the Vertical Large-Scale Model and Establishing the Original Information Database: Based on knowledge in the coal mining field, a vertical large-scale language model is trained, and an information database of various parameters and keywords of different types of monorail cranes already available on the market is obtained. S2. Demand Acquisition and Knowledge Construction: Based on a pre-trained vertical large language model, the model is used to acquire user demand text and scenario data, and a multi-dimensional semantic association network and comprehensive evaluation index system for monorail cranes are constructed. S3. Generate preliminary conceptual design drawings: Based on the correlation network and evaluation indicators, generate 5 sets of monorail crane shape concept design drawings from the vertical large model. The design content must take into account structural strength, installation space adaptability and underground visual recognition. S4. Multi-dimensional comparative analysis and parameter optimization: The conceptual design drawing is compared with the user requirement text and the existing monorail crane product images in the database in multiple dimensions, and the quantitative comparison results are output. The multi-dimensional comparison includes semantic consistency comparison and engineering feasibility comparison. Based on the quantitative comparison results, the parameters driving the image generation model are iteratively adjusted using an optimization algorithm to generate a new conceptual design drawing. S5. Comprehensive Evaluation and Output: The optimized new concept drawing first undergoes an objective comprehensive evaluation. After passing the evaluation, users conduct a subjective satisfaction evaluation of the optimized concept design drawing. If the comprehensive evaluation is passed, the final monorail crane shape design scheme is output. If the objective comprehensive evaluation does not meet the requirements, the association network and evaluation indicators are reconstructed and the process returns to S4. If the subjective comprehensive evaluation does not meet the requirements, subjective modification opinions are collected, the concept design drawing is modified, and the process returns to S4.

[0025] The preliminary steps of the vertical large language model, namely the monorail appearance design generation model, include: training a vertical large model for mining equipment based on knowledge in the coal mining field and applying it to the process of acquiring monorail user needs and scenario data; acquiring monorail user needs and scenario data based on the large model; constructing a multi-dimensional semantic association network for monorails and a comprehensive evaluation module. S2 specifically includes the following steps: S2.1 Information Extraction and Divergence: Extract keywords from the user demand text, and retrieve the top N divergent information with the highest relevance in the semantic association network, starting from the keywords; S2.2 Design Feature Transformation: Based on the divergent information selected by the user, a design path is constructed in the semantic network, and the path information is transformed into a design feature vector that drives the image generation model; S2.3 Image Generation: Input the design feature vector into the image generation model to generate a conceptual design drawing of the monorail crane.

[0026] Specifically, the parameters of the image generation model include the power source and structural form of the exterior of the driver's body, the arrangement of sensors on the vehicle body, and the cockpit space layout, ergonomic design, main operating devices, and instrument panel design of the driver's cabin.

[0027] Specifically, the sensor components include the type, quantity, and installation location of the sensors.

[0028] Specifically, the expression of the semantic network is as follows: ;in, Represents a set of information. This represents the overall shape of a monorail crane. and They represent the first The first piece of information and the first One piece of information, Indicates the first The information and the first The weight of each related edge is set based on the frequency with which the two pieces of information appear in different comment texts. That is, the weight is the number of times the two pieces of information appear in the same comment, and the more times they appear, the greater the weight of the related edge.

[0029] Specifically, the auxiliary comparison part extracts features from user demand information and conceptual design diagrams using the CLIP model to obtain information feature vectors and image feature vectors of the same dimension. Then, a pre-trained deep convolutional neural network is used to input the conceptual design diagrams and existing monorail crane product images to output a feasibility score. The particle swarm optimization algorithm is then used to calculate the optimal solution from the input conceptual design diagrams and existing monorail crane product images to obtain the optimal result. The specific process of combining the CLIP model, the pre-trained deep convolutional neural network, and the particle swarm optimization algorithm is as follows: The CLIP model pre-training phase: Forward propagation: Input a batch of An image-text pair, The images were obtained by an image encoder. Image feature vectors , Each segment of text is processed by a text encoder to obtain N text feature vectors. All of them were L2 normalized.

[0030] Calculate the similarity matrix: Calculate the cosine similarity between all image vectors and all text vectors to form a similarity matrix. The formula for cosine similarity of the matrix is: Since the vector has Normalization simplifies the calculation to The diagonal elements of the matrix are all true matching pairs. The similarity.

[0031] The comparison loss function is calculated to find the corresponding text for each image and the matching image for each text. Each text is treated as a category, and a multi-class loss is calculated for each row of the matrix, ensuring that the similarity at the diagonal positions is significantly higher than the similarity at other rows. Each image is treated as a category, and a multi-class loss is calculated for each column of the matrix, ensuring that the similarity at the diagonal positions is significantly higher than the similarity at other columns. The loss function used is the symmetric cross-entropy loss, which is the average of the results calculated in both the row and column directions.

[0032] For images Its relationship with the text Its similarity score is higher than its scores with other text feature vectors. For text Its relationship with images Its similarity score is higher than its scores with other feature vectors. The process of pre-training a deep convolutional neural network: Preparing the dataset: For image classification tasks, the network needs to learn to map input images to the correct category labels.

[0033] Specifically, the aforementioned objective comprehensive assessment This includes assessments of project feasibility score, ergonomics score, environmental fit score, lifespan analysis, and fault diagnosis. Project feasibility score. The weighted percentage is 30%, and the ergonomics score is... The weighted percentage is 25%, and the environmental matching score is... The weighted percentage is 25%, for fault diagnosis. The weighted average is 15%. Project feasibility score. Based on graph-to-graph comparison and similarity scores from a database, the ergonomics score is... Based on simulation data scores of the simulated driver in the 3D model of the cockpit, the environmental matching score Based on the training of a vertical large language model, the matching analysis between design drawings and mine scene images is used for fault diagnosis. Fault diagnosis is performed based on a trained vertical large language model combined with a historical fault database of coal mine monorail cranes. The final objective evaluation score module... The calculation formula is:

[0034] in, , and For the weighting coefficients, take... It is 0.3. It is 0.25. It is 0.25. It is 0.15. The project feasibility score. Human-computer interaction score, The project feasibility score. For fault diagnosis, an objective evaluation score of 80 or higher is considered passing; otherwise, the concept generation graph needs to be reconstructed, and the evaluation metrics need to be updated.

[0035] Specifically, the subjective satisfaction assessment module Including security perception score and functional perception score Perceived security score The functional perception score is weighted at 50%. The weighted average is 50%. A pre-trained model performs a subjective evaluation of the design drawings, resulting in a safety-aware score. Includes safety protection, maintainability, and operational safety components; functional perception score. This includes operational efficiency and environmental adaptability. The final subjective evaluation score is... The calculation formula is:

[0036] in, and For the weighting coefficients, take... It is 0.5. It is 0.5. For the perceived safety score, This is the functional perception score. A subjective evaluation score of 80 or higher is considered passing. Conceptual drawings that fail to meet the passing score require collecting subjective feedback and revising the concept design.

[0037] In summary, the present invention provides a method for optimizing the shape design of a monorail crane based on generative artificial intelligence, comprising: acquiring user requirements and scenario data; constructing a multi-dimensional semantic association network for the monorail crane; generating a conceptual design diagram based on a vertical large language model pre-trained with coal mine knowledge and the semantic network; comparing the conceptual design diagram with user requirement text and existing product images in multiple dimensions, including semantic consistency comparison and engineering feasibility comparison; based on the comparison results, using a particle swarm optimization algorithm to iteratively optimize the parameters driving image generation until the conceptual design diagram meets the comprehensive evaluation indicators; and finally outputting the optimized solution through subjective and objective evaluation modules. The comprehensive evaluation indicators include a subjective evaluation module and an objective evaluation module. This invention deeply integrates conceptual creativity and engineering feasibility through an optimization cycle, providing designers with efficient and reliable design references.

[0038] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0039] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A method for optimizing the shape design of a monorail crane based on generative artificial intelligence, characterized in that, The method includes the following steps: S1. Based on knowledge in the coal mining field, train a vertical large language model to obtain a database of parameters and keywords for different types of monorail crane shapes; S2. Use the vertical large language model described in S1 to obtain user demand text and scenario data, and construct a multi-dimensional semantic association network and comprehensive evaluation index system for monorail cranes. S3. Based on the multi-dimensional semantic association network and comprehensive evaluation index system of the monorail crane, the conceptual design drawing of the monorail crane shape is generated from the vertical large model; S4. The monorail crane outline concept design drawing described in S3 is compared with the user requirement text and the existing monorail crane product images in the database in multiple dimensions, and the quantitative comparison results are output. The multi-dimensional comparison includes semantic consistency comparison and engineering feasibility comparison. Based on the quantitative comparison results, the parameters of the driving image generation model are iteratively adjusted using an optimization algorithm to generate an optimized concept design drawing. S5. First, conduct an objective comprehensive evaluation of the optimized conceptual design drawing in S4. After the evaluation, the user conducts a subjective satisfaction evaluation of the optimized conceptual design drawing. If the comprehensive evaluation is passed, the final monorail crane shape design scheme is output. If the objective comprehensive evaluation does not meet the requirements, the association network and evaluation indicators are reconstructed and returned to S4. If the subjective comprehensive evaluation does not meet the requirements, subjective modification opinions are collected, the conceptual design drawing is modified, and returned to S4.

2. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, S2 specifically includes: S2.1 Extract keywords from the user's request text, and retrieve the top N divergent information with the highest relevance in the semantic association network, starting from the keywords; S2.2 Based on the divergent information selected by the user, a design path is constructed in the semantic network, and the path information is transformed into a design feature vector that drives the image generation model. S2.3 Input the design feature vector into the image generation model to generate a conceptual design drawing of the monorail crane.

3. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, The parameters of the image generation model described in S4 include the power source, structural form, sensor arrangement of the vehicle body, cockpit space layout, ergonomic design, operating devices, and instrument panel design of the exterior of the driver's cab in the exterior image.

4. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, The semantic consistency comparison described in S4 is implemented using the CLIP model. That is, the CLIP model is used to extract the text feature vector of the user requirement text and the image feature vector of the concept design diagram respectively; the cosine similarity between the two feature vectors is calculated as the score of semantic consistency comparison.

5. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, The engineering feasibility comparison described in S4 is implemented using a pre-trained deep convolutional neural network. Specifically, the pre-trained deep convolutional neural network is used to extract the depth feature vectors of the conceptual design drawing and the existing monorail crane product image, respectively. The cosine similarity between the two depth feature vectors is calculated as the score for the engineering feasibility comparison.

6. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, The objective comprehensive assessment described in S5 includes an engineering feasibility score, based on graph-to-graph comparison and similarity to the product database; and an ergonomics score, based on simulation data of the digital model in the cockpit. Environmental matching score, based on AI model matching analysis of design drawings and mine scene.

7. The method for optimizing the shape design of a monorail crane based on generative artificial intelligence according to claim 1, characterized in that, In S5, subjective satisfaction assessment includes evaluations of perceived safety scores and perceived functionality scores; both the perceived safety scores and perceived functionality scores are obtained by users giving subjective scores.

8. A monorail crane shape design optimization system based on generative artificial intelligence, characterized in that, The system is implemented based on the method of claim 1, and the system includes: The information extraction module is used to train a vertical large language model based on knowledge in the coal mining field, and to obtain a database of various parameters and keywords of different types of monorail cranes. The information dissemination module is used to obtain user demand text and scenario data using the vertical large language model described in the information extraction module, and to construct a multi-dimensional semantic association network and a comprehensive evaluation index system for monorail cranes. The image generation module is used to generate a conceptual design drawing of the monorail crane from a large vertical model, based on the monorail crane's multi-dimensional semantic association network and comprehensive evaluation index system. The comparison module is used to perform multi-dimensional comparison between the monorail crane outline concept design drawing described in the design feature conversion module and the user requirement text, as well as the monorail crane product images already in the database, and output the quantitative comparison results. The multi-dimensional comparison includes semantic consistency comparison and engineering feasibility comparison. Based on the quantitative comparison results, the parameters of the driving image generation model are iteratively adjusted using an optimization algorithm to generate an optimized concept design drawing. The comprehensive evaluation module is used to first conduct an objective comprehensive evaluation of the optimized conceptual design drawings in the comparison module. After passing the evaluation, users conduct a subjective satisfaction evaluation of the optimized conceptual design drawings. If the comprehensive evaluation is passed, the final monorail crane shape design scheme is output. If the objective comprehensive evaluation does not meet the requirements, the association network and evaluation indicators are reconstructed and the project is returned to the comparison module. If the subjective comprehensive evaluation does not meet the requirements, subjective modification opinions are collected, the conceptual design drawings are modified, and the project is returned to the comparison module.

9. A computer storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1-7.

10. A computer device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the method of any one of claims 1-7.

Images