Intelligent design interaction system based on large model

Abstract

The invention provides an intelligent design interaction system based on a large model, relates to the technical field of electrical design, and realizes efficient automation of a design process by utilizing historical lighting power distribution design drawing data and constructing a scene template database and a parameter allocation model, and supports a user to input design requirements in an interaction mode. According to the method, personalized power distribution parameters are generated in a customized mode, specific requirements of different users and projects are met, hidden requirements of the users are deeply mined through questionnaires, matching thresholds are dynamically adjusted, user preferences can be accurately captured, scene templates are adjusted in a personalized mode, and therefore the user experience is improved while functional requirements are met. Compared with the prior art, the method has the advantages that the power distribution parameters are optimized, emotion and aesthetic appeals of users are highly met, scene templates highly matched with user requirements can be screened out, the design parameters are optimal, in addition, the optimized power distribution parameters are output through a parameter allocation model obtained through machine learning model training, and the design quality and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the field of electrical design technology, and in particular to an intelligent design interaction system based on a large model. Background Art

[0002] In the field of lighting power distribution design, with the increasing complexity and scale of construction projects, traditional design methods have been unable to meet current needs. For example, traditional design methods require engineers to manually calculate and analyze a large amount of data. Although the design effect is outstanding, it also leads to a long design cycle and cannot meet the rapidly changing market needs. At the same time, calculations are prone to errors, which increases design costs and time costs. Due to the different experience and levels of engineers, traditional design methods are difficult to ensure the consistency of design quality and are prone to errors, which brings difficulties to subsequent construction and maintenance. Moreover, with the increasing diversification and personalized needs of construction projects, traditional design methods are difficult to meet the specific needs of different users and projects. Engineers need to customize the design for each project, which further increases the difficulty and time cost of the design.

[0003] Therefore, it is necessary to provide an intelligent design interaction system based on large models to solve the above technical problems. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides an intelligent design interaction system based on a large model to solve the problems of low design efficiency, unstable design quality and difficulty in meeting personalized needs in traditional design methods.

[0005] The present invention provides a large model-based intelligent design interaction system, wherein the design interaction system comprises: A data processing module is used to obtain historical lighting distribution design drawing data for preprocessing, identify scene features in the lighting distribution design drawing, and construct a scene template database based on the scene features; A model building module is used to extract key feature parameters from historical lighting power distribution design diagram data, use the key feature parameters as input and the power distribution parameters of the lighting power distribution design diagram as output to train a machine learning model to obtain a parameter adjustment model; An information screening module is used to input design requirements in an interactive manner, extract scene information in the design requirements, and screen out all scene templates whose matching degree exceeds a matching threshold from a scene template database according to the scene information; A dynamic adjustment module is used to analyze the user's implicit preference emotions based on the questionnaire answered in advance by the user to dynamically adjust the matching threshold, perform secondary screening on the screened scene templates and adjust them to obtain the adjusted scene templates; The data analysis module is used to extract key characteristic parameters through the input design requirements, input them into the parameter allocation model, and output the current design power distribution parameters; The identification generation module is used to add the power distribution parameters of the current design to the adjusted scene template, analyze its adaptability, and select the scene template with the highest adaptability as the lighting distribution design diagram of the current design for display.

[0006] Preferably, the acquisition of historical lighting power distribution design drawing data for preprocessing, identifying scene features in the lighting power distribution design drawing, and constructing a scene template database based on the scene features, the specific steps are: Collect historical lighting distribution design data from different projects and regions and pre-process them, including image clarity and format unification; Using computer vision technology to identify scene features in historical lighting distribution design data, where scene features include room type, lamp layout, and switch location; The identified scene features and their corresponding complete historical lighting and power distribution design drawings are stored as entries in the scene template database.

[0007] Preferably, extracting key characteristic parameters from historical lighting power distribution design diagram data, taking the key characteristic parameters as input and the power distribution parameters of the lighting power distribution design diagram as output to train a machine learning model to obtain a parameter allocation model, the specific steps are: Extract key characteristic parameters from historical lighting power distribution design drawings, including lamp type, single lamp power and lighting level; The key characteristic parameters are used as input characteristics, and the distribution parameters of the lighting distribution design drawing are used as output parameters, and the mapping relationship between the input characteristics and the output parameters is marked, where the distribution parameters include line layout, circuit breaker specifications, and cable specifications; A machine learning model is adopted to train the machine learning model according to the mapping relationship between the labeled input features and the output parameters to obtain a parameter adjustment model.

[0008] Preferably, the step of inputting the design requirements in an interactive manner, extracting the scene information in the design requirements, and filtering out all scene templates whose matching degree exceeds a matching threshold from the scene template database according to the scene information comprises the following specific steps: The user inputs design requirements, including room type, lighting style, energy efficiency requirements, etc., through interactive methods, including text box input or integrated voice recognition module to convert voice into text; The scene information in the design requirements is extracted and compared with the entries in the scene template database, and all scene templates whose matching degree exceeds the preset matching threshold are screened out.

[0009] Preferably, the method of analyzing the user's implicit preference emotion based on the questionnaire answered in advance by the user to dynamically adjust the matching threshold, performing secondary screening and adjustment on the screened scene template to obtain the adjusted scene template, specifically comprises the following steps: Setting a questionnaire structure and sending it to a user terminal, wherein the questionnaire structure includes explicit demand collection and implicit demand collection; Collect users' responses to the questionnaire and use a natural language processing model to extract sentiment polarity and keywords from the responses; Based on the extracted sentiment polarity and keywords, the matching thresholds of the features corresponding to the sentiment polarity and keywords are adjusted to obtain an adjusted matching threshold; The screened scene templates are screened again according to the adjusted matching threshold, and the scene templates after the second screening are adjusted based on the corresponding features of the sentiment polarity and the keywords.

[0010] Preferably, the power distribution parameters of the current design are added to all the screened scene templates, and their adaptability is analyzed, and the scene template with the highest adaptability is screened out as the lighting power distribution design diagram of the current design for display, and the specific steps are: Supplementing the generated power distribution parameters into the adjusted scenario template to generate multiple candidate templates; Through the preset evaluation conditions, the adaptability of each candidate template after adding the new power distribution parameters is evaluated respectively, and multiple candidate templates are sorted according to the size of the adaptability to obtain the sorting result; Based on the sorting results, the scene template with the highest adaptability is selected as the optimal template; The optimal template is used as the lighting distribution design diagram of the current design and is displayed and fed back to the user.

[0011] Compared with the related art, the intelligent design interaction system based on large model provided by the present invention has the following beneficial effects: The present invention utilizes historical lighting distribution design drawing data to construct a scene template database and a parameter allocation model, thereby achieving efficient automation of the design process, greatly shortening the design cycle and improving design efficiency. At the same time, it supports users to input design requirements in an interactive manner, and customizes and generates personalized distribution parameters to meet the specific needs of different users and projects. Secondly, through questionnaires, it deeply explores users' hidden needs and dynamically adjusts the matching threshold, which can accurately capture user preferences and make personalized adjustments to scene templates, thereby meeting functional requirements while highly meeting users' emotions and aesthetic demands. Moreover, the present invention can also screen out scene templates that are highly matched with user needs and achieve the best design parameters. In addition, the parameter allocation model obtained through machine learning model training outputs optimized distribution parameters, thereby improving the quality and reliability of the design. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a system block diagram of an intelligent design interaction system based on a large model in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0014] The intelligent design interaction system based on the big model includes: A data processing module is used to obtain historical lighting distribution design drawing data for preprocessing, identify scene features in the lighting distribution design drawing, and construct a scene template database based on the scene features; A model building module is used to extract key feature parameters from historical lighting power distribution design diagram data, use the key feature parameters as input and the power distribution parameters of the lighting power distribution design diagram as output to train a machine learning model to obtain a parameter adjustment model; An information screening module is used to input design requirements in an interactive manner, extract scene information in the design requirements, and screen out all scene templates whose matching degree exceeds a matching threshold from a scene template database according to the scene information; The data analysis module is used to extract key characteristic parameters through the input design requirements, input them into the parameter allocation model, and output the current design power distribution parameters; The identification generation module is used to add the power distribution parameters of the current design to all the screened scene templates, analyze their adaptability, and screen out the scene template with the highest adaptability as the lighting power distribution design diagram of the current design for display; The feedback adjustment module is used to allow users to adjust parameters interactively based on the displayed lighting distribution design diagram, and generate the final lighting distribution design diagram after the user confirms.

[0015] In the specific implementation process, the historical lighting distribution design drawing data is obtained for preprocessing, and the scene features in the lighting distribution design drawing are identified. The specific steps of constructing a scene template database based on the scene features are as follows: The historical lighting distribution design data from different projects and regions are collected and preprocessed, where the preprocessing includes image clarity and format unification.

[0016] Specifically, historical lighting distribution design drawing data are collected from different projects and regions, where these historical lighting distribution design drawing data can come from publicly available data channels such as architectural design companies, power companies or public databases. Then, denoising and contrast enhancement in existing image processing technologies are used to improve the clarity of the lighting distribution design drawings for subsequent scene feature recognition. In addition, the collected design drawings are converted into a unified file format (JPEG, PNG or PDF), and the image size or resolution is adjusted to ensure consistency.

[0017] Computer vision technology is used to identify scene features in historical lighting distribution design diagram data, where scene features include room type, lamp layout and switch location.

[0018] Specifically, existing computer vision technologies, such as traditional image processing algorithms, are used to identify scene features in design drawings, specifically: identifying room types (such as offices, conference rooms, classrooms, etc.) by analyzing the layout, furniture, and decorative elements in the design drawings; identifying the location, type, and quantity of lamps in the design drawings, and how they are distributed in the room; identifying the location of switches in the design drawings, including wall switches, floor sockets, etc.

[0019] The identified scene features and their corresponding complete historical lighting and power distribution design drawings are stored as entries in the scene template database.

[0020] Specifically, the identified scene features and their corresponding complete historical lighting and power distribution design drawings are organized as entries, and a database management system (MySQL, MongoDB) is used to store these entries to build a scene template database.

[0021] In the specific implementation process, key feature parameters are extracted from the historical lighting power distribution design data, and the key feature parameters are used as input, and the power distribution parameters of the lighting power distribution design are used as output to train the machine learning model. The specific steps to obtain the parameter allocation model are as follows: Extract key feature parameters from historical lighting distribution design drawings, including lamp type, single lamp power and lighting level.

[0022] Specifically, key feature parameters are extracted from historical lighting distribution design drawings, specifically: identifying the types of lamps used in the lighting distribution design drawings, such as fluorescent lamps, LED lamps, halogen lamps, etc.; obtaining the single lamp power of each lamp by measuring or consulting the lamp specification table; according to the lighting layout and the number of lamps in the design drawing, in accordance with the existing lighting standards, identifying the lighting level, such as standard lighting or energy-saving lighting.

[0023] The key characteristic parameters are used as input features, the distribution parameters of the lighting distribution design drawing are used as output parameters, and the mapping relationship between the input characteristics and the output parameters is marked. The distribution parameters include line layout, circuit breaker specifications, and cable specifications.

[0024] Specifically, the input features include: lamp type, single lamp power, lighting level; output parameters include: line layout, circuit breaker specifications, cable specifications; for each historical lighting distribution design drawing, the corresponding mapping relationship between input features and output parameters is marked according to its actual distribution parameters, and the accuracy and completeness of the marked data are ensured to facilitate the subsequent training of the machine learning model.

[0025] For example, for an office lighting power distribution design diagram, the corresponding power distribution parameters include: the line layout is a branch circuit, the circuit breaker specification is 16A, and the cable specification is 2.5mm². These power distribution parameters are annotated with the extracted key feature parameters to form a data set for training the machine learning model.

[0026] A machine learning model is adopted to train the machine learning model according to the mapping relationship between the labeled input features and the output parameters to obtain a parameter adjustment model.

[0027] Specifically, a neural network in a machine learning model is selected, and the mapping relationship between labeled input features and output parameters is used as a training data set. The machine learning model is trained so that it can learn the mapping relationship between input features and output parameters. After the training is completed, a parameter adjustment model is obtained.

[0028] In the specific implementation process, the design requirements are input in an interactive manner, and the scene information in the design requirements is extracted. The specific steps of filtering out all scene templates whose matching degree exceeds the matching threshold from the scene template database according to the scene information are as follows: Users input design requirements, including room type, lighting style, energy efficiency requirements, etc., through interactive methods, where the interactive methods include text box input or integrating a voice recognition module to convert voice into text.

[0029] Specifically, the user connects to the system through the web interface on the terminal device (computer, tablet or mobile phone), and enters text into the text box to describe the design requirements, or the voice recognition module recognizes the text to form the input design requirements, where the design requirements include room type (such as office, conference room, classroom, etc.), lighting style (such as modern, simple, retro, etc.), energy efficiency requirements (such as energy saving, high efficiency, etc.) and other related parameters.

[0030] The scene information in the design requirements is extracted and compared with the entries in the scene template database, and all scene templates whose matching degree exceeds the preset matching threshold are screened out.

[0031] Specifically, the system extracts scene information from the design requirements input by the user, such as room type, lighting style, and energy efficiency requirements. Then, it pre-labels each scene template, compares the extracted scene information with the entries in the scene template database, and calculates the label overlap, which is the matching degree. Among them, matching degree = number of overlapping labels / total number of template labels. After calculating the matching degree of all scene templates, all scene templates whose matching degree exceeds the preset matching threshold are screened out.

[0032] In the specific implementation process, based on the questionnaire answered by the user in advance, the user's implicit preference emotion is analyzed to dynamically adjust the matching threshold, and the screened scene templates are screened again and adjusted to obtain the adjusted scene templates. The specific steps are: A questionnaire structure is set and sent to a user terminal, wherein the questionnaire structure includes explicit demand collection and implicit demand collection.

[0033] Specifically, set up a questionnaire structure. For example, set up the collection of explicit needs: such as what lighting brightness level the user expects, whether intelligent control functions are needed, etc.; set up the collection of implicit needs: such as in an ideal working environment, how the lighting effect should affect the user's mood or work efficiency, whether the lighting design is expected to reflect a specific style or atmosphere, and send the designed questionnaire to the user terminal via email, online form, mobile application push, etc.

[0034] The user's responses to the questionnaire are collected, and the natural language processing model is used to extract sentiment polarity and keywords from the responses.

[0035] Specifically, users submit their responses through the questionnaire, and then these responses are collected and sorted, and the responses are analyzed using the natural language processing (NLP) model to extract the emotional polarity (such as positive, negative, neutral) and keywords (such as warmth, energy saving, and intelligence) in the responses. For example, the user hopes that the lighting in the living room is both warm and energy-saving, and it is best to be intelligently controlled, then the extraction result is: emotional polarity = positive, keywords = warmth, energy saving, and intelligent control.

[0036] Based on the extracted sentiment polarity and keywords, the matching thresholds of features corresponding to the sentiment polarity and keywords are adjusted to obtain an adjusted matching threshold.

[0037] Specifically, based on the extracted sentiment polarity and keywords, the degree of importance that users place on different features is analyzed, and the threshold for scene template matching is adjusted so that scene templates that are more in line with user preferences can be more easily screened out. For example, if the user mentions "warmth" multiple times and the sentiment polarity is positive, the matching threshold of scene templates that do not match the "warmth" feature is lowered, while the matching threshold of scene templates that highly match the "warmth" feature is increased.

[0038] The screened scene templates are screened again according to the adjusted matching threshold, and the scene templates after the second screening are adjusted based on the corresponding features of the sentiment polarity and the keywords.

[0039] Specifically, the scene templates initially screened out are screened again using the adjusted matching threshold, and the scene templates whose matching degree exceeds the new threshold are retained. The scene templates after the second screening are adjusted according to the emotional polarity and keywords in the user's response. For example, based on the extracted emotional polarity and keywords, the types of lamps that the user may prefer are increased, and the lighting layout is adjusted to match the user's desired atmosphere.

[0040] In the specific implementation process, key characteristic parameters are extracted through the input design requirements, and input into the parameter allocation model to output the current design distribution parameters.

[0041] In the specific implementation process, the power distribution parameters of the current design are added to the adjusted scene template, and its adaptability is analyzed, and the scene template with the highest adaptability is selected as the lighting power distribution design diagram of the current design. The specific steps for display are: The generated power distribution parameters are added to the adjusted scenario template to generate multiple candidate templates.

[0042] Specifically, the generated power distribution parameters (such as line layout, circuit breaker specifications, cable specifications, etc.) are added to these adjusted scenario templates to form multiple candidate templates containing new power distribution parameters.

[0043] Exemplarily, after adjustment, three scene templates are obtained: Template A (office lighting), Template B (conference room lighting), and Template C (corridor lighting). Then, power distribution parameters for office lighting are generated, including specific line layout, circuit breaker specifications, and cable specifications. Finally, these power distribution parameters are added to Template A to form candidate Template A.

[0044] Through the preset evaluation conditions, the adaptability of each candidate template after adding the new power distribution parameters is evaluated respectively, and multiple candidate templates are sorted according to the size of the adaptability to obtain the sorting result.

[0045] Specifically, the preset evaluation conditions are as follows: define the core evaluation dimensions of the candidate templates, including voltage compatibility, load capacity matching, and safety specification, and set the evaluation coefficients of each core evaluation dimension. In this embodiment, the evaluation coefficients of voltage compatibility, load capacity matching, and safety specification are set to 0.4, 0.3, and 0.3, respectively, and the calculation formula for the adaptability is: adaptability = voltage compatibility * 0.4 + load capacity matching * 0.3 + safety specification * 0.3; after evaluating the adaptability of each candidate template after adding the new power distribution parameters, sort them in descending order to obtain the sorting result.

[0046] Based on the sorting results, the scene template with the highest adaptability is selected as the optimal template.

[0047] For example, if the fitness of candidate template A is 90, the fitness of candidate template B is 7, and the fitness of candidate template C is 60, candidate template A is selected as the optimal template.

[0048] The optimal template is used as the lighting distribution design diagram of the current design and is displayed and fed back to the user.

[0049] Specifically, the optimal template is used as the lighting power distribution design drawing of the current design, and is displayed to the user through a terminal device (such as a computer, tablet, etc.). The user can see detailed line layout, circuit breaker specifications, cable specifications and other information, and make further confirmation or adjustments as needed, such as: The scene template with the highest adaptability is screened out to generate the current lighting distribution design drawing, and it is displayed to the user through the interface of the terminal device (such as a computer, tablet or mobile phone); the user interactively inputs the parameters that need to be adjusted in the lighting distribution design drawing, and the parameter allocation model adjusts them separately to obtain the adjusted lighting distribution design drawing; the user receives the adjusted lighting distribution design drawing through the terminal device, and after confirmation, the adjusted lighting distribution design drawing becomes the final lighting distribution design drawing.

[0050] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0051] The present invention utilizes historical lighting distribution design drawing data to construct a scene template database and a parameter allocation model, thereby achieving efficient automation of the design process, greatly shortening the design cycle and improving design efficiency. At the same time, it supports users to input design requirements in an interactive manner, and customizes and generates personalized distribution parameters to meet the specific needs of different users and projects. Secondly, the present invention can screen out scene templates that are highly matched with user needs and achieve the best design parameters. In addition, the parameter allocation model obtained through machine learning model training outputs optimized distribution parameters, which improves the quality and reliability of the design. The user's interactive method is used to make the final adjustment to the parameters of the lighting distribution design drawing, further improving the design flexibility and user satisfaction.

[0052] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0053] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. Intelligent design interaction system based on large model, characterized by: The design interaction system comprises: A data processing module is used to obtain historical lighting distribution design drawing data for preprocessing, identify scene features in the lighting distribution design drawing, and construct a scene template database based on the scene features; A model building module is used to extract key feature parameters from historical lighting power distribution design diagram data, use the key feature parameters as input and the power distribution parameters of the lighting power distribution design diagram as output to train a machine learning model to obtain a parameter adjustment model; An information screening module is used to input design requirements in an interactive manner, extract scene information in the design requirements, and screen out all scene templates whose matching degree exceeds a matching threshold from a scene template database according to the scene information; A dynamic adjustment module is used to analyze the user's implicit preference emotions based on the questionnaire answered in advance by the user to dynamically adjust the matching threshold, perform secondary screening on the screened scene templates and adjust them to obtain the adjusted scene templates; The data analysis module is used to extract key characteristic parameters through the input design requirements, input them into the parameter allocation model, and output the current design power distribution parameters; The identification generation module is used to add the power distribution parameters of the current design to the adjusted scene template, analyze its adaptability, and select the scene template with the highest adaptability as the lighting distribution design diagram of the current design for display.

2. The intelligent design interaction system based on a large model according to claim 1 is characterized in that: The acquisition of historical lighting power distribution design drawing data for preprocessing, and identification of scene features in the lighting power distribution design drawing, and construction of a scene template database based on the scene features, the specific steps are as follows: Collect historical lighting distribution design data from different projects and regions and pre-process them, including image clarity and format unification; Using computer vision technology to identify scene features in historical lighting distribution design data, where scene features include room type, lamp layout, and switch location; The identified scene features and their corresponding complete historical lighting and power distribution design drawings are stored as entries in the scene template database.

3. The intelligent design interaction system based on a large model according to claim 2 is characterized in that: The key characteristic parameters are extracted from the historical lighting power distribution design diagram data, and the key characteristic parameters are used as input, and the power distribution parameters of the lighting power distribution design diagram are used as output to train the machine learning model to obtain the parameter allocation model, and the specific steps are: Extract key characteristic parameters from historical lighting power distribution design drawings, including lamp type, single lamp power and lighting level; The key characteristic parameters are used as input characteristics, and the distribution parameters of the lighting distribution design drawing are used as output parameters, and the mapping relationship between the input characteristics and the output parameters is marked, where the distribution parameters include line layout, circuit breaker specifications, and cable specifications; A machine learning model is adopted to train the machine learning model according to the mapping relationship between the labeled input features and the output parameters to obtain a parameter adjustment model.

4. The intelligent design interaction system based on a large model according to claim 3 is characterized in that: The design requirements are input in an interactive manner, and scene information in the design requirements is extracted. All scene templates with matching degrees exceeding a matching threshold are screened out from a scene template database according to the scene information. The specific steps are as follows: The user inputs design requirements, including room type, lighting style, energy efficiency requirements, etc., through interactive methods, where the interactive methods include text box input or integrated voice recognition module to convert voice into text; The scene information in the design requirements is extracted and compared with the entries in the scene template database, and all scene templates whose matching degree exceeds the preset matching threshold are screened out.

5. The intelligent design interaction system based on large models according to claim 4 is characterized in that: The method of analyzing the user's implicit preference emotion based on the questionnaire answered in advance by the user to dynamically adjust the matching threshold, performing secondary screening and adjustment on the screened scene templates to obtain the adjusted scene templates, specifically comprises the following steps: Setting a questionnaire structure and sending it to a user terminal, wherein the questionnaire structure includes explicit demand collection and implicit demand collection; Collect users' responses to the questionnaire and use a natural language processing model to extract sentiment polarity and keywords from the responses; Based on the extracted sentiment polarity and keywords, the matching thresholds of the features corresponding to the sentiment polarity and keywords are adjusted to obtain an adjusted matching threshold; The screened scene templates are screened again according to the adjusted matching threshold, and the scene templates after the second screening are adjusted based on the corresponding features of the sentiment polarity and the keywords.

6. The intelligent design interaction system based on large models according to claim 5 is characterized in that: The specific steps of adding the power distribution parameters of the current design to the adjusted scene template, analyzing its adaptability, and selecting the scene template with the highest adaptability as the lighting power distribution design diagram of the current design for display are as follows: Supplementing the generated power distribution parameters into the adjusted scenario template to generate multiple candidate templates; Through the preset evaluation conditions, the adaptability of each candidate template after adding the new power distribution parameters is evaluated respectively, and multiple candidate templates are sorted according to the size of the adaptability to obtain the sorting result; Based on the sorting results, the scene template with the highest adaptability is selected as the optimal template; The optimal template is used as the lighting distribution design diagram of the current design and is displayed to the user.

Images