Artificial intelligence-driven security house furniture modular combination parameter generation method

Abstract

The invention discloses an artificial intelligence-driven security house furniture modular combination parameter generation method, and belongs to the field of modular furniture design, and the method comprises the following steps: S1, obtaining a function demand list and a function demand weight; s2, constructing a structure-material-interface modular unit parameter library based on the function demand list and the function demand weight; s3, based on the modular unit parameter library, generating a multi-level adaptation scheme; s4, based on a multi-stage adaptation scheme, determining an optimal combination parameter scheme through NSGA-II algorithm optimization and AD axiom design verification; and S5, collecting furniture use state data, and dynamically updating the modular unit parameter library. By adopting the artificial intelligence-driven security house furniture modular combination parameter generation method, precise adaptation, efficient generation and continuous optimization of security house furniture combination parameters are realized, and the pain points of poor adaptability, low efficiency and insufficient individuation of traditional design are overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of modular furniture design, and particularly relates to an AI-driven affordable housing furniture modular combination parameter generation method. BACKGROUND

[0002] At present, the house type of affordable housing is mainly small and medium-sized (main area 40-60m 2 ), the space layout is compact (such as the average face width of the living room is only 2.5-3m, and the net depth of the bedroom is less than 3m), and the user group structure is diverse (including newly married families, elderly people living alone, disabled families, etc.), the space adaptability, function complexity and cost controllability of furniture are significantly higher than those of ordinary commercial housing. At the same time, the existing furniture is mostly standardized mass production products (such as fixed size wardrobe, non-expandable dining table), or is designed only for ordinary families, which is difficult to meet the special needs of high utilization of small space, low budget and high quality, and multi-scene adaptation of affordable housing, resulting in additional self-reform of users (reform rate more than 60%), or space waste and function redundancy (such as large size sofa occupying 30% of the living room space) due to the mismatch between furniture and space / demands, which seriously affects the living experience.

[0003] From the user demand dimension, there are three core pain points of affordable housing users: first, space anxiety, the furniture size does not match the space in small house type (such as traditional 600mm deep bookcase leads to narrow movement line in bedroom), the furniture needs to have the ability of expansion, folding and modular combination; second, cost sensitivity, the user's per capita furniture expenditure is generally less than 2000 yuan (only 1 / 3 of that of commercial housing users), the cost of materials and manufacturing needs to be controlled, while the durability (expected service life ≥8 years) is also considered; third, diverse needs, the elderly family needs barrier-free handrails and lightweight operation, the family of three needs child safety protection and more storage space, and traditional standardized furniture cannot meet such individualized needs.

[0004] The existing demand modeling method mainly relies on questionnaire survey + expert judgment, which has the following shortcomings in the affordable housing scene: 1. Insufficient ability to capture implicit needs: traditional research (such as paper questionnaire, offline interview) can only obtain explicit needs such as "need storage" and "need dining table", and cannot identify the implicit demands of affordable housing users - for example, the "expandable size demand" behind the user feedback "the space is too small" and the "barrier-free handrail demand" behind the "difficulty in getting up" of the elderly. Although existing research has introduced Kano model, user portrait and other methods, it has not combined with emotional tendency analysis in the affordable housing scene, resulting in less than 50% accuracy rate of implicit demand capture.

[0005] 2. Large subjective bias in demand priority calculation: existing demand weight determination mostly uses a single method: either relying on expert subjective scoring of AHP (analytic hierarchy process), or relying on objective data dispersion calculation of entropy weight method. Among them, the single AHP method is easily affected by the experience of experts (such as setting the "aesthetics" weight to 0.2, far exceeding the actual concern of 0.05 of low-income housing users), and the weight error can reach 15%; the single entropy weight method is only based on data dispersion, and cannot reflect the special appeal of "space adaptation priority over aesthetics" of low-income housing. SUMMARY

[0006] The purpose of the present application is to provide an artificial intelligence driven low-income housing furniture modular combination parameter generation method to solve the above technical problems.

[0007] To achieve the above purpose, the present application provides an artificial intelligence driven low-income housing furniture modular combination parameter generation method, comprising the following steps: S1, combine the low-income housing house type database, user portrait, online questionnaire and network text to construct the original demand data set, calculate the demand priority weight based on AHP-entropy weight method, and obtain the function demand list and function demand weight by using QFD function demand mapping; S2, based on the function demand list and function demand weight output by S1, combined with the low-income housing space characteristic division, through the mortise and tenon parameter optimization, sustainable material selection, finite element strength verification and general interface design, the modular unit parameter library of structure-material-interface is constructed; S3, based on the modular unit parameter library constructed in S2, combined with the low-income housing target space parameters, construct the labeled data set and train the CGAN-dynamic rule hybrid model, generate the preliminary combination parameter scheme, and then screen the multi-level adaptive scheme through expert semantic differentiation evaluation; S4, based on the multi-level adaptive scheme output by S3, combined with the function demand weight of S1 to construct the space-strength-economy-sustainability optimization model, and determine the optimal combination parameter scheme through NSGA-II algorithm optimization and AD axiom design verification; S5, based on the optimal combination parameter scheme output by S4 to generate a standardized executable parameter file, deploy IoT sensors to collect furniture usage state data, train parameter adjustment model and dynamically update the modular unit parameter library described in step S2.

[0008] Therefore, the present application adopts the above-mentioned artificial intelligence driven low-income housing furniture modular combination parameter generation method, which has the beneficial effects of: 1. Precise improvement of space adaptability: based on the dynamic condition vector integrating low-income housing space size, layout constraints and other parameters, combined with multi-objective optimization algorithm, the furniture combination space utilization rate is improved to more than 85%, avoiding problems such as size overrun and layout conflict.

[0009] 2. Significantly improved generation efficiency: Automatically generate combination parameters with CGAN-dynamic rule hybrid model, replace traditional manual repeated adjustment process, compress single-family furniture parameter design period from several days to hours, greatly reduce design cost; 3. Deeply meet personalized needs: Quantify user space adaptation, economic cost, and needs weight of elderly and young children through AHP-entropy weight method, generate scheme to accurately match different family structure and special demands, and improve living experience; 4. Balance between economy and sustainability: Multi-objective optimization considers material cost, recyclability and environmental protection requirements, uses standardized modules to reduce production and maintenance cost, and increase recyclable material ratio, which meets the low-carbon economic demand of indemnificatory housing; 5. Dynamic iteration of parameter system: Continuously update modular parameter library through parameter adjustment formula based on IoT collected usage data, adapt to actual usage scenarios such as furniture wear and deformation, and prolong product life cycle; 6. Consideration of universality and compliance: Compatible with different house types, budgets and policy specifications, the generated scheme automatically meets the requirements of structural safety, environmental protection standards, etc., reduces construction rework rate, and adapts to the scenarios of indemnificatory housing projects in multiple regions.

[0010] The technical solutions of the present application will be further described in detail below with the aid of drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0011] Figure 1 The flowchart of the present application is a kind of artificial intelligence driven indemnificatory housing furniture modular combination parameter generation method. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the embodiments of the present application will be further described in detail below with the aid of drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present application and not to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application. The examples of the embodiments are shown in the drawings, wherein the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout.

[0013] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those clearly listed steps or units, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0014] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0015] like Figure 1 As shown, an AI-driven method for generating modular combination parameters for affordable housing furniture includes the following steps: S1. Combine the affordable housing unit type database, user profiles, online questionnaires and online texts to construct the original demand dataset, calculate the demand priority weights based on the AHP-entropy weight method, and then use QFD functional demand mapping to obtain the functional demand list and functional demand weights. Step S1 specifically includes the following steps: S11. Construct the original demand dataset and mine the implicit demands from the original demand dataset to obtain a structured demand list: First, construct a corpus of affordable housing furniture demand text containing labeled samples. Design a BERT classification head for sentiment tendency recognition and user demand type classification for the affordable housing furniture demand text in the affordable housing furniture demand corpus to achieve joint sentiment-type recognition, output sentiment classification probability and user demand type classification probability, and use the SentiStrength algorithm to calculate sentiment score. ; S12. Calculate the demand priority weights based on the AHP-entropy weight method: ; in, ; ; ; ; ; In the formula, Indicates the first User needs in a sample of emotional intensity The weights, and These respectively represent spatial adaptation requirements, economic cost requirements, functional fulfillment requirements, accessibility requirements, and sustainability requirements; Indicate user needs Subjective weights in AHP; Indicate user needs Entropy weight is an objective weight. This represents the total number of categories of user needs. ; Indicate user needs relative to user needs The importance of; Indicate user needs relative to user needs The importance of; entropy value representing user demand ; entropy value representing user demand in the first emotional intensity sample; total number of emotional intensity samples; standardized emotional score; emotional score of the user demand in the first emotional intensity sample, ; and respectively represent the maximum and minimum values in the emotional score; S13, QFD function requirement mapping: first, determine the user demand and the corresponding comprehensive weight based on the requirement priority weight, and after adding the constraint list, standardize to obtain the standardized user demand and the corresponding comprehensive weight, then establish the user demand-function demand mapping relationship to obtain the indemnificatory housing quality house, then use expert scoring to obtain the association strength of user demand and function demand, and finally calculate the function demand weight: ; wherein, ; ; In the formula, represent the weight of the function requirement ; represent the initial weight of the function requirement ; represent the indemnificatory housing constraint correction coefficient of the function requirement ; represent the total number of function requirements; represent the association strength between the user demand and the function requirement ; represent the number of experts; represent the expert 's scoring of the association strength between the user demand and the function requirement .

[0016] Step S11 specifically includes the following steps: S111, define the task target as emotional tendency classification and user demand type classification, wherein the emotional tendency classification includes positive, negative and neutral; S112, construct an indemnificatory housing furniture demand corpus containing labeled samples, wherein the labeling includes emotional tendency labeling and user demand type labeling, to obtain a labeled data set; S113, divide the labeled data set into training set, validation set and test set according to 7:2:1; S114, adopt BERT-Base-Chinese as pre-training model, design sentiment tendency recognition-user demand type classification parallel branch type classification head and weighted joint loss function, and iterate training until reaching the standard, wherein the expression of the weighted joint loss function is as follows: ; Among them, ; ; In the formula, Total loss is represented by L; And Sentiment classification weight and user demand type classification weight are represented by And Sentiment classification loss and user demand type classification loss are represented by The true label of the th sentiment tendency annotation is represented by The sentiment type is represented by The probability that the th sentiment tendency annotation is predicted as the sentiment type is represented by The true label of the th sentiment tendency annotation is represented by The user demand type is represented by The probability that the th sentiment tendency annotation is predicted as the user demand type is represented by The total number of sentiment tendency annotations is represented by N; S115, calculate the sentiment score by using SentiStrength algorithm, and obtain the sentiment intensity sample: ; ; ; ; In the formula, The basic sentiment word score is represented by The security housing furniture field weight of the th sentiment word in the th security housing furniture demand text is represented by The security housing furniture field weight of the TF-IDF value of the sentiment word in the text of the 1st th guarantee house furniture demand text, number of sentiment words in the text of the 1st th guarantee house furniture demand text; length normalization adjustment factor; context correction term; the number of occurrences of the sentiment word in a certain sentiment category; the total number of occurrences of the sentiment word in all sentiment categories; the number of occurrences of the 1st th sentiment word in the text of the 1st th guarantee house furniture demand text; inverse document frequency, and , total number of guarantee house furniture demand texts; effective sentiment word number of the 1st th guarantee house furniture demand text; average attention weight of the sentiment word

[0017] S2, based on the function requirement list and the function requirement weight output by S1, combined with the guarantee house layout space characteristics division, through the mortise and tenon parameter optimization, sustainable material selection, finite element strength verification and general interface design, a modular unit parameter library of structure-material-interface is constructed; Step S2 specifically includes the following steps: S21, divide the modular unit according to the space reuse scene, match the function requirement described in step S1, determine the core module type and function boundary, and ensure that each module independently realizes 1 basic function and reserves a general interface; S22, optimize the mortise and tenon structure parameters for the light weight demand of guarantee house, based on the module plate thickness and the main body width , determine the benefit mortise and tenon width under the constraint condition , and determine the concession mortise and tenon depth under the constraint condition ; wherein, assembly gap; processing safety allowance; S23, screen the material combination based on the material database, and match the function requirement in step S1; S24, under the premise that the maximum stress of the module and the material allowable stress and the maximum displacement are not greater than the corresponding set threshold, the Autodesk Inventor Nastran is used to perform finite element strength verification on the material combination screened in step S23, the furniture use load and frequency of the security housing are simulated, the material combination screened in step S23 is verified, if qualified, step S25 is performed, otherwise, step S23 is returned; S25, a mixed universal interface of buckle and mortise is designed, and the matching degree of the universal interface is judged whether the requirement is met, if yes, step S26 is performed, otherwise, step S23 is returned; wherein the matching degree of the universal interface is calculated according to the following formula: ; In the formula, represents the size coincidence area of the universal interface; represents the total area of the interface; represents the tolerance adaptation range; represents the total tolerance range; S26, the structure parameters, material parameters and interface parameters of steps S22-S25 are integrated to obtain a modular unit parameter library containing module ID, function type and adaptation scenario.

[0018] S3, based on the modular unit parameter library constructed in S2, the target space parameters of the security housing are combined to construct a labeled data set and train a CGAN-dynamic rule hybrid model, and a preliminary combination parameter scheme is generated, and then a multi-level adaptation scheme is screened through expert semantic differentiation evaluation; Step S3 specifically includes the following steps: S31, based on the modular unit parameter library, the security housing house type data is combined to construct a labeled data set, and the space length-width ratio and family population number characteristics are introduced to adapt to different user portraits in S1; S32, the CGAN-dynamic rule hybrid model is trained by using the labeled data set, and the generated combination parameter scheme is iterated adaptation rate ≥ 85%; wherein the generator G of the CGAN-dynamic rule hybrid model inputs random noise and a dynamic condition vector which includes a security housing space parameter sub-vector, a user demand weight sub-vector, a family structure and scene sub-vector, and a budget and sustainability sub-vector, outputs a combination parameter scheme , and the loss function of the CGAN-dynamic rule hybrid model is expressed as follows: ; wherein, ; In the formula, Indicates random noise and dynamic condition vector Mathematical expectation operation; Indicates the discriminator; Indicates a generator; The weighting coefficients represent the rule loss; Represents the loss of dynamic rules; This indicates that the combined parameter scheme satisfies the constraint rules. The degree; Represents dynamic condition vector The The minimum satisfaction threshold for a constraint rule; The input to discriminator D is a combination parameter scheme. and dynamic condition vector Output a truth score, and its loss function is... The expression is as follows: ; In the formula, Represents the actual combination of parameters scheme and dynamic condition vector Mathematical expectation operation; This indicates that the discriminator operates under a given dynamic condition vector. At that time, for the actual combined parameter scheme The probability of predicting the accuracy of the prediction; S33. Experts review the generated multiple combined parameter schemes. Conduct a scoring and evaluation: ; In the formula, Indicates the scoring result; Indicates dimensional weights. These respectively represent spatial adaptability, module compatibility, and functional satisfaction. Experts Dimension The rating; Indicates the total number of experts; S34. Combining multiple parameter schemes Rating results Sort in descending order and take the first few. A combination of parameter schemes is used as the adaptation scheme.

[0019] S4. Based on the multi-level adaptation scheme output by S3, a space-intensity-economic-sustainability optimization model is constructed by combining the functional requirement weights of S1. The optimal combination parameter scheme is determined by the NSGA-II algorithm for optimization and the AD axiom design verification. Step S4 specifically includes the following steps: S41, construct a space-strength-economy-sustainability optimization model, the objective function of which is The expression is as follows: ; Wherein, ; ; ; ; In the formula, , , And Respectively represent the space utilization rate, the structural strength safety factor, the economy, the sustainability target weight, and ; , , , Respectively represent the space utilization rate, the strength safety factor, the economy, and the sustainability; Indicates the specific volume of furniture; Indicates the space volume; Indicates the allowable stress of furniture material; Indicates the maximum stress of furniture; Indicates the total cost; Indicates the upper limit of the budget for affordable housing furniture; Indicates the material degradation period; Indicates the service life; Indicates the proportion of recyclable materials; Constraint conditions: ; S42, optimization using NSGA-II algorithm: set the population size to 50, the number of iterations to 100, the crossover probability to 0.8, and the mutation probability to 0.1, and take the objective function As the fitness function, traverse the Adaptation schemes output in step S3, and output the Pareto optimal solution as the candidate combination parameter scheme; S43, AD axiom-based design verification: construct a function requirement-design parameter mapping matrix: ; In the formula, , , And Respectively represent space adaptation, structural strength, economic cost, and sustainability; , , And respectively represent the module size, material thickness, material cost and recyclable material proportion; determine whether the function requirement-design parameter mapping matrix satisfies non-diagonal element ≤0.05 to ensure that the function is not coupled, if yes, pass the verification, otherwise delete the corresponding combination parameter scheme; S44, screening target function The scheme with the maximum value and satisfying the constraint condition is taken as the optimal combination parameter scheme.

[0020] S5, based on the optimal combination parameter scheme output by S4, generate a standardized executable parameter file, deploy IoT sensors to collect furniture usage state data, train parameter adjustment model and dynamically update the modular unit parameter library in step S2.

[0021] Step S5 specifically includes the following steps: S51, generate a JSON format standardized parameter file, the standardized parameter file contains scheme ID, adaptive scene, module ID, module number, module position coordinates, structure parameters, material parameters, assembly and production instructions, for use in security housing furniture production and manufacturing; S52, collect stress, deformation and usage frequency at the interface of the furniture module to obtain a usage state data set; S53, input the usage state data set into the random forest parameter adjustment model to obtain the parameter adjustment amount in the modular unit parameter library ; In the formula, represents the adjustment margin; represents the optimal combination parameter obtained in step S4; represents the current combination parameter based on prediction; represents the usage state data set; S54, based on the parameter adjustment amount update the modular unit parameter library in step S2.

[0022] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application and not to limit it, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that: it can still modify or equivalently replace the technical solutions of the present application, and these modifications or equivalent replacements also cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.​

Claims

1. A method for generating modular combination parameters for affordable housing furniture driven by artificial intelligence, characterized in that: Includes the following steps: S1. Combine the affordable housing unit type database, user profiles, online questionnaires and online texts to construct the original demand dataset, calculate the demand priority weights based on the AHP-entropy weight method, and then use QFD functional demand mapping to obtain the functional demand list and functional demand weights. S2. Based on the functional requirement list and functional requirement weights output by S1, and combined with the spatial characteristics of affordable housing units, a modular unit parameter library of structure-material-interface is constructed through mortise and tenon parameter optimization, sustainable material selection, finite element strength verification and general interface design. S3. Based on S2, a modular unit parameter library is built. Combined with the target spatial parameters of affordable housing, a labeled dataset is constructed and a CGAN-dynamic rule hybrid model is trained. After generating a preliminary combination parameter scheme, a multi-level adaptation scheme is selected through expert semantic differentiation evaluation. S4. Based on the multi-level adaptation scheme output by S3, a space-intensity-economic-sustainability optimization model is constructed by combining the functional requirement weights of S1. The optimal combination parameter scheme is determined by the NSGA-II algorithm for optimization and the AD axiom design verification. S5. Based on the optimal combination parameter scheme output by S4, generate a standardized executable parameter file, deploy IoT sensors to collect furniture usage status data, train the parameter adjustment model, and dynamically update the modular unit parameter library described in step S2.

2. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 1, characterized in that: Step S1 specifically includes the following steps: S11. Construct the original demand dataset and mine the implicit demands from the original demand dataset to obtain a structured demand list: First, construct a corpus of affordable housing furniture demand text containing labeled samples. Design a BERT classification head for sentiment tendency recognition and user demand type classification for the affordable housing furniture demand text in the affordable housing furniture demand corpus to achieve joint sentiment-type recognition, output sentiment classification probability and user demand type classification probability, and use the SentiStrength algorithm to calculate sentiment score. ; S12. Calculate the demand priority weights based on the AHP-entropy weight method: ； in, ； ； ; ； ; In the formula, Indicates the first User needs in a sample of emotional intensity The weights, and These respectively represent spatial adaptation requirements, economic cost requirements, functional fulfillment requirements, accessibility requirements, and sustainability requirements; Indicate user needs Subjective weights in AHP; Indicate user needs Entropy weight is an objective weight. This represents the total number of categories of user needs. ; Indicate user needs relative to user needs The importance of; Indicate user needs relative to user needs The importance of; Indicate user needs The entropy value; Indicate user needs In the The percentage of emotional intensity in each emotional intensity sample; This represents the total number of samples indicating emotional intensity. This represents the standardized sentiment score; Indicates the first User needs in a sample of emotional intensity Emotional score ; and These represent the maximum and minimum values ​​in the sentiment score, respectively. S13, QFD Functional Requirements Mapping: First, user requirements and their corresponding comprehensive weights are determined based on requirement priority weights. After adding a constraint list, these are standardized to obtain standardized user requirements and their corresponding comprehensive weights. Then, a user requirement-functional requirement mapping relationship is established to obtain the quality of affordable housing. Next, expert scoring is used to obtain the correlation strength between user requirements and functional requirements. Finally, the functional requirement weights are calculated. ； in, ； ； In the formula, Indicate functional requirements The weights; Indicate functional requirements The initial weights; Indicate functional requirements The adjustment coefficient for affordable housing constraints; Indicates the total number of functional requirements; Indicate user needs Functional requirements The strength of the correlation between them; Indicates the number of experts; Experts User needs Functional requirements The strength of the correlation between them is scored.

3. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 2, characterized in that: Step S11 specifically includes the following steps: S111. Define the task objectives as sentiment classification and user need type classification, where sentiment classification includes positive, negative and neutral. S112. Construct a corpus of affordable housing furniture demand containing labeled samples, where the labels include sentiment labels and user demand type labels, to obtain a labeled dataset; S113. Divide the labeled dataset into a training set, a validation set, and a test set in a 7:2:1 ratio; S114. BERT-Base-Chinese is used as the pre-trained model, and a parallel branching classifier head for sentiment tendency recognition and user demand type classification, along with a weighted joint loss function, are designed. Iterative training is performed until the target is met. The expression for the weighted joint loss function is as follows: ； in, ； ； In the formula, Indicates the total loss; and These represent the weights for emotion classification and user need type classification, respectively. and These represent the sentiment classification loss and the user demand type classification loss, respectively. Indicates the first The sentiment tendency is labeled as sentiment type. The true label, These represent positive, negative, and neutral, respectively. Indicates the first The sentiment tendency label predicts the sentiment type. The probability of; Indicates the first The sentiment index is labeled as the user's need type. The true label; Indicates the first Sentiment labeling predicts user need type The probability of; The total number of entries indicating sentiment bias; S115. Calculate sentiment score using the SentiStrength algorithm. We obtained an emotional intensity sample: ； in, ； ； ； In the formula, Indicates the score of basic sentiment words, Indicates the first Article 1 of the document on furniture requirements for affordable housing The weight of emotional keywords in the affordable housing furniture sector. Indicates the first Article 1 of the document on furniture requirements for affordable housing The TF-IDF value of each sentiment word. Indicates the first The number of sentiment words in the text of the furniture demand for affordable housing; Indicates the first The number of effective words in a text describing the furniture requirements for affordable housing; This represents the length normalization adjustment factor; Indicates a context-corrected item; Words expressing emotions Frequency of occurrence in a particular emotion category; Words expressing emotions Total frequency of occurrence across all emotion categories; Indicates the first Article 1 of the document on furniture requirements for affordable housing The number of times each emotion word appears; Indicates inverse document frequency, and , This indicates the total number of documents related to the demand for furniture in affordable housing. Indicates the first The number of effective sentiment words in a text describing the demand for furniture in affordable housing; Indicates the first Emotional words in the text regarding furniture requirements for affordable housing and non-emotional words The average attention weight.

4. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 3, characterized in that: Step S2 specifically includes the following steps: S21. Divide the modular units according to the space reuse scenario, match the functional requirements described in step S1, determine the core module type and functional boundaries, and ensure that each module independently implements one basic function and reserves a general interface. S22. Optimize the mortise and tenon structure parameters to meet the lightweight requirements of affordable housing, based on the thickness of the modular panels. and body width According to the constraints Determine the width of the beneficial tenon. and according to the constraints Determine the depth of the concession tenon ;in, Indicates assembly clearance; Indicates the machining safety margin; S23. Based on the material database, filter material combinations to match the functional requirements in step S1; S24. Under the premise of ensuring that the maximum stress of the module material, the allowable stress, and the maximum displacement are not greater than the corresponding set threshold, Autodesk Inventor Nastran is used to perform finite element strength verification on the material combination screened in step S23. The load and frequency of use of the affordable housing furniture are simulated to verify the material combination screened in step S23. If it is qualified, step S25 is executed; otherwise, step S23 is returned. S25. Design a universal interface combining snap-fit ​​and mortise-tenon joints, and determine the compatibility of the universal interface. If the requirements are met, proceed to step S26; otherwise, return to step S23. The general interface matching degree calculation formula is as follows: ； In the formula, Indicates the overlapping area of ​​the general interface size; Indicates the total area of ​​the interface; Indicates the tolerance range; Indicates the total tolerance range; S26. Integrate the structural parameters, material parameters, and interface parameters from steps S22-S25 to obtain a modular unit parameter library containing module ID, function type, and applicable scenarios.

5. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 4, characterized in that: Step S3 specifically includes the following steps: S31. Based on the modular unit parameter library and combined with the affordable housing unit type data, a labeled dataset is constructed, and spatial aspect ratio and family population features are introduced to adapt to different user profiles in S1. S32. Train the CGAN-dynamic rule hybrid model using the labeled dataset, iterating until the generated combined parameter scheme is reached. The fit rate is ≥85%; among them, the generator G of the CGAN-dynamic rule hybrid model is input with random noise. and dynamic condition vector Output combined parameter scheme Its loss function The expression is as follows: ； in, ； In the formula, Indicates random noise and dynamic condition vector Mathematical expectation operation; Indicates the discriminator; Indicates a generator; The weighting coefficients represent the rule loss; Represents the loss of dynamic rules; This indicates that the combined parameter scheme satisfies the constraint rules. The degree; Represents dynamic condition vector The The minimum satisfaction threshold for a constraint rule; The input to discriminator D is a combination parameter scheme. and dynamic condition vector Output a truth score, and its loss function is... The expression is as follows: ； In the formula, Represents the actual combination of parameters scheme and dynamic condition vector Mathematical expectation operation; This indicates that the discriminator operates under a given dynamic condition vector. At that time, for the actual combined parameter scheme The probability of predicting the accuracy of the prediction; S33. Experts review the generated multiple combined parameter schemes. Conduct a scoring and evaluation: ； In the formula, Indicates the scoring result; Indicates dimensional weights. These respectively represent spatial adaptability, module compatibility, and functional satisfaction. Experts Dimension The rating; Indicates the total number of experts; S34. Combining multiple parameter schemes Rating results Sort in descending order and take the first few. A combination of parameter schemes is used as the adaptation scheme.

6. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 5, characterized in that: Step S4 Specifically, the following steps are included: S41. Construct a spatial-intensity-economic-sustainability optimization model with the objective function... The expression is as follows: ； in, ； ； ； ； In the formula, , , and These represent the target weights for space utilization, structural strength safety factor, economy, and sustainability, respectively. ; , , , These respectively represent space utilization rate, strength safety factor, economy, and sustainability; Indicates the volume of the furniture; Indicates spatial volume; Indicates the allowable stress of the furniture material; Indicates the maximum stress on the furniture; Indicates the total cost; This indicates the upper limit of the budget for furniture in affordable housing; Indicates the material degradation cycle; Indicates the service life; Indicates the percentage of recyclable materials; Constraints: ; S42. Optimization using the NSGA-II algorithm: Set the population size to 50, the number of iterations to 100, the crossover probability to 0.8, and the mutation probability to 0.1, with the objective function... For the fitness function, iterate through the output of step S3. Each adaptation scheme is used to output the Pareto optimal solution as a candidate combination parameter scheme. S43. Design Verification Based on AD Axioms: Constructing a Functional Requirements-Design Parameter Mapping Matrix: ； In the formula, , , and These respectively represent spatial adaptability, structural strength, economic cost, and sustainability; , , and These represent module size, material thickness, material cost, and the percentage of recyclable materials, respectively. Determine if the functional requirement-design parameter mapping matrix satisfies that off-diagonal elements ≤ 0.05 to ensure functional decoupling. If so, pass the verification; otherwise, delete the corresponding combined parameter scheme. S44. Selecting the objective function The scheme that maximizes the parameters and satisfies the constraints is taken as the optimal combination parameter scheme.

7. The method for generating modular combination parameters of affordable housing furniture driven by artificial intelligence according to claim 6, characterized in that: Step S5 specifically includes the following steps: S51. Generate a standardized parameter file in JSON format. The standardized parameter file contains the scheme ID, adapted scenario, module ID, number of modules, module location coordinates, structural parameters, material parameters, assembly and production instructions, for use in the production and manufacturing of affordable housing furniture. S52. Collect stress, deformation, and usage frequency at the interface of the furniture module to obtain a usage status dataset; S53. Input the state dataset into the random forest parameter tuning model to obtain the parameter tuning amounts from the modular unit parameter library. : ； In the formula, Indicates the adjustment margin; This represents the optimal combination parameters obtained in step S4; Indicates based on Predicted current combination parameters; This indicates the use of a state dataset; S54, Based on parameter adjustment amount Update the modular unit parameter library in step S2.

Images