AI-based building design optimization system and method

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an AI-based building design optimization system and method, and the system comprises a visual parameter collection module, an interface adjustment judgment module, a behavior path construction module, a space arrangement replacement module and a sequence adaptation updating module. According to the method, the functional area boundary sight line passing length and the angle deviation value are extracted, the visual difference sequence is constructed to determine the spatial visual connection difference, sight line obstacle and non-uniform perception areas are identified, and the interface adjustment list is constructed in combination with the functional attributes and the guide paths, so that the structure adjustment pertinence and efficiency are improved; the method comprises the following steps of: establishing a weight structure chart through a ratio of behavior calling frequency to path section change, reflecting spatial behavior association, screening an arrangement optimization scheme by combining a connection relationship and a path expectation difference value, enhancing logic consistency and behavior matching degree, identifying a behavior sequence deviation, updating an operation path chart, and realizing functional area sequence dynamic adaptation and behavior logic reconstruction. And the space responsiveness and the use rationality are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an AI-based architectural design optimization system and method. Background Art

[0002] The field of artificial intelligence technology includes the research and application of theories, methods and technologies related to computer science and artificial intelligence, mainly covering machine learning, deep learning, natural language processing, computer vision and other aspects. The core content of artificial intelligence is to enable computer systems to automatically perform tasks and support decisions by simulating and extending human intelligence. Its application range is wide, including speech recognition, image processing, robotics, intelligent recommendations, autonomous driving, etc. This technical field is committed to improving the automation, accuracy and efficiency of the system in specific tasks through the innovation of artificial intelligence algorithms and models.

[0003] Among them, the architectural design optimization system refers to the automatic generation and optimization of architectural design solutions based on artificial intelligence technology. The technical matters of this patent subject mainly cover data analysis and processing in the architectural design process, intelligent reasoning and optimization decision-making of design parameters, etc. Specifically, machine learning algorithms are used to analyze design data, and artificial intelligence models such as deep neural networks are used to predict and optimize design solutions. The model is iterated and adjusted based on feedback information to improve design quality and efficiency. This system promotes the automation and intelligence of architectural design by combining data input with intelligent reasoning, solving the complexity and efficiency problems in the traditional manual design process.

[0004] Existing technologies rely on static modeling for visual element analysis and lack systematic identification of differences in visual accessibility, resulting in errors in spatial guidance perception. In the process of interface adjustment, the functional attributes are not effectively associated with the path sequence, and interface modifications often fall into structural isolation and lack overall logical support. The construction of behavioral paths lacks a comprehensive weight analysis of call frequency and path continuity, and the path setting does not match the actual usage logic. In the process of spatial layout adjustment, the connection relationship is not combined to maintain the expected difference control with the path, and the replacement plan is easy to deviate from usage habits. The adjustment of the functional area sorting under situational changes lacks a weight deviation judgment mechanism, resulting in insufficient responsiveness of the behavioral sequence and poor spatial adaptability. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an AI-based architectural design optimization system and method.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: The AI-based architectural design optimization system includes: The visual parameter acquisition module obtains the boundary surfaces of all functional areas in the building's indoor layout, collects the sight line length and angle offset values ​​in the forward and reverse directions, calculates the average and performs difference normalization and superposition to obtain a sequence of visual difference values ​​between functional areas; The interface adjustment determination module calls the items exceeding the threshold value in the sequence of visual difference values ​​between the functional areas, determines the interface closure, light transmittance and the number of openable components, marks the material replacement items and the structural opening items, associates the sorting position of the interface in the guide path with the functional area usage attributes, and establishes an operation sequence for all the suggested items to obtain an interface reset operation list; The behavior path construction module extracts the calling frequency of the function area based on the interface reset operation list, calculates the frequency difference and change ratio of the path segments, marks the combination exceeding the path coherence threshold, and obtains the path node weight structure diagram; The spatial arrangement replacement module calls the functional area combination in the path node weight structure diagram, calculates the difference between the spatial grid distance and the path expectation, screens the replacement schemes whose connection relationship has not been reduced, and obtains a feasible arrangement replacement sequence.

[0007] As a further solution of the present invention, the sequence of visual difference values ​​between functional intervals specifically includes line of sight accessibility difference, angle offset difference, and difference normalization index; the interface reset operation list includes material replacement items, structural opening items, and operation sorting labels; the path node weight structure diagram specifically refers to path node call frequency, path continuity weight, and node combination mark; the feasibility arrangement alternative sequence includes spatial grid distance evaluation value, path expected deviation value, and connection relationship retention label.

[0008] As a further solution of the present invention, the visual parameter acquisition module includes: The boundary extraction submodule collects the boundary surfaces of all functional areas in the building interior layout, converts the outer contour lines of the functional areas in the two-dimensional plane image into boundary vertex coordinate data, uses the contour boundary closure condition to perform integrity verification on the collected boundaries, obtains the boundary line segment information of all functional areas according to the connection relationship between the spatial contours of the differentiated functional areas, and generates the boundary line segment data of the functional areas; The sight parameter calculation submodule calls the boundary line segment data of the functional area, sets a plurality of equally spaced monitoring points in the positive and negative directions of the boundary line segment, projects a sight line to each monitoring point, collects the sight line length without penetrating other boundaries, and records the angle offset value relative to the boundary normal, calculates the average length and angle offset mean of all positive sight lines, and the average length and angle offset mean of all reverse sight lines, and obtains the mean value of the two-way sight line parameter; The visual difference generation submodule calls the mean of the two-way sight line parameters, refers to the influence of the difference of sight line parameters between multi-functional areas on spatial perception, and introduces the sight line difference index Using the formula: ; Calculate the visual difference coefficient of the multifunctional area to form a sequence of visual difference values ​​of the functional area; in, represents the visual difference index, Representative The average length of the positive sight line of each functional area, Representative The average length of reverse sight lines in each functional area, Representative The threshold constant used in the calculation of sight difference between functional areas, is the adjustment constant used to maintain the denominator in a non-zero state, Representative The weight factor of the sight line parameters of each functional area in the difference calculation, Represents the total number of functional areas.

[0009] As a further solution of the present invention, the interface adjustment determination module includes: The difference item screening submodule calls a preset visual difference threshold based on the visual difference value sequence of the functional area, compares and screens multiple items in the difference value sequence one by one, obtains the difference items in the difference value sequence that exceed the visual difference threshold, determines the spatial distribution block where the difference items are located in the functional area, calculates the number of difference items and their corresponding interface unit numbers, and generates an over-threshold difference item set; The interface characteristic judgment submodule calls the above-threshold difference item set to obtain the closure parameter, light transmittance parameter and the number of openable elements of the corresponding interface unit, and judges whether the interface is closed according to the closure parameter, judges whether it has a light-transmitting structure by using the light transmittance parameter, judges the openable state by the number of openable elements, calculates the interface transparency characteristic value and performs weighted integration with the frequency of use coefficient, using the formula: ; Calculate the interface transparency difference value, call the transparency difference threshold to determine whether it is an item that needs to be adjusted, and obtain the interface attribute item that needs to be adjusted; in, Indicates The difference in the permeability of the interface, Indicates The number of different items corresponding to each interface, Indicates In the interface The closure parameter value of the term, represents the corresponding transmittance parameter, For the The light intensity coverage factor of the term, For the In the interface The number of components that can be opened corresponding to the difference item, For the The frequency of use of each interface, For the The activity value of the functional area where the interface is located, The sorting number of this interface in the navigation path. Sort and number the benchmark guides for the functional areas; The reset operation list generation submodule calls the attribute items that need to be adjusted in the interface, obtains the navigation path sorting number and function area usage attributes of the corresponding interface, builds an operation priority list based on the sorting number and usage attributes, arranges the material items and structural opening items that need to be replaced in sequence, establishes the interface replacement operation sequence, and obtains the interface reset operation list.

[0010] As a further solution of the present invention, the behavior path construction module includes: The interface operation reset submodule extracts the function area name, trigger time and usage frequency corresponding to each operation in the list based on the interface reset operation list, classifies them by function area, obtains the calling frequency of the multi-function area in the complete interaction process, summarizes and sorts the calling frequency, and generates the function area calling frequency value; The path segment difference calculation submodule calls the function area call frequency value to obtain the call frequency difference between consecutive path segments using the formula: ; Calculate the path segment stability score, and determine the continuity of multiple path segments based on the score to obtain the path segment frequency change rate; in, represents the stability score of the path segment, k represents the kth path segment, N represents the total number of path segments to be evaluated, Represents the calling frequency of the starting point function area of ​​the kth path segment, Represents the calling frequency of the functional area at the end point of the k-th path segment, Represents the absolute value of the difference in call frequency between the starting point and the end point of the kth path segment, represents the decay coefficient of the interval time between path segment operations, represents the operation interval time corresponding to the kth path segment, shows the attenuation factor based on the operation interval time, represents the weighting factor of the kth path segment, which is used to adjust the weight of the path segment in the overall stability score. It represents the weighted impact value of the calling frequency of the endpoint of the k-th path segment. represents adjustment parameters; The coherence structure marking submodule calls the path segment frequency change rate, and judges each item against the path coherence threshold according to the frequency change ratio in the path segment combination, and marks the path segment combination whose comparison value is greater than the path coherence threshold, maps the marked path segments to the path node structure diagram, assigns corresponding weights to multiple nodes, and generates a path node weight structure diagram.

[0011] As a further solution of the present invention, the spatial arrangement replacement module includes: The path weight extraction submodule extracts the connection edge lengths, node numbers and corresponding functional area combinations between multiple path nodes based on the path relationship marked in the path node weight structure diagram, constructs a path segment matrix according to the node connection sequence, and reassigns the weight values ​​according to the physical position parameters and combination relationship parameters between the nodes to generate a path segment weight matrix; The expected difference calculation submodule calls the path segment weight matrix and calculates the functional adaptability difference between nodes according to the node function attributes and the actual distance, using the formula: ; Calculate the multi-node functional adaptability deviation and establish the expected deviation value of the functional path; in, Represents the functional adaptability deviation value between path nodes, Represents the path The actual physical distance between the nodes connecting the segments, Represents the path The functional difference level value between the functional area combinations corresponding to the segment nodes, Represents the path The structural strength weight value of the functional area combination in the node connected by the segment, Represents the path The participation level of the segment node in the functional connectivity network, Indicates the upper limit of the count of all segments in the path. Represents the total number of nodes involved in weight evaluation in the path segment; The alternative sequence screening submodule performs connection number statistics, structure preservation judgment and offset value sorting on all replaceable path combinations according to the screening conditions that the expected deviation value of the functional path and the node connection relationship in the path segment have not been reduced, screens out path combinations with reduced number of connection edges, sorts and selects the structural alternative sequences with the highest offset value, and generates a feasible arrangement alternative sequence.

[0012] As a further solution of the present invention, the system further includes: The sequence adaptation update module arranges the functional area sequence adjustment scheme in the alternative sequence according to the feasibility, extracts the functional area behavior sequence weight distribution table corresponding to the current context label in the marked exhibition and teaching scenes, calculates the deviation value between the actual arrangement sequence and the weight sorting, and if the deviation value is greater than the functional area sorting offset threshold, marks the functional area set that needs to be updated, generates a sequence update operation queue corresponding to the current context label, and obtains the functional area sorting update path diagram; The functional area sorting update path diagram specifically includes a sorting offset mark, an updated functional area set, and a context adaptation order.

[0013] As a further solution of the present invention, the sequence adaptation update module includes: The context label extraction submodule obtains the marked display and teaching scene data according to the functional area sequence adjustment plan in the feasibility arrangement alternative sequence, calls the functional area behavior sequence weight distribution table corresponding to the context label, extracts the matching items between the context label and the functional area behavior, determines the weight arrangement order according to the matching relationship, and generates the context label weight sequence table; The behavior sequence deviation calculation submodule calls the current actual arrangement sequence according to the context label weight sequence table, calculates the difference between the actual sequence and the weight sequence, and uses the formula: ; Calculate the behavior order deviation index, determine whether it is greater than the benchmark according to the function area sorting deviation threshold, select the function areas that need to be adjusted, and generate a set of function areas to be updated; in, represents the behavior order deviation index, Representative The position of each functional area in the actual layout, Representative The position of each functional area in the weight sorting, For the The visibility coefficient of each functional area, Indicates the total number of functional areas; The sequential update path generation submodule extracts the arrangement information of the set of functional areas to be updated in the current order and their target positions in the weight sorting according to the set, constructs a sorting alternative path according to the difference sequence adjustment method, forms an adjustment action queue and sorts it, and screens feasible operation paths in combination with the actual site feasibility comparison structure required for the sequential adjustment to obtain a functional area sorting update path diagram.

[0014] The AI-based architectural design optimization method is performed based on the above-mentioned AI-based architectural design optimization system, and includes the following steps: S1: Obtain the boundary surfaces of all functional areas in the building's indoor layout, collect the sight line length and angle offset values, calculate the sight line difference and angle difference between each combination, normalize them and superimpose them, and generate a sequence of visual difference values ​​between functional areas; S2: Based on the visual difference value sequence of the functional interval, the sight line accessibility difference threshold is compared, the combinations exceeding the threshold are extracted and their closure, light transmittance and the number of openable elements are called, the interface material replacement and structural opening requirements are judged, the interface and path association is established by combining the guide path sorting and functional attributes, the suggestions are arranged in the judgment order, and the interface reset operation list is generated; S3: calling the function area number in the interface reset operation list, obtaining the calling frequency of each function area and calculating the frequency difference and ratio of the path segments, and generating a path node weight structure diagram; S4: calling the functional area combination in the path node weight structure diagram, calculating the spatial grid distance and the path expectation difference, screening the connection relationship unchanged combination, and obtaining a feasible arrangement alternative sequence; S5: extracting the behavior order weight distribution table under the situation label according to the feasibility arrangement alternative sequence, calculating the current sorting deviation value and screening the functional areas with excessive deviation, and generating a functional area sorting update path diagram.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by extracting the line of sight length and angle offset value of the functional area boundary, a visual difference value sequence is constructed to clarify the actual difference in visual connectivity of the space, which helps to identify the position of visual obstacles and areas of uneven perception. According to the visual difference results, combined with the functional attributes and the order of the guide path, an interface adjustment operation list is constructed to improve the pertinence and execution efficiency of the interface structure adjustment. In the path construction, a weight structure diagram is established through the ratio of the behavior call frequency and the path segment change to accurately reflect the spatial correlation relationship of the usage behavior. The spatial arrangement replacement combines the connection relationship and the path expectation difference to screen the optimization plan, and enhances the logical consistency and behavior matching of the arrangement adjustment. The behavior sequence deviation is identified and the operation path map is updated to achieve dynamic adaptation of the functional area order and reconstruction of the behavior logic under the situation, and improve the spatial responsiveness and rationality of use. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a system flow chart of the present invention; Figure 2 This is a flow chart of the visual parameter acquisition module of the present invention; Figure 3 This is a flow chart of the interface adjustment determination module of the present invention; Figure 4 A flow chart of the module for constructing the behavior path of the present invention; Figure 5 This is a flow chart of the spatial arrangement replacement module of the present invention; Figure 6 This is a flow chart of the sequence adaptation update module of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0019] Example 1: Please refer to Figure 1 , the present invention provides a technical solution: the AI-based architectural design optimization system includes: The visual parameter acquisition module obtains the boundary surfaces of all functional areas in the building's indoor layout, collects the sight line length and angle offset values ​​in the forward and reverse directions, calculates the average and performs difference normalization and superposition to obtain a sequence of visual difference values ​​between functional areas; The interface adjustment judgment module calls the items exceeding the threshold in the sequence of visual difference values ​​between functional areas, judges the interface closure, light transmittance and the number of openable components, marks material replacement items and structural opening items, associates the sorting position of the interface in the guidance path with the usage attributes of the functional area, and establishes an operation sequence for all recommended items to obtain an interface reset operation list; The behavior path construction module extracts the frequency of function area calls based on the interface reset operation list, calculates the frequency difference and change ratio of path segments, marks the combinations that exceed the path coherence threshold, and obtains the path node weight structure diagram; The spatial arrangement replacement module calls the functional area combination in the path node weight structure diagram, calculates the difference between the spatial grid distance and the path expectation, screens the replacement schemes whose connection relationship has not been reduced, and obtains a feasible arrangement replacement sequence; The sequence adaptation update module arranges the functional area order adjustment plan in the alternative sequence according to the feasibility, extracts the functional area behavior order weight distribution table corresponding to the current context label in the marked exhibition and teaching scenarios, calculates the deviation value between the actual arrangement order and the weight sorting, and if the deviation value is greater than the functional area sorting offset threshold, marks the set of functional areas that need to be updated, generates a sequential update operation queue corresponding to the current context label, and obtains the functional area sorting update path diagram.

[0020] The sequence of visual difference values ​​between functional areas specifically includes the line of sight accessibility difference, angle offset difference, and difference normalization index; the interface reset operation list includes material replacement items, structural opening items, and operation sorting labels; the path node weight structure diagram specifically refers to the path node call frequency, path continuity weight, and node combination mark; the feasibility arrangement alternative sequence includes the spatial grid distance evaluation value, the path expected deviation value, and the connection relationship retention label; the functional area sorting update path diagram specifically includes the sorting offset mark, the updated functional area set, and the situational adaptation order.

[0021] See also Figure 2 , the visual parameter acquisition module includes: The boundary extraction submodule collects the boundary surfaces of all functional areas in the building interior layout, converts the outer contour lines of the functional areas in the two-dimensional plane image into boundary vertex coordinate data, uses the contour boundary closure condition to perform integrity verification on the collected boundaries, obtains the boundary line segment information of all functional areas according to the connection relationship between the spatial contours of the differentiated functional areas, and generates the boundary line segment data of the functional areas; The boundary extraction submodule collects the boundary surfaces of all functional areas in the building's indoor layout. The boundary collection process is first carried out in a two-dimensional plane image of the actual indoor layout (for example, a CAD two-dimensional layout diagram of an office building), and the outer contour line of the two-dimensional image of each functional area is extracted, such as the office area, conference room, corridor, rest area and other areas are extracted one by one, wherein the contour line of each area is scanned point by point, and the coordinate position of each corner point on the contour line of the functional area is marked and numbered in turn, such as the corner points A (3.5m, 2.0m), B (7.0m, 2.0m), C (7.0m, 5.0m) and D (3.5m, 5.0m) of the office area, and similarly, the contour points E (7.2m, 2.0m), F (10.0m, 2.0m), G (10.0m, 5.0m) and H (7.2m, 5.0m) of the conference room are processed similarly to obtain a specific two-dimensional vertex coordinate data array; secondly, according to the contour The closure condition of the boundary is used to check the integrity of the contour boundaries of all functional areas one by one. The closure judgment is made by checking the coordinates of both ends of each boundary segment of each functional area, that is, whether the first and last coordinates are consistent. For example, the office area passes through B, C, and D from the starting point A and returns to the starting point A. A coordinate closure comparison is performed to ensure the integrity of the contour. Then, based on the connection relationship between the spatial contours of the differentiated functional areas, whether the boundaries of adjacent functional areas are collinear is judged. By comparing the boundary coordinates of adjacent areas point by point, the common segments are determined (for example, the boundary points B and E, and C and H between the office area and the conference room are collinear). After the connection relationship is confirmed, the complete boundary segment data sets of each functional area are stored separately to complete the standardization of the boundary information. Finally, the functional area boundary segment data sets are generated, such as {office area: [(A, B), (B, C), (C, D), (D, A)], conference room: [(E, F), (F, G), (G, H), (H, E)]}.

[0022] The sight parameter calculation submodule calls the boundary line segment data of the functional area, sets multiple equally spaced monitoring points in the positive and negative directions of the boundary line segment, projects the sight line to each monitoring point, collects the sight line length without penetrating other boundaries, and records the angle offset value relative to the boundary normal, calculates the average length and angle offset mean of all positive sight lines, and the average length and angle offset mean of all reverse sight lines, and obtains the mean value of the two-way sight line parameter; The line of sight parameter calculation submodule calls the functional area boundary segment data. Taking the actual office area boundary segment data as an example, for the line segment (A, B), a monitoring point is set every 0.5m in its forward direction (A→B) and reverse direction (B→A), such as A1 (4.0m, 2.0m), A2 (4.5m, 2.0m) to A6 (6.5m, 2.0m), a total of 6 monitoring points. Each monitoring point projects the line of sight into the functional area until it encounters other boundary segments, and calculates the line of sight length of each monitoring point. For example, the line of sight length of A1 is 3.0m, A2 is 3.0m, and A6 is 3.0m. m, and then the average length of the forward line of sight is calculated to be 3.0 m through the length data; at the same time, the angular offset value of the line of sight of each monitoring point and the normal direction of the boundary segment is recorded, for example, the offset angles of each monitoring point are measured to be 5°, 6°, 5°, 4°, 5°, and 4° in sequence, and the average angular offset of the forward line of sight is calculated to be 4.83°; the above process is repeated to measure the reverse (B→A) monitoring points B1 to B6, and the average length is measured to be 2.8 m, and the average offset angle is 5.16°. The above process is repeated for all boundary segments of the functional area, and finally the mean value of the two-way line of sight parameters of the functional area is calculated and integrated.

[0023] The visual difference generation submodule calls the mean of the two-way sight line parameters, refers to the impact of the difference in sight line parameters between multi-functional areas on spatial perception, and introduces the sight line difference index Using the formula: ; Calculate the visual difference coefficient of the multifunctional area to form a sequence of visual difference values ​​of the functional area; in, represents the visual difference index, Representative The average length of the positive sight line of each functional area, Representative The average length of reverse sight lines in each functional area, Representative The threshold constant used in the calculation of sight difference between functional areas, is the adjustment constant used to maintain the denominator in a non-zero state, Representative The weight factor of the sight line parameters of each functional area in the difference calculation, Represents the total number of functional areas.

[0024] The visual difference generation submodule calls the mean of the two-way sight line parameters of the functional area. Taking the three functional areas of office area, conference room and corridor as examples, the parameters are shown in Table 1: Table 1 Functional area sight line parameter weight table; ; Calculate the visual difference index for each functional area : Office Area: ; Meeting Room: ; corridor: ; According to the above results, the sequence of visual difference values ​​of functional areas is obtained: ; Furthermore, the visual difference index of each functional area is accumulated according to the formula to obtain the visual difference coefficient of the multi-functional area (i.e. the total value of the overall visual difference): ; The visual difference coefficient reflects the degree of difference in spatial perception caused by the line of sight difference between functional areas in the overall space. The higher the value, the weaker the spatial visual coherence and the stronger the sense of difference. Conversely, a low difference coefficient means a more natural spatial visual transition.

[0025] Both the sequence and total value of visual difference values ​​can be used as a quantitative evaluation basis for spatial layout optimization, channel adjustment or visual permeability enhancement, further improving the functional adaptability and human perception performance of buildings or interior spaces.

[0026] The innovation of the formula is that by introducing the line of sight parameter difference index , quantify subjective visual perception into objective values, clearly point out the spatial positions or orientations that can be optimized in the spatial layout of different functional areas, and improve the accuracy and applicability of the overall layout design.

[0027] See also Figure 3 , the interface adjustment determination module includes: The difference item screening submodule is based on the visual difference value sequence of the functional area, calls the preset visual difference threshold, compares and screens multiple items in the difference value sequence one by one, obtains the difference items in the difference value sequence that exceed the visual difference threshold, determines the spatial distribution block of the difference items in the functional area, calculates the number of difference items and their corresponding interface unit numbers, and generates a super-threshold difference item set; The difference item screening submodule operates based on the visual difference value sequence of the functional area. Specifically, it calls the calculated visual difference index sequence (such as 0.4 for office area, 0.429 for conference room, and 0.5 for corridor) and sets the visual difference threshold. The threshold setting refers to the statistical distribution of the visual difference index in actual projects. For example, the statistical analysis of the visual difference value of the indoor layout of office buildings shows that the general interval of the visual difference index is [0.1, 0.6]. If it exceeds 0.35, it is judged that the visual difference is obvious, so the threshold is set to 0.35. Next, the difference value sequence is compared item by item, that is, the office area (0.4), conference room (0.429) and corridor (0.5) are compared respectively. The difference indicators of the three functional areas of the corridor (0.5) are numerically compared with the set threshold of 0.35 one by one, and it is determined that all three data exceed the threshold (for example: office area 0.4>0.35), and the specific spatial distribution positions of the three difference values ​​are clearly defined as the office area interface unit (interface numbers one and two), the conference room interface unit (interface number three) and the corridor interface unit (interface number four), and the difference items of the identified positions are counted, and the total number of difference items is 3, and finally a super-threshold difference item set marked by the interface unit number is formed, for example: super-threshold difference item set = [interface one (office area), interface two (conference room), interface four (corridor)].

[0028] The interface characteristic judgment submodule calls the above-threshold difference item set to obtain the closure parameter, light transmittance parameter and the number of openable elements of the corresponding interface unit, and judges whether the interface is closed according to the closure parameter, judges whether it has a light-transmitting structure by using the light transmittance parameter, judges the openable state by the number of openable elements, calculates the interface transparency characteristic value and performs weighted integration with the frequency of use coefficient, using the formula: ; Calculate the interface transparency difference value, call the transparency difference threshold to determine whether it is an item that needs to be adjusted, and obtain the interface attribute item that needs to be adjusted; in, Indicates The difference in the permeability of the interface, Indicates The number of different items corresponding to each interface, Indicates In the interface The closure parameter value of the term, represents the corresponding transmittance parameter, For the The light intensity coverage factor of the term, For the In the interface The number of components that can be opened corresponding to the difference item, For the The frequency of use of each interface, For the The activity value of the functional area where the interface is located, The sorting number of this interface in the navigation path. Sort and number the benchmark guides for the functional areas; Taking interface 1 as an example, the total area measured on site is 12㎡, of which the area of ​​the airtight glass curtain wall is 10㎡. The closure parameters are: ; Transmittance parameters are for single-layer glass, transmittance: ; Light intensity coverage factor (measured coverage 75%): ; Number of components that can be opened: ; Assume that the interface contains only one difference item ( ), the first part is calculated as: ; Interface usage frequency: ; The second part is calculated as: ; Then the permeability difference value of interface 1 is: ; The total area of ​​interface 3 is 10㎡, and the area of ​​airtight material is 7㎡. The closure parameters are: ; Using double-layer insulating glass, light transmittance: ; The measured light intensity coverage area is 70%, so: ; The openable elements are 1 window and 1 door: ; Assume there is only one difference term ( ), the first part is calculated as: ; The total area of ​​interface 3 is 10㎡, and the area of ​​airtight material is 7㎡. The closure parameters are: ; Using double-layer insulating glass, light transmittance: ; The measured light intensity coverage area is 70%, so: ; The openable elements are 1 window and 1 door: ; Assume there is only one difference term ( ), the first part is calculated as: ; The difference in the four-way transparency of the interface is: ; Finally, the transparency difference values ​​of interfaces one, three, and four are as follows: the reset operation list generates a sub-module that calls the interface to adjust the attribute items, obtains the sorting number of the navigation path and the usage attributes of the functional area of ​​the corresponding interface, builds an operation sequence list based on the sorting number and the usage attributes, arranges the material items and structural opening items that need to be replaced in order, establishes the interface replacement operation sequence, and obtains the interface reset operation list.

[0029] The reset operation list generation submodule call interface needs to adjust the attribute items (interface 1, interface 3 and interface 4). By obtaining the guide path sorting number of each interface (interface 1 sorting number is 2, interface 3 sorting number is 4, interface 4 sorting number is 5) and the functional area usage attribute (the office area usage attribute is daily office, the conference room usage attribute is intermittent use, and the corridor usage attribute is access connection), the reset operation order is determined based on the ascending order of sorting number and usage attribute importance (office area> conference room> corridor): interface 1→interface 3→interface 4, and further for each interface Analyze the material items or structural opening items that need to be replaced. Take interface one as an example. Since the calculated transparency difference value is too high, it is necessary to replace the interface material with high-transmittance double-layer glass or increase the structural opening (such as opening window). Interface three has high closure, so 2 opening elements need to be added. Interface four has low activity, so 1 opening element should be appropriately added. The specific replacement content is clarified to form a detailed sequence of operation steps, that is, the reset operation list is: [Interface one: replace double-layer light-transmitting glass, replace 1 opening window; Interface three: add 2 window openings; Interface four: add 1 opening door], and generate an interface reset operation list.

[0030] See also Figure 4 , the behavior path building modules include: The interface operation reset submodule extracts the function area name, trigger time and usage frequency corresponding to each operation in the list based on the interface reset operation list, classifies them by function area, obtains the calling frequency of the multi-function area in the complete interaction process, summarizes and sorts the calling frequency, and generates the function area calling frequency value; The interface operation reset submodule is based on the interface reset operation list. First, it extracts three data items, namely, the function area name, specific trigger time, and function area usage frequency, from the list one by one. For example, the operation record of interface one includes the function area name as office area, the trigger time as 08:00-18:00, and the usage frequency as 50 times / day. The function area corresponding to interface three is conference room, the trigger time as 09:00-12:00 and 14:00-17:00, and the usage frequency as 30 times / day. The function area corresponding to interface four is corridor, the trigger time as 24 hours a day, and the usage frequency as 100 times / day. The extracted data were classified separately, and the usage frequencies of different interfaces under the same functional area were added and merged. For example, in the office area, only interface 1 was used 50 times / day, interface 3 in the conference room was used 30 times / day, and interface 4 in the corridor was used 100 times / day. Then, cross-functional area frequency comparison and sorting operations were performed, that is, the combined usage frequency values ​​were directly sorted from high to low. After sorting, the corridor (100 times / day) was ranked first, the office area (50 times / day) was ranked second, and the conference room (30 times / day) was ranked third. In this way, a sequence of call frequency values ​​for each functional area was generated, and the final call frequency value of the functional area was recorded as: [corridor 100, office area 50, conference room 30].

[0031] The path segment difference calculation submodule calls the function area call frequency value to obtain the call frequency difference between consecutive path segments using the formula: ; Calculate the path segment stability score, and determine the continuity of multiple path segments based on the score to obtain the path segment frequency change rate; in, represents the stability score of the path segment, k represents the kth path segment, N represents the total number of path segments to be evaluated, Represents the calling frequency of the starting point function area of ​​the kth path segment, Represents the calling frequency of the functional area at the end point of the k-th path segment, Represents the absolute value of the difference in call frequency between the starting point and the end point of the kth path segment, represents the decay coefficient of the interval time between path segment operations, represents the operation interval time corresponding to the kth path segment, represents the attenuation factor based on the operation interval time, represents the weighting factor of the kth path segment, which is used to adjust the weight of the path segment in the overall stability score. It represents the weighted impact value of the calling frequency of the endpoint of the k-th path segment. represents adjustment parameters; Taking the three continuous paths of “corridor→office area→conference room” as an example, the stability scores of the path segments are calculated.

[0032] Table 2 Path segment parameter statistics; ; The parameters are set as follows: (Attenuation coefficient); (Coherence adjustment parameter).

[0033] Substitute into the formula to calculate: Molecular part: ; Denominator: ; Stability score results: ; Calculate the path segment stability score , indicating that there is a certain degree of call frequency fluctuation between the three path combinations. If compared with the preset path continuity threshold (for example, 0.5): when , indicating that there are obvious frequency differences between the path segments, which belongs to the path segment combination with large fluctuations; when , it means that the path segment switches smoothly and has good continuity.

[0034] In this example Therefore, it is judged that the continuity between path segments is weak. It is recommended to improve the path connectivity experience through path guidance optimization, buffer space setting, etc. in actual application.

[0035] This path segment difference calculation method introduces the attenuation factor of the operation interval time and the weight modeling of the node frequency difference to quantify the difference in path segment continuity. It can be used as a decision basis for spatial layout optimization, guided path generation and usage heat analysis, and enhance the adaptability and logical coherence of multi-path segment layout design.

[0036] The coherence structure marking submodule calls the frequency change rate of the path segment, and judges each item against the path coherence threshold according to the frequency change ratio in the path segment combination. The path segment combination with a comparison value greater than the path coherence threshold is marked, and the marked path segments are mapped to the path node structure diagram, and corresponding weights are assigned to multiple nodes to generate a path node weight structure diagram.

[0037] The coherence structure marking submodule calls the path segment frequency change rate of 0.574, and performs a specific numerical comparison operation on the path segment frequency change rate, that is, the path segment frequency change rate is compared with the path coherence threshold (the threshold 0.5 is set according to the use frequency history of the spatial structure and the user's subjective comfort survey results) item by item. If the above result 0.574>0.5, it is determined that the frequency change rate of the path segment combination (corridor→office area→conference room) exceeds the threshold, and the path segment combination is clearly marked, and then the mapping operation of the path node structure diagram is performed, that is, according to the spatial positions of the three nodes of corridor, office area and conference room, the corresponding path segment is clearly marked as a path segment with obvious fluctuations on the node structure diagram, and the corresponding node weight is assigned to the node. For example, according to the function area call frequency value sequence, the corridor node weight is assigned to 1.0 (highest frequency), the office area node weight is assigned to 0.8 (medium frequency), and the conference room node weight is assigned to 0.6 (lowest frequency). Finally, a path node weight structure diagram with path segment coherence marks and node weights is generated.

[0038] See also Figure 5 , the space arrangement replacement modules include: The path weight extraction submodule extracts the connection edge lengths, node numbers, and corresponding functional area combinations between multiple path nodes based on the path relationships marked in the path node weight structure diagram, constructs a path segment matrix according to the node connection sequence, and reassigns the weight values ​​with the physical position parameters and combination relationship parameters between nodes to generate a path segment weight matrix. The path weight extraction submodule is based on the path relationship marked in the path node weight structure diagram. First, the physical connection side lengths between the path nodes, the node numbers themselves, and the combination information of the corresponding functional areas are extracted from the structure diagram one by one. Specifically, the physical coordinates of the nodes are processed in two dimensions and then the actual connection distance between the nodes is calculated. For example, the physical distance from node number one (corridor) to node number two (office area) is confirmed to be 12 meters through on-site measurement or CAD drawings, and the physical distance from node number two to node number three (conference room) is confirmed to be 15 meters. At the same time, the combination relationship of the functional areas corresponding to the nodes is extracted. For example, the functional area combination corresponding to node one is defined as the traffic connection area, node two is the daily office functional area, and node three is the temporary office functional area. Next, the nodes are connected section by section according to the actual connection order. Construct a path segment matrix, and reset the weight values ​​of the matrix according to the physical location parameters (i.e., the measured physical distance) between nodes and the parameters of the functional combination relationship. For example, a higher weight value of 0.9 to 1.0 is assigned to a physical distance within 10 meters, a medium weight value of 0.7 to 0.89 is assigned to a distance of 10 to 20 meters, and a lower weight value of 0.5 to 0.69 is assigned to a distance exceeding 20 meters. The functional combination relationship is assigned according to the similarity of functional properties. A high functional similarity (such as office area and conference room) is assigned a weight of 0.9, a medium similarity (such as corridor and office area) is assigned a weight of 0.7, and a low similarity is assigned a weight of 0.5. The path 01→02 and the path 02→03 are assigned path weight values ​​of 0.7 and 0.9 respectively. Finally, the path segment weight matrix is ​​obtained as follows: ; The expected difference calculation submodule calls the path segment weight matrix and calculates the functional adaptability difference between nodes based on the node function attributes and the actual distance, using the formula: ; Calculate the multi-node functional adaptability deviation and establish the expected deviation value of the functional path; in, Represents the functional adaptability deviation value between path nodes, Represents the path The actual physical distance between the nodes connecting the segments, Represents the path The functional difference level value between the functional area combinations corresponding to the segment nodes, Represents the path The structural strength weight value of the functional area combination in the node connected by the segment, Represents the path The participation level of the segment node in the functional connectivity network, Indicates the upper limit of the count of all segments in the path. Represents the total number of nodes involved in weight evaluation in the path segment; The expected difference calculation submodule calls the path segment weight matrix mentioned above and calculates the functional adaptability difference between nodes based on the node functional attributes and the actual physical distance. Taking the actual path segment as an example, the actual physical distance The distance from path 01 to 02 is 12 meters, and the distance from path 02 to 03 is 15 meters. The functional difference level value is The expert ratings determined that the significant difference was 0.8-1.0, the medium difference was 0.5-0.79, and the low difference was 0.1-0.49. For example, the medium difference between the corridor and the office area was 0.6, and the low difference between the office area and the conference room was 0.4. The functional combination structure strength weight value It is obtained through the on-site structural connection stability assessment. The reinforced concrete frame structure is assigned a value of 0.9, the lightweight partition wall structure is 0.7, and the glass curtain wall structure is 0.6. For example, 01→02 is the reinforced concrete structure weight 0.9, 02→03 is the lightweight partition wall structure weight 0.7, and the node participation level The value is determined by the average number of times the actual node participates in the path segment every day. The value is 0.9 for high participation (more than 80 times), 0.7 for medium participation (50-79 times), and 0.5 for low participation (less than 50 times). For example, if the participation of 01→02 is 100 times, the value is 0.9, and if the participation of 02→03 is 60 times, the value is 0.7. The parameters are shown in Table 3: Table 3 Node function adaptability parameter table; ; Based on the above data, enter the formula to calculate the functional adaptability deviation value : ; Substitute the specific data into: ; The result shows that the functional adaptability deviation value of the path node is 4.0199, which is greater than the preset deviation benchmark (usually set between 1 and 3), indicating that there is an obvious deviation in the functional adaptability of the path.

[0039] The benefit of the formula is that it effectively quantifies the functional adaptability deviation between nodes by introducing the functional difference level and the structural strength and node participation for combined calculation.

[0040] The alternative sequence screening submodule counts the number of connections, makes structural preservation judgments and sorts the offset values ​​of all replaceable path combinations according to the screening conditions that the expected deviation value of the functional path and the node connection relationship in the path segment have not been reduced. It screens out path combinations with reduced number of connection edges, sorts and selects the structural alternative sequences with the highest offset values, and generates a feasible arrangement alternative sequence.

[0041] The alternative sequence screening submodule performs screening operations based on the calculated expected deviation value of the functional path of 4.0199. First, the number of node connections of all replaceable path combinations is counted one by one. If the number of connections of the original path is 2 segments, the number of connections of the alternative combination must be no less than the number of connections of the original path (2 segments). If the number of connection segments of the alternative combination is less than 2 segments, it is directly excluded. Then, a clear preservation judgment is made on the path structure, that is, whether the alternative path combination maintains the node connection relationship without reduction. For example, a valid connection must be maintained between node 1 and node 2. If the alternative path does not maintain this connection, the alternative is directly excluded. Path, then calculate the offset values ​​of the path combinations that meet the above two conditions respectively, and sort them from large to small according to the calculated functional adaptability deviation values. For example, the functional deviation value of alternative path A (node ​​1 → node 4 → node 2 → node 3) is 4.5, and the deviation value of path B (node ​​1 → node 5 → node 2 → node 3) is 3.8. After sorting, the deviation value of path A ranks first, followed by path B. A feasibility arrangement alternative sequence is generated based on the sorting results, that is, the generated structural alternative sequence is [A: node 1 → node 4 → node 2 → node 3, B: node 1 → node 5 → node 2 → node 3].

[0042] See also Figure 6 , the sequence adaptation update module includes: The context label extraction submodule arranges the functional area sequence adjustment plan in the alternative sequence according to the feasibility, obtains the marked exhibition and teaching scene data, calls the functional area behavior sequence weight distribution table corresponding to the context label, extracts the matching items between the context label and the functional area behavior, determines the weight arrangement order according to the matching relationship, and generates the context label weight sequence table; The context label extraction submodule arranges the functional area sequence adjustment plan in the alternative sequence according to the feasibility. First, it extracts the pre-labeled scene data of different functional areas such as the exhibition area and the teaching area, and obtains the specific context labels corresponding to the functional areas from the site or layout data. For example, the exhibition area corresponds to "visual guidance" and "immersive experience", and the teaching area corresponds to "interactive demonstration" and "concentrated explanation" and other labels. The functional area behavior sequence weight distribution table corresponding to the context label is called, and the label matching degree is determined by matching and comparing the acquired context labels with each label in the weight distribution table one by one. For example, the exhibition area label The matching weight of "visual guidance" is 0.9, the weight of "immersive experience" is 0.85, the weight of the teaching area label "interactive demonstration" is 0.8, and the weight of "concentrated explanation" is 0.95. The labels are sorted and reorganized from high to low according to the label weights, such as teaching area (concentrated explanation 0.95, interactive demonstration 0.8), exhibition area (visual guidance 0.9, immersive experience 0.85). The functional area behavior order corresponding to the sorted situational labels is rearranged into teaching area → exhibition area, and finally the situational label weight sequence table is generated, that is: [teaching area (weight 0.95), exhibition area (weight 0.90)].

[0043] The behavior sequence deviation calculation submodule calls the current actual arrangement order according to the context label weight sequence table, calculates the difference between the actual order and the weight order, and uses the formula: ; Calculate the behavior order deviation index, determine whether it is greater than the benchmark according to the function area sorting deviation threshold, select the function areas that need to be adjusted, and generate a set of function areas to be updated; in, represents the behavior order deviation index, Representative The position of each functional area in the actual layout, Representative The position of each functional area in the weight sorting, For the The visibility coefficient of each functional area, Indicates the total number of functional areas; The behavior sequence deviation calculation submodule calls the above generated context label weight sequence table and compares it item by item with the current actual functional area arrangement order. For example, the current actual arrangement order is: exhibition area (position 1), teaching area (position 2), and the weight order is: teaching area (position 1), exhibition area (position 2). Then, the actual arrangement position is clearly extracted. (Exhibition area position 1, teaching area position 2) and weighted ranking position (Exhibition area location 2, teaching area location 1), then call the visibility coefficient of the functional area (The actual visible area of ​​the functional area accounts for the total layout area, for example, the exhibition area accounts for 0.7, and the teaching area accounts for 0.8), and the difference between the actual position and the weighted position of each functional area is calculated in turn, and then divided by the corresponding visibility coefficient, and brought into the formula to calculate the behavior order offset index Calculation: ; The actual calculation example is as follows: ; The calculated behavior order offset index of 3.6033 is compared with the preset functional area sorting offset threshold (generally between 2.0 and 3.0 based on user feedback statistics), and the result is 3.6033>3.0. Therefore, it is determined that the actual arrangement order of the exhibition area and the teaching area needs to be adjusted, and the set of functional areas that need to be adjusted is determined to be: [exhibition area, teaching area]. This result shows that there is a significant difference between the current actual layout order and the expected weight sorting.

[0044] The benefit of the formula is that, by comparing the actual position with the weighted position and introducing the visibility coefficient, the degree of deviation between the actual layout order of the functional areas and the expected order of the situation can be clarified.

[0045] The sequential update path generation submodule extracts the arrangement information of the functional areas to be updated in the current order and their target positions in the weight sorting according to the set of functional areas to be updated, constructs a sorting alternative path according to the difference order adjustment method, forms an adjustment action queue and sorts it, and screens feasible operation paths in combination with the actual site feasibility comparison structure required for the sequential adjustment to obtain the functional area sorting update path diagram.

[0046] The sequential update path generation submodule first extracts the position information of each functional area in the current actual layout order (the exhibition area is currently at position 1, and the teaching area is at position 2) according to the above-determined set of functional areas to be updated (exhibition area, teaching area), as well as their target positions in the context label weight sequence table (exhibition area target position 2, teaching area target position 1), and clearly determines the position difference between the actual position and the target position (the exhibition area needs to be adjusted backward by 1, and the teaching area needs to be adjusted forward by 1), and then determines the specific position adjustment action according to the numerical value of the difference position, for example, the teaching area is adjusted from the 2nd position to the 1st position, and the exhibition area is adjusted from the 1st position to the 2nd position. Adjust to the second position, form a specific sorting alternative path plan, and arrange all adjustment actions in sequence to form a clear functional area adjustment action queue: [(teaching area: position 2→position 1), (exhibition area: position 1→position 2)]. Next, compare the action queue with the feasibility of the actual site space item by item. For example, determine whether there are structural obstacles on the path from position 2 to position 1 of the teaching area. If not, it is determined to be a feasible operation path. The same is true for the adjustment of the exhibition area. Finally, a clear functional area sorting update path map is formed, which clearly indicates that the order of the new functional areas after adjustment is teaching area (position 1) and exhibition area (position 2).

[0047] Table 4 Functional area behavior sequence adjustment parameter table; ; As shown in Table 4, the specific parameters and operations required for adjusting the functional area sequence are clearly given. The data in the table can be directly used to clarify the generation and execution of the functional area sequence update path diagram in the actual implementation process.

[0048] The AI-based architectural design optimization method is performed based on the above-mentioned AI-based architectural design optimization system, and includes the following steps: S1: Obtain the boundary surfaces of all functional areas in the building's indoor layout, collect the sight line length and angle offset values, calculate the sight line difference and angle difference between each combination, normalize them and superimpose them, and generate a sequence of visual difference values ​​between functional areas; S2: Compare the sight accessibility difference threshold based on the sequence of visual difference values ​​between functional areas, extract combinations exceeding the threshold and call their closure, light transmittance and the number of openable components, determine the interface material replacement and structural opening requirements, combine the guide path sorting and functional attributes to establish the interface and path association, arrange the suggestions in the judgment order, and generate an interface reset operation list; S3: calling the function area number in the reset operation list of the interface, obtaining the calling frequency of each function area and calculating the frequency difference and ratio of the path segments, and generating a path node weight structure diagram; S4: Call the functional area combination in the path node weight structure diagram, calculate its spatial grid distance and path expectation difference, screen the connection relationship unchanged combination, and obtain the feasible arrangement alternative sequence; S5: Extract the behavior order weight distribution table under the situation label according to the feasibility arrangement alternative sequence, calculate the current sorting deviation value and filter out the functional areas with excessive deviation, and generate the functional area sorting update path diagram.

[0049] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. AI-based architectural design optimization system, characterized by: The system comprises: The visual parameter acquisition module obtains the boundary surfaces of all functional areas in the building's indoor layout, collects the sight line length and angle offset values ​​in the forward and reverse directions, calculates the average and performs difference normalization and superposition to obtain a sequence of visual difference values ​​between functional areas; The interface adjustment determination module calls the items exceeding the threshold value in the sequence of visual difference values ​​between the functional areas, determines the interface closure, light transmittance and the number of openable components, marks the material replacement items and the structural opening items, associates the sorting position of the interface in the guide path with the functional area usage attributes, and establishes an operation sequence for all the suggested items to obtain an interface reset operation list; The behavior path construction module extracts the calling frequency of the function area based on the interface reset operation list, calculates the frequency difference and change ratio of the path segments, marks the combination exceeding the path coherence threshold, and obtains the path node weight structure diagram; The spatial arrangement replacement module calls the functional area combination in the path node weight structure diagram, calculates the difference between the spatial grid distance and the path expectation, screens the replacement schemes whose connection relationship has not been reduced, and obtains a feasible arrangement replacement sequence.

2. The AI-based architectural design optimization system according to claim 1, characterized in that: The sequence of visual difference values ​​between functional intervals specifically includes line of sight accessibility difference, angle offset difference, and difference normalization index; the interface reset operation list includes material replacement items, structural opening items, and operation sorting labels; the path node weight structure diagram specifically refers to path node call frequency, path continuity weight, and node combination mark; the feasibility arrangement alternative sequence includes spatial grid distance evaluation value, path expected deviation value, and connection relationship retention label.

3. The AI-based architectural design optimization system according to claim 2, characterized in that: The visual parameter acquisition module includes: The boundary extraction submodule collects the boundary surfaces of all functional areas in the building interior layout, converts the outer contour lines of the functional areas in the two-dimensional plane image into boundary vertex coordinate data, uses the contour boundary closure condition to perform integrity verification on the collected boundaries, obtains the boundary line segment information of all functional areas according to the connection relationship between the spatial contours of the differentiated functional areas, and generates the boundary line segment data of the functional areas; The sight parameter calculation submodule calls the boundary line segment data of the functional area, sets a plurality of equally spaced monitoring points in the positive and negative directions of the boundary line segment, projects a sight line to each monitoring point, collects the sight line length without penetrating other boundaries, and records the angle offset value relative to the boundary normal, calculates the average length and angle offset mean of all positive sight lines, and the average length and angle offset mean of all reverse sight lines, and obtains the mean value of the two-way sight line parameter; The visual difference generation submodule calls the mean of the two-way sight line parameters, refers to the influence of the difference of sight line parameters between multi-functional areas on spatial perception, and introduces the sight line difference index Using the formula: ； Calculate the visual difference coefficient of the multifunctional area to form a sequence of visual difference values ​​of the functional area; in, represents the visual difference index, Representative The average length of the positive sight line of each functional area, Representative The average length of reverse sight lines in each functional area, Representative The threshold constant used in the calculation of sight difference between functional areas, is the adjustment constant used to maintain the denominator in a non-zero state, Representative The weight factor of the sight line parameters of each functional area in the difference calculation, Represents the total number of functional areas.

4. The AI-based architectural design optimization system according to claim 3, characterized in that: The interface adjustment determination module comprises: The difference item screening submodule calls a preset visual difference threshold based on the visual difference value sequence of the functional area, compares and screens multiple items in the difference value sequence one by one, obtains the difference items in the difference value sequence that exceed the visual difference threshold, determines the spatial distribution block where the difference items are located in the functional area, calculates the number of difference items and their corresponding interface unit numbers, and generates an over-threshold difference item set; The interface characteristic judgment submodule calls the above-threshold difference item set to obtain the closure parameter, light transmittance parameter and the number of openable elements of the corresponding interface unit, and judges whether the interface is closed according to the closure parameter, judges whether it has a light-transmitting structure by using the light transmittance parameter, judges the openable state by the number of openable elements, calculates the interface transparency characteristic value and performs weighted integration with the frequency of use coefficient, using the formula: ； Calculate the interface transparency difference value, call the transparency difference threshold to determine whether it is an item that needs to be adjusted, and obtain the interface attribute item that needs to be adjusted; in, Indicates The difference in the permeability of the interface, Indicates The number of different items corresponding to each interface, Indicates In the interface The closure parameter value of the term, represents the corresponding transmittance parameter, For the The light intensity coverage factor of the term, For the In the interface The number of components that can be opened corresponding to the difference item, For the The frequency of use of each interface, For the The activity value of the functional area where the interface is located, The sorting number of this interface in the navigation path. Sort and number the benchmark guides for the functional areas; The reset operation list generation submodule calls the attribute items that need to be adjusted in the interface, obtains the navigation path sorting number and function area usage attributes of the corresponding interface, builds an operation priority list based on the sorting number and usage attributes, arranges the material items and structural opening items that need to be replaced in sequence, establishes the interface replacement operation sequence, and obtains the interface reset operation list.

5. The AI-based architectural design optimization system according to claim 4, characterized in that: The behavior path building module includes: The interface operation reset submodule extracts the function area name, trigger time and usage frequency corresponding to each operation in the list based on the interface reset operation list, classifies them by function area, obtains the calling frequency of the multi-function area in the complete interaction process, summarizes and sorts the calling frequency, and generates the function area calling frequency value; The path segment difference calculation submodule calls the function area call frequency value to obtain the call frequency difference between consecutive path segments using the formula: ； Calculate the path segment stability score, and determine the continuity of multiple path segments based on the score to obtain the path segment frequency change rate; in, represents the stability score of the path segment, k represents the kth path segment, N represents the total number of path segments to be evaluated, Represents the calling frequency of the starting point function area of ​​the kth path segment, Represents the calling frequency of the functional area at the end point of the k-th path segment, Represents the absolute value of the difference in call frequency between the starting point and the end point of the kth path segment, represents the decay coefficient of the interval time between path segment operations, represents the operation interval time corresponding to the kth path segment, represents the attenuation factor based on the operation interval time, represents the weighting factor of the kth path segment, which is used to adjust the weight of the path segment in the overall stability score. It represents the weighted impact value of the calling frequency of the endpoint of the k-th path segment. represents adjustment parameters; The coherence structure marking submodule calls the path segment frequency change rate, and judges each item against the path coherence threshold according to the frequency change ratio in the path segment combination, and marks the path segment combination whose comparison value is greater than the path coherence threshold, maps the marked path segments to the path node structure diagram, assigns corresponding weights to multiple nodes, and generates a path node weight structure diagram.

6. The AI-based architectural design optimization system according to claim 5, characterized in that: The space arrangement replacement module includes: The path weight extraction submodule extracts the connection edge lengths, node numbers and corresponding functional area combinations between multiple path nodes based on the path relationship marked in the path node weight structure diagram, constructs a path segment matrix according to the node connection sequence, and reassigns the weight values ​​according to the physical position parameters and combination relationship parameters between the nodes to generate a path segment weight matrix; The expected difference calculation submodule calls the path segment weight matrix and calculates the functional adaptability difference between nodes according to the node function attributes and the actual distance, using the formula: ； Calculate the multi-node functional adaptability deviation and establish the expected deviation value of the functional path; in, Represents the functional adaptability deviation value between path nodes, representing the The actual physical distance between the nodes connected by the segment, representing the The functional difference level value between the functional area combinations corresponding to the segment nodes, Represents the path The structural strength weight value of the functional area combination in the node connected by the segment, Represents the path The participation level of the segment node in the functional connectivity network, Indicates the upper limit of the count of all segments in the path. Represents the total number of nodes involved in weight evaluation in the path segment; The alternative sequence screening submodule performs connection number statistics, structure preservation judgment and offset value sorting on all replaceable path combinations according to the screening conditions that the expected deviation value of the functional path and the node connection relationship in the path segment have not been reduced, screens out path combinations with reduced number of connection edges, sorts and selects the structural alternative sequences with the highest offset value, and generates a feasible arrangement alternative sequence.

7. The AI-based architectural design optimization system according to claim 6, characterized in that: The system further comprises: The sequence adaptation update module arranges the functional area sequence adjustment scheme in the alternative sequence according to the feasibility, extracts the functional area behavior sequence weight distribution table corresponding to the current context label in the marked exhibition and teaching scenes, calculates the deviation value between the actual arrangement sequence and the weight sorting, and if the deviation value is greater than the functional area sorting offset threshold, marks the functional area set that needs to be updated, generates a sequence update operation queue corresponding to the current context label, and obtains the functional area sorting update path diagram; The functional area sorting update path diagram specifically includes a sorting offset mark, an updated functional area set, and a context adaptation order.

8. The AI-based architectural design optimization system according to claim 7, characterized in that: The sequence adaptation update module comprises: The context label extraction submodule obtains the marked display and teaching scene data according to the functional area sequence adjustment plan in the feasibility arrangement alternative sequence, calls the functional area behavior sequence weight distribution table corresponding to the context label, extracts the matching items between the context label and the functional area behavior, determines the weight arrangement order according to the matching relationship, and generates the context label weight sequence table; The behavior sequence deviation calculation submodule calls the current actual arrangement sequence according to the context label weight sequence table, calculates the difference between the actual sequence and the weight sequence, and uses the formula: ； Calculate the behavior order deviation index, determine whether it is greater than the benchmark according to the function area sorting deviation threshold, select the function areas that need to be adjusted, and generate a set of function areas to be updated; in, represents the behavior order deviation index, Representative The position of each functional area in the actual layout, Representative The position of each functional area in the weight sorting, For the The visibility coefficient of each functional area, Indicates the total number of functional areas; The sequential update path generation submodule extracts the arrangement information of the set of functional areas to be updated in the current order and their target positions in the weight sorting according to the set, constructs a sorting alternative path according to the difference sequence adjustment method, forms an adjustment action queue and sorts it, and screens feasible operation paths in combination with the actual site feasibility comparison structure required for the sequential adjustment to obtain a functional area sorting update path diagram.

9. The AI-based architectural design optimization method is characterized by: The AI-based architectural design optimization system according to any one of claims 1 to 8 comprises the following steps: S1: Obtain the boundary surfaces of all functional areas in the building's indoor layout, collect the sight line length and angle offset values, calculate the sight line difference and angle difference between each combination, normalize them and superimpose them, and generate a sequence of visual difference values ​​between functional areas; S2: Based on the visual difference value sequence of the functional interval, the sight line accessibility difference threshold is compared, the combinations exceeding the threshold are extracted and their closure, light transmittance and the number of openable elements are called, the interface material replacement and structural opening requirements are judged, the interface and path association is established by combining the guide path sorting and functional attributes, the suggestions are arranged in the judgment order, and the interface reset operation list is generated; S3: calling the function area number in the interface reset operation list, obtaining the calling frequency of each function area and calculating the frequency difference and ratio of the path segments, and generating a path node weight structure diagram; S4: calling the functional area combination in the path node weight structure diagram, calculating the spatial grid distance and the path expectation difference, screening the connection relationship unchanged combination, and obtaining a feasible arrangement alternative sequence; S5: extracting the behavior order weight distribution table under the situation label according to the feasibility arrangement alternative sequence, calculating the current sorting deviation value and screening the functional areas with excessive deviation, and generating a functional area sorting update path diagram.

Images