Building model family parameter generation method and system based on AI driving

Abstract

The invention relates to a building model family parameter generation method and system based on AI driving, and belongs to the technical field of building information models.The method comprises the steps that a design intention proposed by a user is received, the design intention is analyzed based on a preset AI large model, and all intention points are extracted; based on a preset mapping relationship, mapping a family parameter for each intention point; defining constraint conditions among the family parameters correspondingly mapped by different intention points, and determining a family parameter value range corresponding to each intention point based on a definition result; the definition result meets the condition that the family parameter range corresponding to any intention point does not conflict with other intention points; and generating and outputting a parameter combination with the family parameter value range corresponding to all the intention points. The method and the device have the effects of improving family parameter setting efficiency and reducing parameter setting errors caused by artificial factors.

Description

Technical Field

[0001] The present application relates to the field of building information modeling technology, and in particular to an AI-driven method and system for generating building model family parameters. Background Art

[0002] Currently, the application of Building Information Modeling (BIM) technology in the stages of building design, construction, and operation and maintenance is becoming increasingly widespread. Its core lies in constructing building models through digital means to achieve accurate expression and management of building construction. With the popularization of BIM technology, parametric design of building construction has become the key to improving design efficiency and accuracy. In the BIM model, the family is the basic component unit of the construction model. It is used to describe a collection of building components with similar properties, behaviors, and geometric shapes. Components such as doors, windows, lamps, and walls can all be regarded as a family. Family parameters are specifically used to describe the size, material, and performance (such as thermal insulation, heat insulation, moisture resistance, etc.) properties of the family objects. Users can adjust the family parameter values according to design requirements to achieve the setting of building components in the building model.

[0003] Based on the above, it can be seen that in the existing technology, the family parameter setting of building components needs to rely heavily on manual experience. Designers need to manually adjust the parameters to meet the design intent. This process is not only time-consuming and labor-intensive, but also prone to parameter setting errors due to human factors, which in turn affects the model quality and subsequent applications. Therefore, there is room for improvement. Summary of the Invention

[0004] In order to improve the efficiency of family parameter setting and reduce parameter setting errors caused by human factors, the present application provides an AI-driven building model family parameter generation method and system.

[0005] In a first aspect, the present application provides an AI-driven method for generating parameters of a building model family, comprising: Receive the design intent proposed by the user, analyze the design intent based on the preset AI big model and extract all the intent points; Based on the preset mapping relationship, map family parameters for each intention point; Define constraints between the family parameters mapped to different intent points, and based on the definition results, determine the family parameter value range corresponding to each intent point; and the definition results satisfy: there is no conflict between the family parameter range corresponding to any intent point and other intent points; Generates and outputs parameter combinations with family parameter value ranges corresponding to all intent points.

[0006] By adopting the above technical solution, based on the complex design intentions proposed by the user, a comprehensive analysis is performed through the AI big model to decompose the complex intentions into several intention points, and automatically correspond the relevant family parameters and parameter value ranges to each intention point. In this process, the dependencies and constraints between the parameters are specially taken into account to avoid generating contradictory parameter combinations. Finally, the above solution is combined to replace the manual parameter adjustment method to improve the accuracy and efficiency of parameter determination.

[0007] Optionally, each of the intention points corresponds to a family object, and the intention points include a first intention point for describing the correspondence of a single family, and a second intention point for describing the association relationship between objects of different families; The method of defining the constraints between the family parameters mapped to different intention points, determining the family parameter value range corresponding to each intention point based on the definition results, and generating and outputting a parameter combination with the family parameter value ranges corresponding to all intention points includes: Define the constraints between the family parameters corresponding to different first intention points, and based on the definition results, determine the initial range of family parameter values corresponding to each first intention point; The second intention point is used to modify the initial range of family parameter values of the first intention point related to the second intention point, and the modified initial range of family parameter values is used as the family parameter value range corresponding to each first intention point; wherein, if there is a target first intention point that satisfies: the family parameter corresponding to the target first intention point is included in the family parameter corresponding to the target second intention point, then the target first intention point is considered to be the first intention point related to the target second intention point; For each of the second intention points, generating and storing a collaborative relationship between all first intention points related to the same second intention point; Generate and output a parameter combination with a family parameter value range corresponding to all the first intention points; The method further comprises: Receive a parameter adjustment instruction from a user, adjust the family parameter value of the family parameter contained in the parameter adjustment instruction, and if there is a second family parameter with a collaborative relationship in the parameter adjustment instruction, adaptively adjust the family parameter value with the second family parameter having a collaborative relationship according to the collaborative relationship.

[0008] By adopting the above technical solution, this solution further proposes a parameter combination processing solution for complex design intentions associated with multiple families (such as the size of family A needs to adapt to the interface of family B, or the style of family A needs to be coordinated with family B). The above processing solution realizes dynamic relationship modeling and cross-family collaborative optimization to further optimize the adaptability of this application to complex design intentions.

[0009] Optionally, the method further includes: At every designated period, the intention points corresponding to the design intentions acquired within the designated period are analyzed using a preset learning analysis algorithm, and high-frequency co-occurring intention points are extracted to form composite intention points, wherein the composite intention point is formed by combining more than one high-frequency co-occurring intention points; The design intent is analyzed based on the preset AI big model and all intent points are extracted, including: The design intent is analyzed based on the preset AI large model, and the design intent is preferentially matched with all pre-stored composite intent points, and then other intent points except the composite intent points are extracted from the design intent; and when a composite intent point is obtained by matching, the composite intent point is used as the intent point, and finally all intent points are extracted.

[0010] By adopting the above technical solution, based on all the design intentions proposed by users in the historical period, we can learn and summarize the common intention points proposed by users, and combine them to generate composite intention points, so as to facilitate the improvement of the efficiency of intent parsing and the efficiency of determining the specific family parameter value range.

[0011] Optionally, the family parameter value range corresponding to each intention point includes several family parameter value sub-ranges, and each family parameter value sub-range corresponds to a sensitive interval; wherein the sensitive interval includes the intention point that is in conflict with the family parameter value sub-range to which it belongs; The method further comprises: Regularly update the sensitive interval of the sub-range of family parameter values corresponding to each intention point, and perform sensitivity analysis on the sensitivity period; Perform sensitivity analysis on the sub-range of family parameter values for each intention point, and adjust and update the sub-range of family parameter values and its corresponding sensitive interval based on the sensitivity analysis results; The step of defining the constraint conditions between the family parameters corresponding to different first intention points and determining the initial range of family parameter values corresponding to each first intention point based on the definition results includes: Analyze whether there is a sensitive intention point in the sensitive interval corresponding to each family parameter value sub-range contained in each first intention point, where the sensitive intention point is any first intention point. If so, use the family parameter value sub-range corresponding to the sensitive interval containing the sensitive intention point as the sensitive sub-range; For each first intention point, all other sub-ranges of the family parameter values in the corresponding sub-range of the family parameter value except the sensitive sub-range are taken as a union to obtain the corresponding initial range of the family parameter value.

[0012] By adopting the above technical solution, the family parameter value range contained in the intention point is further stratified, and the sensitive interval is calibrated for each of the several family parameter value sub-ranges obtained by stratification. The sensitive interval contains the intention points that conflict with the family parameter sub-range to which they belong, thereby establishing a correspondence between the intention point and the family parameter sub-range, and using the sensitive interval to help determine the family parameter value range corresponding to each intention point in the subsequent determination, thereby improving the determination efficiency.

[0013] Optionally, the step of defining constraints between family parameters corresponding to different first intention points and determining a preliminary range of family parameter values corresponding to each first intention point based on the definition results may further include: Determine whether there is a first intention point that meets a preset condition. If so, use the first intention point that meets the preset condition as the third intention point, and use the sensitive intention points corresponding to the third intention point to form a conflicting intention point set; wherein the preset condition is that all family parameter value subranges corresponding to the first intention point are sensitive subranges; Obtaining the conflict intensity between each sensitive sub-range corresponding to the third intention point and the corresponding sensitive intention point, as well as the priority and adjustable elasticity of each sensitive intention point, and inputting the obtained content into a preset conflict resolution model. The conflict resolution model then outputs a compromise solution for each sensitive sub-range included in the third intention point, the compromise solution including: adjustments to the sensitive intention points corresponding to the retained sensitive sub-ranges in order to retain the corresponding sensitive sub-ranges; Output the fallback solution corresponding to the third intention point, obtain user feedback results, adjust the sensitive intention points related to the third intention point according to the feedback results, and re-determine the initial range of the family parameter value corresponding to the third intention point based on the adjusted sensitive intention points.

[0014] By adopting the above technical solution, when all family parameter value sub-ranges corresponding to the first intention point are determined to be sensitive sub-ranges, the initial range of the family parameter value finally determined will be 0. At this time, the first intention point is considered to be the third intention point that meets the preset conditions. In order to avoid this situation, that is, in order to retain at least one family parameter value sub-range corresponding to the first intention point, it is necessary to adjust the intention point that conflicts with it. For this purpose, the present application specially sets up a conflict resolution model to implement the above solution to ensure that the user's various intention points are realized to the greatest extent.

[0015] Optionally, the method further includes: Regularly, based on the concession plans determined in historical periods, take the intention point corresponding to each concession plan as the target intention point, and determine and establish the corresponding relationship between the specific content of the target intention point before and after adjustment according to the corresponding concession plan; Whenever the design intent input by the user is received and all the intent points are extracted, it is determined whether the target intent point is included in all the intent points. If so, a suggestion is generated based on the corresponding relationship between the target intent point and the target intent point, and the suggestion is output to the user. The target intention point is updated according to the user's output and the adoption result of the proposed solution; wherein the proposed solution includes the specific content of the target intention point after being adjusted in the historical period.

[0016] By adopting the above technical solution, correction suggestions (i.e., recommended solutions) are provided to users when they propose design intentions, so as to help avoid the occurrence of third intention points in advance and help achieve the efficiency of the entire operation process of obtaining the corresponding parameter combination from the analysis of design intentions.

[0017] Optionally, the method further includes: Periodically generating optimal specific content for describing each target intention point based on the adjusted specific content in the corresponding relationship determined for each target intention point in the historical period, and updating and storing the target intention point and its corresponding optimal specific content; the optimal specific content is used to reduce the probability of the corresponding target intention point becoming a sensitive intention point; The proposed solution includes the optimal specific content corresponding to the target intention point.

[0018] By adopting the above technical solution, the optimal specific content is recommended for the target intent point, the adjustment effect of the specific content of the target intent point is improved, and the probability of the corresponding target intent point becoming a sensitive intent point is helped to be reduced.

[0019] In the second aspect, the present application provides an AI-driven building model family parameter generation system, including: The intent recognition module is used to receive the design intent proposed by the user, analyze the design intent based on the preset AI model, and extract all intent points; A family parameter mapping module is used to map family parameters for each intent point based on a preset mapping relationship; A family parameter value determination module is used to define the constraints between the family parameters mapped to different intent points and, based on the definition results, determine the family parameter value range corresponding to each intent point; and the definition results satisfy the following conditions: the family parameter range corresponding to any intent point does not conflict with that of other intent points; The family parameter combination feedback module is used to generate and output parameter combinations with family parameter value ranges corresponding to all intention points.

[0020] In a third aspect, the present application provides an AI-driven architectural model family parameter generation device, comprising a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute any method described in the first aspect.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and execute any of the methods described in the first aspect.

[0022] In summary, this application has the following beneficial technical effects: This application uses a large AI model to comprehensively analyze the complex design intent proposed by the user, breaking it down into several intent points. It then automatically maps the relevant family parameters and parameter value ranges to each intent point. During this process, the dependencies and constraints between parameters are specifically considered to avoid generating conflicting parameter combinations. Ultimately, the aforementioned solutions are combined to replace manual parameter adjustment, improving the accuracy and efficiency of parameter determination. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flow chart of an AI-driven architectural model family parameter generation method disclosed in an embodiment of the present application.

[0025] Figure 2 This is a structural block diagram of an AI-driven architectural model family parameter generation system disclosed in an embodiment of the present application.

[0026] Explanation of reference numerals: 201, intention recognition module; 202, family parameter mapping module; 203, family parameter value determination module; 204, family parameter combination feedback module. DETAILED DESCRIPTION

[0027] The following is combined with Figure 1-2 This application is described in further detail.

[0028] The present application discloses an AI-driven architectural model family parameter generation method (hereinafter referred to as the family parameter generation method), which aims to identify the user's design intent, match the corresponding family parameters to the user's design intent with the help of an AI big model, and automatically determine and feedback the corresponding family parameter value range, thereby achieving efficient and accurate determination of the family parameters and their corresponding family parameter value ranges. The execution subject of the family parameter generation method is an AI-driven architectural model family parameter generation system (hereinafter referred to as the family parameter generation method), which will be combined with Figure 1 The execution process of the family parameter generation method by the family parameter generation system is specifically explained.

[0029] S101, receiving the design intent proposed by the user, analyzing the design intent based on the preset AI big model and extracting all intent points.

[0030] S102: Mapping family parameters for each intention point based on a preset mapping relationship.

[0031] In practice, users can be architectural designers who access the family parameter generation system via a webpage. The system's access interface includes a pre-set input interface for users to enter design intent. Design intent refers to the specific building components the user wishes to define and the corresponding design requirements, such as "generate an energy-saving exterior window." Input can be done through various methods, including text input, voice input, and sketch / image input.

[0032] The family parameter generation system is used to receive the design intent entered by the user and identify the design intent through a preset AI large model, such as using GPT-4 to parse text content, using speech recognition technology (such as whisper) to convert speech content into text and then parse it, and using a CV model to parse image content. Based on a preset knowledge base that stores several benchmark words, it searches for whether the design intent contains content that is consistent with or contains the benchmark words in the knowledge base. If so, it is used as an intention point. For example, if the benchmark word is sound insulation, and the design intent contains descriptive content such as "sound insulation", "good sound insulation effect", and "sound insulation effect reduced to 35 decibels", the aforementioned content is used as the intention point; then, according to the pre-stored mapping relationship between the relevant benchmark words and family parameters (such as the family parameter corresponding to "good sound insulation" at least includes the decibel parameter of the sound insulation volume), the corresponding family parameters are mapped for each intention point.

[0033] Furthermore, each intention point corresponds to a family object, and a family object refers to the object described by the corresponding intention point. For example, for the design intention of "generating a soundproof window", the family object corresponding to the "soundproof" intention point is "window". In order to be able to determine the one-to-one correspondence between each intention point and the family object from the design intention, especially when the design intention proposed by the user involves multiple family objects, such as "the windows in the living room should be soundproof and the exhaust fan in the kitchen should be powerful", the present application proposes to achieve one-to-one correspondence by limiting the input format of the design intention entered by the user. For example, when the user enters the design intention, he is prompted to manually calibrate all family objects and the intention points corresponding to each family object.

[0034] In other embodiments, the NER model can also be used to extract entities in the design intent (such as "living room window" and "kitchen exhaust fan"). If the user uploads a floor plan or sketch, a CV model (such as Mask R-CNN) can be used to segment the space and mark the entity area to generate an entity-ID: spatial coordinate mapping table; then SpanBERT can be used to process pronoun references (such as "it" and "the former") and bind them to specific entities to achieve the correspondence between the intent point and the family object.

[0035] S103, defining constraints between the family parameters mapped to different intent points, and based on the definition results, determining the family parameter value range corresponding to each intent point; and the definition results satisfy: there is no conflict between the family parameter range corresponding to any intent point and that of other intent points; S104: Generate and output a parameter combination with a family parameter value range corresponding to all intention points.

[0036] During implementation, the family parameter generation system is used to further perform structured analysis on the determined intention points, and preliminarily determine the corresponding family parameter value range for each intention point: specifically, first determine whether the intention point contains the corresponding family parameter, such as the intention point "the sound insulation effect is reduced to 35 decibels", which contains the family parameter "decibel" related to sound insulation. If it exists, the rule template + AI model (such as T5) is used to extract the parameter expression from the specific content of the intention point, such as the corresponding sound insulation volume ≥35dB for the aforementioned intention point "the sound insulation effect is reduced to 35 decibels"; if the intention point does not always have a corresponding family parameter "such as good sound insulation", the corresponding family parameter value range is determined for the intention point based on the pre-stored industry standards (such as residential RW ≥40dB, office RW ≥35dB).

[0037] Next, the family parameter generation system will detect whether there are any contradictions or conflicts between each intention point contained in the design intent and its corresponding preliminarily determined group parameter value range and other intention points. For example, there may be a conflict between intention point A "good window sound insulation" and intention point B "window cost ≤ 800 yuan / square meter". Accordingly, the family parameter generation system pre-stores multiple conflict relationships, which include conflicting intention points (such as sound insulation and cost) and the corresponding relationships between the family parameter value ranges corresponding to the conflicting intention points, such as the cost ranges corresponding to different sound insulation volume ranges, so that the family parameter generation system can realize conflict detection based on the aforementioned conflict relationships.

[0038] If there are any contradictions or conflicts, the family parameter system will determine the constrained intention points and the constrained intention points for the conflicting intention points, and use the constrained intention points as constraints to adjust the preliminarily determined family parameter value range corresponding to the constrained intention points. For example, using the above-mentioned intention point B as a constraint for intention point A, the preliminarily determined family parameter value range corresponding to intention point A is narrowed, and the family parameter value range of intention point A is finally obtained. If there are no contradictions or conflicts, the preliminarily determined family parameter value range for the intention point is used as its final family parameter value range.

[0039] Finally, the family parameters corresponding to all the intention points and their corresponding family parameter value ranges (i.e., parameter combinations) are output. For example, the parameter combinations are displayed on the access interface of the family parameter generation system for the user to know.

[0040] Optionally, when a design intent includes several family objects, and the design intent includes content describing the relationships between different family objects, such as "the long table in the conference room must be consistent in style with the lockers on both sides," the design intent includes two family objects, namely family object E: "long table in the conference room" and family object F: "lockers." The content describing the relationship is "consistent in style." For this complex design intent, the content describing the relationship will also serve as a constraint to further limit the range of family parameter values. To this end, S102 and S103 of this application further include the following steps: S1021, defining constraints between family parameters corresponding to different first intention points, and determining a preliminary range of family parameter values corresponding to each first intention point based on the definition results; S1022: Using the second intention point, the initial range of family parameter values of the first intention point associated with the second intention point is revised, and the revised initial range of family parameter values is used as the family parameter value range corresponding to each first intention point. If a target first intention point exists and satisfies: the family parameter corresponding to the target first intention point is included in the family parameter corresponding to the target second intention point, then the target first intention point is considered to be the first intention point associated with the target second intention point. S1023, for each second intention point, generating and storing a collaborative relationship between all first intention points related to the same second intention point; S1031, generating and outputting a parameter combination with a family parameter value range corresponding to all first intention points; The family parameter generation method also includes the following steps: Receive a parameter adjustment instruction from the user, adjust the family parameter value of the family parameter contained in the parameter adjustment instruction, and if there is a second family parameter with a collaborative relationship in the parameter adjustment instruction, adaptively adjust the family parameter value with the second family parameter in a collaborative relationship according to the collaborative relationship.

[0041] In practice, the group parameter generation system is used to further classify intent points into two categories: first intent points and second intent points, based on whether the number of family objects described is unique. First intent points are used to describe a single family object, meaning the family object corresponding to the first intent point is unique. As can be seen, the intent point described above is a first intent point. Second intent points are used to describe the relationship between two or more family objects, meaning the family objects corresponding to the second intent point are not unique.

[0042] The aforementioned solution can be combined to clarify the description template of the second intention point when the user enters the design intent (such as using the specified connector "&" to connect the family objects described by the second intention point, such as "E&F&G style consistency"), so that the family parameter generation system can determine all family objects corresponding to the second intention point based on the description template, and then parse the content used to describe the association relationship between the corresponding family objects in the second intention according to the aforementioned parsing method (such as "style consistency"), map family parameters and family parameter value ranges, such as family parameters mapped for style consistency are material parameters, and the corresponding family parameter value range can be expressed as "material parameter consistency". First, the family parameter generation system can pre-set association templates for different association relationships and corresponding family parameter value ranges through the knowledge base, such as associating "pipeline connection" with "equal interface diameters of family objects" and "style consistency" with "same material of family objects".

[0043] Therefore, when the design intent proposed by the user only includes the first intention point, the family parameter value range of all first intention points is determined one by one according to the above scheme, and the family parameter value range is further limited based on the conflict detection between the first intention points. If the design intent point also includes the second intention point, the second intention point is used as a constraint condition to limit the family parameter value range of the first intention point related to the second intention point, and if the family object and family parameters corresponding to the first intention point are both included in the second intention point, the first intention point is considered to be related to the second intention point.

[0044] Optionally, the family parameter generation method also includes: At every specified period, the preset learning analysis algorithm is used to analyze the intention points corresponding to the design intentions obtained within the specified period, extract high-frequency co-occurring intention points, and form composite intention points, where a composite intention point is formed by combining more than one high-frequency co-occurring intention points; S101's "analyzing design intent based on a preset AI big model and extracting all intent points" further includes: The design intent is analyzed based on the preset AI large model, and the design intent is matched with all pre-stored composite intent points first, and then other intent points except the composite intent points are extracted from the design intent; and when the composite intent point is matched, the composite intent point is used as the intent point, and finally all intent points are extracted.

[0045] During implementation, the family parameter generation system analyzes the intent points corresponding to the design intents received within each specified period, clustering the intent points corresponding to the same benchmark word into several sets. That is, the intent points contained in each set correspond to the same benchmark word. Then, for all the intent points contained in the sets, it is determined whether there are two or more intent points that are simultaneously included in the same design intent within the specified period, and the number of inclusions exceeds a preset number. If so, they are determined to be high-frequency co-occurring intent points, and the high-frequency shared intent points are combined to generate a composite intent point. For example, if intent points M and N are simultaneously included in design intents X, Y, and Z, then intent points M and N are considered high-frequency co-occurring intent points, and are combined to generate a composite intent point. The corresponding benchmark word is generated for the composite intent point and stored.

[0046] When the design intent sent by the user is received subsequently, the benchmark words corresponding to all stored compound intent points will be compared one by one with the specific content of the design intent to determine whether the design intent contains the content of the benchmark words corresponding to the compound intent point. If so, the match is successful, and the benchmark word of the compound intent point is used as an independent intent point extracted from the current design intent. After completing the comparison of all compound intent points, other intent points except the compound intent points will be further extracted from the design intent to improve the efficiency of extracting intent points.

[0047] In other embodiments, if the high-frequency co-occurring intention points that combine to form a composite intention point have contradicted and conflicted with each other in the historical period, then the family parameter value range finally determined for the aforementioned high-frequency co-occurring intention points in the historical period will be used as the family parameter value range corresponding to the composite intention point.

[0048] Optionally, the family parameter value range corresponding to each intention point includes several family parameter value sub-ranges, and each family parameter value sub-range corresponds to a sensitive interval; wherein the sensitive interval includes the intention point that conflicts with the family parameter value sub-range to which it belongs; The family parameter generation method also includes the following steps: Regularly update the sensitive interval of the sub-range of family parameter values corresponding to each intention point, and perform sensitivity analysis on the sensitivity period; Perform sensitivity analysis on the sub-range of family parameter values for each intention point, and adjust and update the sub-range of family parameter values and its corresponding sensitive interval based on the sensitivity analysis results; S1021 further includes the following sub-steps: Analyze whether there is a sensitive intention point in the sensitive interval corresponding to each family parameter value sub-range contained in each first intention point. The sensitive intention point is any first intention point. If it exists, the family parameter value sub-range corresponding to the sensitive interval containing the sensitive intention point is used as the sensitive sub-range; For each first intention point, the corresponding family parameter value sub-ranges except the sensitive sub-range are combined to obtain the corresponding family parameter value initial range; Determine whether there is a first intention point that meets the preset conditions. If so, use the first intention point that meets the preset conditions as the third intention point, and use the sensitive intention points corresponding to the third intention point to form a conflicting intention point set. The preset condition is that all family parameter value subranges corresponding to the first intention point are sensitive subranges. Obtain the conflict intensity between each sensitive sub-range corresponding to the third intention point and the corresponding sensitive intention point, as well as the priority and adjustable elasticity of each sensitive intention point. Input the obtained content into a preset conflict resolution model, and output a concession plan for each sensitive sub-range included in the third intention point through the conflict resolution model. The concession plan includes: in order to retain the corresponding sensitive sub-range, the adjustment content required for the sensitive intention point corresponding to the retained sensitive sub-range; Output the concession plan corresponding to the third intention point, obtain user feedback results, adjust the sensitive intention points related to the third intention point according to the feedback results, and re-determine the initial range of the family parameter value corresponding to the third intention point based on the adjusted sensitive intention points.

[0049] In implementation, the family parameter value range corresponding to the intention point is divided based on whether a conflict has occurred and the object where the conflict has occurred (i.e., the intention point) to form several family parameter value sub-ranges, and a sensitive interval is set for each family parameter value sub-range. The intention point is stored in the sensitive interval, and the intention point is a constrained intention point that has had a conflict with the intention point (hereinafter referred to as the constrained intention point) belonging to the corresponding family parameter value sub-range (hereinafter referred to as the constrained sub-range) in the historical period, and the constrained sub-range cannot be selected due to the limitation of the constrained intention point.

[0050] For example, the family parameter corresponding to the intention point J is "window width", and the corresponding family parameter value range is [800, 2500], which is divided into the following three family parameter value sub-ranges: J1 [800, 1200], J2 [12, 1800], and J3 [1800, 2500]. The sensitive interval of J1 is 0, that is, there is no intention point that conflicts with it; the sensitive interval corresponding to J2 is ("structural lightweight"); and the sensitive interval corresponding to J3 is ("structural lightweight", "cost control"). When the two intention points "appropriate window width" and "structural lightweight" appear simultaneously in a certain design intention, then when determining the family parameter value range for "appropriate window width", J2 and J3 will be affected by the constraint of "structural lightweight" and will not be selected. That is, J2 and J3 will not exist in the family parameter value range finally determined by the intention point "appropriate width". Therefore, the sensitive interval can be used to determine whether there are any conflicts between all the intention points in the design intent, and the sensitive interval can be used to help ultimately determine the family parameter value range of the intention point.

[0051] In addition, the above further mentions that the first intention point that meets the preset conditions is used as the third intention point, that is, all family parameter value sub-ranges contained in the third intention point are constrained by other first intention points (hereinafter referred to as conflicting intention points) contained in the design intent and cannot be selected. In this case, the present application proposes to use a preset conflict resolution model to correct the conflicting intention point so that the corrected conflicting intention point does not exist in the sensitive interval of the family parameter value sub-range corresponding to the third intention point. In other words, the corrected conflicting intention point will not constrain the family parameter value sub-range contained in the third intention point, so that there is a family parameter value sub-range that can be selected in the family parameter value sub-range contained in the third intention point.

[0052] The above amendments are as follows: After all the intention points are extracted from the design intent, the family parameter generation system is used to feed back all the intention points to the user so that the user can set the priority values of all the intention points, and the priority values of all the intention points are combined to be 10. Then, the family parameter generation system is used to determine the adjustable elasticity corresponding to each intention point according to the preset correspondence between the relevant priorities and the adjustable elasticity, and according to the priority values of the intention points. The adjustable elasticity is specifically used to describe the adjustable range of the family parameter value of the corresponding intention point. For example, the adjustable elasticity of the "cost" parameter can be "acceptable increase of 10%". Conflict intensity reflects the number of sensitive intent points corresponding to each family parameter value subrange. A sensitive intent point is the first intent point in the design intent that constrains the family parameter value subrange. A greater number of sensitive intent points indicates a higher conflict intensity for the corresponding family parameter value subrange. This also indicates that to preserve the corresponding family parameter value subrange, all corresponding sensitive intent points must be corrected.

[0053] In this regard, the family parameter generation method is used to select the family parameter value sub-range with the corresponding conflict intensity lower than the preset impact intensity from all the family parameter value sub-ranges corresponding to the third intention point as the spare reserved sub-range, and then perform weighted summation on the priority and adjustable elasticity of the sensitive intention point corresponding to each spare reserved sub-range to obtain the comprehensive score of each spare sub-range, and select the spare reserved sub-range with the lowest comprehensive priority as the final reserved sub-range, and determine a concession plan for each sensitive sub-range corresponding to the third intention point, but calibrate the recommendation index for the concession plan of each sensitive sub-range separately. The high or low recommendation index is used to characterize the degree of recommendation of the corresponding concession plan by the family parameter generation system, and the recommendation index of the final reserved sub-range is > the spare reserved sub-range > the non-spare reserved sub-range.

[0054] The specific content of the concession plan is the modified sensitive intention point. The modification method is to expand the family parameter value range of the sensitive intention point according to the elasticity rule. For example, the preset adjustable elasticity of the "cost" family parameter (such as an acceptable increase of 10%) is used to expand the value range of the "cost" family parameter.

[0055] The compromise plan is then fed back to the user so that the user can select the final compromise plan and execute the corresponding compromise plan to correct the corresponding sensitive intention point, and then re-determine the corresponding family parameter value range based on the corrected intention point.

[0056] Optionally, the family parameter generation method further includes the following steps: Regularly, based on the concession plans determined in historical periods, take the intention point corresponding to each concession plan as the target intention point, and determine and establish the corresponding relationship between the specific content of the target intention point before and after adjustment according to the corresponding concession plan; Whenever the design intent input by the user is received and all the intent points are extracted, it is determined whether the target intent point is included in all the intent points. If so, a suggestion is generated based on the corresponding relationship between the target intent point and the target intent point, and the suggestion is output to the user. Updating the target intention point based on the user's output and the adoption result of the proposed solution; wherein the proposed solution includes the specific content of the target intention point after adjustment in the historical period; Periodically generating optimal specific content for describing each target intention point based on the adjusted specific content in the corresponding relationship determined for each target intention point in the historical period, and updating and storing the target intention point and its corresponding optimal specific content; the optimal specific content is used to reduce the probability of the corresponding target intention point becoming a sensitive intention point; The proposed solution includes the optimal specific content corresponding to the target intention point.

[0057] During implementation, whenever a concession plan is generated, the intention point (i.e., the target intention point) corresponding to each generated concession plan and the specific content of the target intention point before and after the execution of the concession plan are recorded, and a correspondence between the specific contents is established, and the correspondence is further associated with the benchmark word corresponding to the target intention point.

[0058] Regularly analyze all the corresponding relationships determined in the historical period: all the corresponding relationships corresponding to the same benchmark word, and the specific content after the implementation of the concession plan (hereinafter referred to as the specific content to be evaluated) to select the best specific content. The optimal specific content is selected as follows: according to the family parameter value range corresponding to the specific content to be estimated in the historical period, the maximum endpoint value is selected from the two endpoint values of the family parameter value range corresponding to all the specific contents to be estimated to form the optimal family parameter value range, and then the optimal family parameter value range is described to obtain the optimal specific content, so as to reduce the probability that the target intention point described in the optimal specific content becomes a sensitive intention point.

[0059] The present application also discloses an AI-driven building model family parameter generation system. Figure 2 : Intent recognition module 201, used to receive the design intent proposed by the user, analyze the design intent based on the preset AI big model and extract all intent points; A family parameter mapping module 202 is used to map family parameters for each intention point based on a preset mapping relationship; The family parameter value determination module 203 is used to define the constraints between the family parameters mapped to different intent points. Based on the definition results, the family parameter value range corresponding to each intent point is determined. The definition results satisfy the following conditions: the family parameter range corresponding to any intent point does not conflict with that of other intent points. The family parameter combination feedback module 204 is configured to generate and output a parameter combination with family parameter value ranges corresponding to all intention points.

[0060] Optionally, the family parameter combination feedback module 204 is used to define constraint conditions between family parameters corresponding to different first intention points, and based on the definition results, determine an initial range of family parameter values corresponding to each first intention point; The second intention point is used to correct the initial range of family parameter values of the first intention point related to the second intention point, and the corrected initial range of family parameter values is used as the family parameter value range corresponding to each first intention point; wherein, if there is a target first intention point, it satisfies: the family parameter corresponding to the target first intention point is included in the family parameter corresponding to the target second intention point, then the target first intention point is considered to be the first intention point related to the target second intention point; for each second intention point, the collaborative relationship between all first intention points related to the same second intention point is generated and stored; and a parameter combination with the family parameter value ranges corresponding to all first intention points is generated and output.

[0061] It also includes a parameter range adjustment module, which is used to receive the user's parameter adjustment instructions, adjust the family parameter values of the family parameters contained in the parameter adjustment instructions, and if there is a second family parameter with a collaborative relationship in the parameter adjustment instruction, adaptively adjust the family parameter values with a collaborative relationship with the second family parameter according to the collaborative relationship.

[0062] Optionally, a composite intention point generation module is further included, which is used to analyze the intention points corresponding to the design intentions obtained within the specified period through a preset learning analysis algorithm at each specified period, extract high-frequency co-occurring intention points, and form composite intention points, wherein the composite intention point is formed by combining more than one high-frequency co-occurring intention points; The intent recognition module 201 is also used to analyze the design intent based on the preset AI large model, prioritize matching the design intent with all pre-stored composite intent points, and then extract other intent points except the composite intent points from the design intent; and when the composite intent point is matched, the composite intent point is used as the intent point, and finally all intent points are extracted.

[0063] Optionally, it also includes a parameter range division module, which is used to regularly update the sensitive interval of the family parameter value sub-range corresponding to each intention point, and perform sensitivity analysis on the sensitivity period; perform sensitivity analysis on the family parameter value sub-range of each intention point, and adjust and update the family parameter value sub-range and its corresponding sensitive interval based on the sensitivity analysis results.

[0064] The family parameter combination feedback module 204 is further used to analyze whether there is a sensitive intention point in the sensitive interval corresponding to each family parameter value sub-range contained in each first intention point. The sensitive intention point is any first intention point. If it exists, the family parameter value sub-range corresponding to the sensitive interval containing the sensitive intention point is used as the sensitive sub-range; for each first intention point, the union of all other family parameter value sub-ranges in the corresponding family parameter value sub-range except the sensitive sub-range is taken to obtain the corresponding initial range of family parameter values.

[0065] Optionally, the family parameter combination feedback module 204 is further configured to determine whether there is a first intention point that meets a preset condition. If so, the first intention point that meets the preset condition is used as the third intention point, and the sensitive intention points corresponding to the third intention point constitute a conflict intention point set. The preset condition is that all family parameter value subranges corresponding to the first intention point are sensitive subranges. The conflict intensity between each sensitive subrange corresponding to the third intention point and the corresponding sensitive intention point, as well as the priority and adjustable elasticity of each sensitive intention point, are obtained, and the obtained content is input into a preset conflict resolution model. The conflict resolution model outputs a concession plan for each sensitive subrange included in the third intention point. The concession plan includes: in order to retain the corresponding sensitive subrange, the adjustment content that needs to be made to the sensitive intention points corresponding to the retained sensitive subrange; It is also used to output the concession plan corresponding to the third intention point and obtain user feedback results, adjust the sensitive intention points related to the third intention point according to the feedback results, and re-determine the initial range of the family parameter value corresponding to the third intention point based on the adjusted sensitive intention points.

[0066] Optionally, it also includes a conflict resolution module, which is used to regularly use the intention point corresponding to each concession plan as the target intention point based on the concession plan determined in the historical period, determine and establish the corresponding relationship between the specific content of the target intention point before and after adjustment according to the corresponding concession plan; whenever the design intention input by the user is received and all intention points are extracted, determine whether the target intention point is included in all intention points. If so, generate a recommended plan based on the corresponding relationship corresponding to the target intention point, and output the recommended plan to the user; update the target intention point based on the user's output and the adoption result of the recommended plan; wherein the recommended plan includes the specific content of the target intention point after adjustment in the historical period.

[0067] Optionally, it also includes an intention point content optimization module, which is used to regularly generate the optimal specific content for describing the corresponding target intention point for each target intention point based on the adjusted specific content in the corresponding relationship determined in the historical period, and update the stored target intention points and their corresponding optimal specific content; the optimal specific content is used to reduce the probability of the corresponding target intention point becoming a sensitive intention point; the recommended plan includes the optimal specific content corresponding to the target intention point.

[0068] An embodiment of the present application also discloses an AI-driven building model family parameter generation device, which includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and execute the above-mentioned AI-driven building model family parameter generation method.

[0069] An embodiment of the present application also discloses a computer-readable storage medium, which stores a computer program that can be loaded by a processor and execute the above-mentioned AI-driven architectural model family parameter generation method. The computer-readable storage medium includes, for example: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0070] It should be noted that, in this document, relational terms such as first and second, etc. are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0071] The above embodiments are intended only to illustrate the technical solutions of this application and are not intended to limit the scope of protection of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on these embodiments, all other embodiments obtained by persons of ordinary skill in the art without inventive effort are also within the scope of protection to be protected by this application.

Claims

1. A method for generating architectural model family parameters based on AI, characterized in that: include: Receive the design intent proposed by the user, analyze the design intent based on the preset AI big model and extract all the intent points; Based on the preset mapping relationship, map family parameters for each intention point; Define constraints between the family parameters mapped to different intent points, and based on the definition results, determine the family parameter value range corresponding to each intent point; and the definition results satisfy: there is no conflict between the family parameter range corresponding to any intent point and other intent points; Generates and outputs parameter combinations with family parameter value ranges corresponding to all intent points.

2. The AI-driven building model family parameter generation method according to claim 1 is characterized in that: Each of the intention points corresponds to a family object, and the intention points include a first intention point for describing the correspondence of a single family, and a second intention point for describing the association relationship between objects of different families; The method of defining the constraints between the family parameters mapped to different intention points, determining the family parameter value range corresponding to each intention point based on the definition results, and generating and outputting a parameter combination with the family parameter value ranges corresponding to all intention points includes: Define the constraints between the family parameters corresponding to different first intention points, and based on the definition results, determine the initial range of family parameter values corresponding to each first intention point; The second intention point is used to modify the initial range of family parameter values of the first intention point related to the second intention point, and the modified initial range of family parameter values is used as the family parameter value range corresponding to each first intention point; wherein, if there is a target first intention point that satisfies: the family parameter corresponding to the target first intention point is included in the family parameter corresponding to the target second intention point, then the target first intention point is considered to be the first intention point related to the target second intention point; For each of the second intention points, generating and storing a collaborative relationship between all first intention points related to the same second intention point; Generate and output a parameter combination with a family parameter value range corresponding to all the first intention points; The method further comprises: Receive a parameter adjustment instruction from a user, adjust the family parameter value of the family parameter contained in the parameter adjustment instruction, and if there is a second family parameter with a collaborative relationship in the parameter adjustment instruction, adaptively adjust the family parameter value with the second family parameter having a collaborative relationship according to the collaborative relationship.

3. The AI-driven building model family parameter generation method according to claim 1 is characterized in that: The method further comprises: At every designated period, the intention points corresponding to the design intentions acquired within the designated period are analyzed using a preset learning analysis algorithm, and high-frequency co-occurring intention points are extracted to form composite intention points, wherein the composite intention point is formed by combining more than one high-frequency co-occurring intention points; The design intent is analyzed based on the preset AI big model and all intent points are extracted, including: The design intent is analyzed based on the preset AI large model, and the design intent is preferentially matched with all pre-stored composite intent points, and then other intent points except the composite intent points are extracted from the design intent; and when a composite intent point is obtained by matching, the composite intent point is used as the intent point, and finally all intent points are extracted.

4. The AI-driven building model family parameter generation method according to claim 2 is characterized in that: The family parameter value range corresponding to each intention point includes several family parameter value sub-ranges, and each family parameter value sub-range corresponds to a sensitive interval; wherein the sensitive interval includes the intention point that conflicts with the family parameter value sub-range to which it belongs; The method further comprises: Regularly update the sensitive interval of the sub-range of family parameter values corresponding to each intention point, and perform sensitivity analysis on the sensitivity period; Perform sensitivity analysis on the sub-range of family parameter values for each intention point, and adjust and update the sub-range of family parameter values and its corresponding sensitive interval based on the sensitivity analysis results; The step of defining the constraint conditions between the family parameters corresponding to different first intention points and determining the initial range of family parameter values corresponding to each first intention point based on the definition results includes: Analyze whether there is a sensitive intention point in the sensitive interval corresponding to each family parameter value sub-range contained in each first intention point, where the sensitive intention point is any first intention point. If so, use the family parameter value sub-range corresponding to the sensitive interval containing the sensitive intention point as the sensitive sub-range; For each first intention point, all other sub-ranges of the family parameter values in the corresponding sub-range of the family parameter value except the sensitive sub-range are taken as a union to obtain the corresponding initial range of the family parameter value.

5. The AI-driven building model family parameter generation method according to claim 4 is characterized in that: The step of defining the constraint conditions between the family parameters corresponding to different first intention points and determining the initial range of family parameter values corresponding to each first intention point based on the definition results may further include: Determine whether there is a first intention point that meets a preset condition. If so, use the first intention point that meets the preset condition as the third intention point, and use the sensitive intention points corresponding to the third intention point to form a conflicting intention point set; wherein the preset condition is that all family parameter value subranges corresponding to the first intention point are sensitive subranges; Obtaining the conflict intensity between each sensitive sub-range corresponding to the third intention point and the corresponding sensitive intention point, as well as the priority and adjustable elasticity of each sensitive intention point, and inputting the obtained content into a preset conflict resolution model. The conflict resolution model then outputs a compromise solution for each sensitive sub-range included in the third intention point, the compromise solution including: adjustments to the sensitive intention points corresponding to the retained sensitive sub-ranges in order to retain the corresponding sensitive sub-ranges; Output the fallback solution corresponding to the third intention point, obtain user feedback results, adjust the sensitive intention points related to the third intention point according to the feedback results, and re-determine the initial range of the family parameter value corresponding to the third intention point based on the adjusted sensitive intention points.

6. The AI-driven building model family parameter generation method according to claim 5 is characterized in that: The method further comprises: Regularly, based on the concession plans determined in historical periods, take the intention point corresponding to each concession plan as the target intention point, and determine and establish the corresponding relationship between the specific content of the target intention point before and after adjustment according to the corresponding concession plan; Whenever the design intent input by the user is received and all the intent points are extracted, it is determined whether the target intent point is included in all the intent points. If so, a suggestion is generated based on the corresponding relationship between the target intent point and the target intent point, and the suggestion is output to the user. The target intention point is updated according to the user's output and the adoption result of the proposed solution; wherein the proposed solution includes the specific content of the target intention point after being adjusted in the historical period.

7. The AI-driven building model family parameter generation method according to claim 6 is characterized in that: The method further comprises: Periodically generating optimal specific content for describing each target intention point based on the adjusted specific content in the corresponding relationship determined for each target intention point in the historical period, and updating and storing the target intention point and its corresponding optimal specific content; the optimal specific content is used to reduce the probability of the corresponding target intention point becoming a sensitive intention point; The proposed solution includes the optimal specific content corresponding to the target intention point.

8. An AI-driven architectural model family parameter generation system, characterized in that: include, An intention recognition module (201) is used to receive a design intention proposed by a user, analyze the design intention based on a preset AI big model, and extract all intention points; A family parameter mapping module (202) is used to map family parameters for each intention point based on a preset mapping relationship; A family parameter value determination module (203) is used to define the constraint conditions between the family parameters mapped to different intention points, and based on the definition results, determine the family parameter value range corresponding to each intention point; and the definition results satisfy: there is no conflict between the family parameter range corresponding to any intention point and other intention points; The family parameter combination feedback module (204) is used to generate and output a parameter combination with family parameter value ranges corresponding to all intention points.

9. An AI-driven building model family parameter generation device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that A computer program is stored which can be loaded by a processor and execute the method according to any one of claims 1 to 7.

Images