An experimental functional space layout optimization method based on building renewal constraints

Abstract

The present application relates to the technical field of architectural design and optimization, and particularly relates to an experimental functional space layout optimization method based on building renewal constraints. The method comprises the following steps: obtaining target building plane data, generating a basic cutting line set through grid discretization, and using a weighted priority cutting line set to perform constraint-aware discretization on the building plane; analyzing the discretized functional requirements, merging them into functional groups according to group labels to generate a candidate rectangular size pool, dividing the building plane into non-overlapping minimum rectangular atomic regions, enumerating atomic region combinations for each functional group to meet connectivity, and generating a group-level layout scheme based on constraint programming for non-overlapping space assignment; performing room-level space allocation based on full coverage and column network accommodation constraints in the group division area, generating service space candidate sites near obstacle neighborhoods and corridor intersection nodes, and obtaining a preliminary layout scheme by inserting the candidate sites, and finally obtaining a final layout scheme through re-planning.

Description

Technical Field

[0001] This invention relates to the field of architectural design and optimization technology, and in particular to a method for optimizing the layout of laboratory functional spaces in the renovation and upgrading of protected buildings. Background Technology

[0002] In the renovation and upgrading of protected buildings, the layout design of experimental functional spaces is an extremely challenging task. Protected buildings typically have irregular spatial forms, preserved load-bearing structures (such as column grids), and strict requirements for historical preservation. Furthermore, laboratory functions often require close connections between different laboratory modules and concise circulation to meet the specific experimental needs and safety requirements of researchers. Currently, the conventional approach to the interior spatial layout of such buildings relies primarily on the architect's experience and manual drawing. Designers first identify the locations of major walls, column grids, and corridors based on architectural survey drawings, and then manually divide the space on the drawings according to the list of experimental functional requirements (such as laboratories, preparation rooms, equipment rooms, etc.). This process often employs a trial-and-error method, repeatedly adjusting room boundaries and positions to meet basic requirements such as area, connectivity, and lighting. In addition, while some computer-aided design tools support parametric modeling, they are usually only used to generate rectangular grids or divide spaces based on fixed modules, failing to effectively handle the rigid constraints of non-orthogonal walls, oblique boundaries, and column positions on spatial division common in protected buildings. In terms of functional allocation, existing technologies mostly adopt manual grouping or simple surface accumulation methods, lacking a systematic consideration of the internal connectivity of functional groups and the overall space utilization.

[0003] Conventional layout adjustments rely on manual experience, which is inefficient and struggles to guarantee optimal solutions. This is especially true when dealing with complex, irregularly shaped spaces and multiple constraints (such as column grid avoidance and minimizing circulation routes), easily leading to spatial fragmentation or functional inconsistencies and resulting in high costs for subsequent modifications. Existing automated layout methods are mostly designed for standard rectangular spaces in new buildings and cannot be directly transferred to protected building environments. These methods generally lack the ability to perceive key architectural features such as morphological transition lines and main corridor edges, and fail to incorporate column grid axes as rigid constraints into the layout solution process. This often results in layout solutions that conflict with the actual structure of the protected building; for example, room boundaries may intersect with columns, corridors may be blocked, or experimental functional areas may fail to meet basic clearance requirements for equipment placement and personnel passage. Therefore, there is an urgent need for a layout method that can automatically adapt to the irregular shapes of protected buildings and effectively handle multiple constraints. Summary of the Invention

[0004] This invention provides a method for optimizing the layout of experimental functional spaces based on building renewal constraints, which can solve the problems in the prior art.

[0005] A first aspect of the present invention provides a method for the layout of experimental functional spaces in the renovation of protected buildings based on constraint optimization, comprising:

[0006] The target building plan data includes building geometry data, basic structural component data, basic access organization data, protection element data, and functional requirement data. Building geometry data is used to determine the effective layout range and outer contour transition characteristics of the building plan; basic structural component data is used to determine the column grid axis, inaccessible obstacles, and column placement constraints; basic access organization data is used to determine the main corridor edge line, entrance and exit locations, and circulation distances; protection element data is used to determine the spatial boundaries that cannot be cut, cannot be used to arrange functional spaces, or are subject to modification restrictions; functional requirement data is used to determine the area, group affiliation, and layout restrictions of experimental functional rooms.

[0007] The acquired target building plan data is discretized by grid to generate a basic cutting line set. From this set, the shape turning line, the main corridor edge line and the column grid axis are selected as priority cutting lines. The weighted priority cutting line set is sorted by weight and used to perform constraint perception discretization on the building plan.

[0008] Functional requirements are analyzed on the discretized building plan. The functional requirements are grouped into functional groups according to the group labels and a candidate rectangular size pool is generated for each functional group. The building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions and the atomic region combinations that satisfy connectivity are enumerated for each functional group. Based on constraint planning, non-overlapping spatial assignment is performed on each functional group to generate a group-level layout scheme.

[0009] Based on the cluster-level layout scheme, within the division area of ​​each functional cluster, the fixed protection area corresponding to each protection element is taken as an unoccupiable locked area. Based on full coverage and column grid accommodation constraints, room-level space allocation is carried out to complete the subdivision of rooms within the cluster. The wet laboratory functional room is arranged around the outside of the fixed protection area. Service space candidate positions are generated near the intersection of obstacle neighborhood and corridor and placed into the service space to obtain the preliminary layout scheme.

[0010] The preliminary layout scheme is scored globally, and problem areas are identified based on circulation distance, column conflicts, and morphological quality. The problem areas are then locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

[0011] From the basic cutting line set, shape transition lines, main corridor edge lines, and column grid axes are selected as priority cutting lines and sorted by weight to form a weighted priority cutting line set, including:

[0012] Extract the coordinates of each vertex of the building's outer contour and calculate the turning angle of the adjacent side at each vertex to determine the concavity / convexity attribute. Locate all concave vertices and generate candidate segmentation lines that pass through the concave vertices and are parallel to the adjacent contour sides. Identify the candidate segmentation lines that can completely divide the concave area into independent rectangular areas as shape transition lines and assign them the highest weight.

[0013] Extract continuous line segments that are consistent with the direction of the corridor along both sides of the main corridor, merge the continuous line segments into the edge line of the main corridor and assign them the second highest weight;

[0014] The column direction axis is formed by sequentially connecting the center points of the same column along each column direction and sequentially connecting the center points of the same row along each row direction. The column direction axis and the row direction axis are then merged into the column grid axis and assigned a third weight.

[0015] The morphological transition line is inserted into the basic cutting line set with the highest priority, the main corridor edge line is inserted into the basic cutting line set with the second highest priority, and the column grid axis is inserted into the basic cutting line set with the third priority. This allows the high-weight priority cutting line to obtain priority division weight during the discretization process. The three types of priority cutting lines are then added to the basic cutting line set in a weighted overlapping manner to form the weighted priority cutting line set.

[0016] Candidate segmentation lines that can completely divide a concave region into independent rectangular regions are identified as shape transition lines, including:

[0017] Extract all vertices of the building's planar outline and number each vertex in a clockwise direction, recording the coordinates of each vertex. For each vertex, calculate the cross product sign formed by the previous outline edge vector and the next outline edge vector, and determine whether the current vertex is a convex vertex or a concave vertex based on the sign of the cross product sign.

[0018] Traverse all concave vertices and, starting from the current concave vertex, draw rays along the extension directions of the two contour edges adjacent to that concave vertex. The line connecting the contour point where each ray first intersects the outer contour of the building plane and the concave vertex is taken as the candidate dividing line.

[0019] The building plan is divided into two sub-regions by each candidate dividing line. The rectangular integrity of each sub-region is calculated as the ratio of the area of ​​the sub-region to the area of ​​its smallest bounding rectangle. The weighted average of the rectangular integrity of the two sub-regions is then taken. At the same time, the ratio of the sum of the areas of the two sub-regions to the total area of ​​the building plan is calculated as the area utilization rate. The weighted average of the rectangular integrity and the area utilization rate are used as a comprehensive index to evaluate the dividing effect of the candidate dividing line.

[0020] Sort all candidate dividing lines according to the comprehensive index from high to low, and determine the candidate dividing line corresponding to the maximum value of the comprehensive index as the morphological turning line.

[0021] The building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions, and combinations of atomic regions satisfying connectivity are enumerated for each functional cluster, including:

[0022] Using the weighted priority cutting line set as the dividing boundary, the discretized building plane is successively divided along the extension direction of each priority cutting line, so that the area between two adjacent priority cutting lines is divided into rectangular grid units. Each rectangular grid unit is defined as an atomic region and all atomic regions are non-overlapping and completely cover the entire building plane.

[0023] An adjacency graph of atomic regions is established with each atomic region as the basic operation unit. Each node in the graph represents an atomic region and is connected to the nodes corresponding to adjacent atomic regions by an edge. The length of the shared boundary between the two atomic regions connected by each edge is recorded as the adjacency weight.

[0024] Based on the area requirements of each functional group, the adjacency graph of the atomic regions is used as the search space. Starting from the initial atomic region corresponding to the location of each functional group, all connected subgraphs including the initial atomic region are enumerated by breadth-first traversal. The set of atomic regions contained in each connected subgraph is taken as a connected region combination. Connected region combinations whose absolute difference between the total area of ​​the connected region combination and the target area of ​​the corresponding functional group is within the preset tolerance range are selected as candidate assignment combinations for the functional group.

[0025] Based on constraint programming, non-overlapping spatial assignments are performed on each of the aforementioned functional clusters, including:

[0026] Using each candidate assignment combination of each functional group as a Boolean decision variable, construct a non-overlapping constraint that each atomic region is occupied by at most one functional group and a mutual exclusion constraint that each functional group has one and only one candidate assignment combination selected, and construct a cardinality constraint that the number of atomic regions contained in the selected candidate assignment combination must be within a preset range.

[0027] The absolute value of the difference between the actual total area of ​​the selected candidate assignment combination and the target area of ​​the corresponding functional group is used as the area deviation penalty term, and the difference between the total length of the boundary line of the selected candidate assignment combination and the perimeter of the square with the same area is used as the morphological density reward term. The weighted sum of the area deviation penalty term and the morphological density reward term is used as the objective function.

[0028] Using the non-overlapping constraints, the mutual exclusion constraints, and the cardinal constraints as constraints, and minimizing the objective function as the optimization direction, the values ​​of each Boolean decision variable are solved through constraint programming to minimize the objective function. The set of atomic regions corresponding to the selected candidate assignment combinations in each functional group is taken as the occupied area of ​​the functional group on the building plan. The set of occupied areas of all functional groups constitutes the cluster-level layout scheme.

[0029] Room-level space allocation is carried out within the defined areas of each functional cluster, and service spaces are placed near obstacle neighborhoods and corridor intersections, including:

[0030] Within the division area of ​​each functional group, the area requirements and rectangular shape requirements of each room are used as input parameters. The hard constraint is full coverage, that is, the sum of the rectangular areas of each room must be equal to the area of ​​the division area of ​​the functional group. The spatial constraint is column grid accommodation, that is, when the minimum area of ​​a room is greater than three times the area of ​​a column grid unit, the room is allowed to accommodate columns. By enumerating the candidate rectangular size combinations of each room and verifying whether each combination meets the full coverage constraint and column grid accommodation constraint, the combination that meets all the constraints is taken as the room-level subdivision scheme.

[0031] The union of the preset neighborhood range of each obstacle and the preset neighborhood range of the corridor intersection node is taken as the service space candidate area. Multiple candidate placement positions are generated in the candidate area at preset intervals, and the target placement position is determined by the minimum distance between each candidate placement position and the boundary of the adjacent assigned room as the evaluation index.

[0032] After placing the service space at the target placement location, it is detected whether there is an overlap or gap between the boundary of the service space and the boundary of the allocated room. If there is an overlap or gap, the boundaries of the remaining allocated rooms are kept unchanged, and the functional groups involved in the overlapping or gap area are reassigned to room-level spaces separately.

[0033] The preliminary layout scheme is globally scored, and problem areas are identified for local release and re-solution, including:

[0034] The sum of the shortest path lengths between each functional group and between each functional group and the entrance / exit is used as the circulation distance score; the cumulative value of the number of columns contained in a single room in all rooms is used as the column conflict score; the arithmetic mean of the ratio of the area of ​​each room to the area of ​​its smallest bounding rectangle is used as the morphological quality score; and the weighted sum of the above three scores is used as the global score.

[0035] The contribution of each functional group to the circulation distance score and morphological quality score in the global score is decomposed from the global score. Functional groups with a contribution exceeding a preset threshold are marked as problem areas and the room subdivision results in the problem areas are released. The room subdivision results of functional groups in non-problem areas are retained as locked layouts.

[0036] The problem area after release is taken as the solution space. The locked layout is taken as an immutable boundary condition. The solution space is redistributed at the room level based on the full coverage partitioning constraint and the column grid accommodation constraint. The acceptance criterion is whether to retain the current rediscovery result based on whether the global score after rediscovery is improved compared with the previous round of global score. The iterative process of global score, problem area identification and local rediscovery is repeated until the global score is not improved in multiple consecutive rounds of iteration. Then the iteration is terminated and the final layout scheme is output.

[0037] A second aspect of the present invention provides a constraint-optimized experimental functional space layout system for the renovation of protected buildings, comprising:

[0038] The discretization unit is used to generate a basic cutting line set by discretizing the acquired target building plan data through a grid, and to select the shape turning line, the main corridor edge line and the column grid axis as priority cutting lines. The weighted priority cutting line set is sorted according to weight and used to perform constraint-aware discretization of the building plan.

[0039] The cluster assignment unit is used to analyze the functional requirements of the discretized building plan, group the functional requirements into functional clusters according to the cluster label, generate a candidate rectangular size pool for each functional cluster, divide the building plan into non-overlapping and tessellated minimum rectangular atomic regions, enumerate the atomic region combinations that satisfy connectivity for each functional cluster, and generate a cluster-level layout scheme by non-overlapping spatial assignment of each functional cluster based on constraint planning.

[0040] The room allocation unit is used to perform room-level space allocation within the division area of ​​each functional group based on the group-level layout scheme and full coverage and column grid accommodation constraints to complete the subdivision of rooms within the group. It generates service space candidate positions near obstacle neighborhoods and corridor intersection nodes and places them into service spaces to obtain a preliminary layout scheme.

[0041] The iterative optimization unit is used to perform a global score on the preliminary layout scheme and identify problem areas based on the distance of the movement line, the conflict of the column positions and the shape quality. The problem areas are locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

[0042] A third aspect of the present invention provides an electronic device, comprising:

[0043] processor;

[0044] Memory used to store processor-executable instructions;

[0045] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0046] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0047] A constraint-optimized method for the layout of experimental functional spaces in the renovation of protected buildings significantly improves the rationality and efficiency of such layouts. This method discretizes the building plan using a priority cutting line set, effectively preserving key structural features such as the original form transitions, main corridor edges, and column grid axes. This ensures that the layout scheme is generated while respecting the spatial logic of the historical building, avoiding arbitrary fragmentation or destruction of protected elements. Functional requirements are merged into a candidate rectangular size pool, enabling each functional cluster to obtain highly adaptable rectangular spaces based on actual needs, reducing space waste and improving area utilization. Constraint programming is used for non-overlapping space assignment, ensuring the compactness and conflict-free nature of the cluster-level layout. Connectivity enumeration ensures internal accessibility within functional clusters, preventing isolated areas. In the room subdivision stage within clusters, allocation is based on full coverage and column grid constraints, ensuring that the room layout conforms to the column grid structure while maximizing space utilization. The intelligent generation and placement of service space candidate locations optimizes the distribution of auxiliary functions, reducing circulation intersections and spatial fragmentation. The global scoring mechanism, combining multi-dimensional assessments of circulation distance, column conflicts, and morphological quality, can accurately locate problem areas in the layout. Through iterative optimization of local release and re-solution, it gradually corrects defects and ultimately outputs a highly adaptable layout solution. This method achieves hierarchical and automated layout from macro-clusters to micro-rooms, significantly shortening the design cycle and reducing manual adjustment costs. It is particularly suitable for the renovation projects of protected buildings with complex spatial forms and stringent constraints, providing scientific and efficient technical support for the functional regeneration of historical buildings. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating an experimental functional space layout method for constrained optimization-based renovation of protected buildings.

[0049] Figure 2 A schematic diagram illustrating the process of generating candidate combinations for dividing atomic regions of a building plan and assigning functional groups. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0052] Figure 1 This is a flowchart illustrating the experimental functional space layout method for the renovation of protected buildings based on constraint optimization, according to an embodiment of the present invention.

[0053] Experimental functional space layout methods for the renewal of protected buildings based on constraint optimization include:

[0054] The acquired target building plan data is discretized by grid to generate a basic cutting line set. From this set, the shape turning line, the main corridor edge line and the column grid axis are selected as priority cutting lines. The weighted priority cutting line set is sorted by weight and used to perform constraint perception discretization on the building plan.

[0055] Functional requirements are analyzed on the discretized building plan. The functional requirements are grouped into functional groups according to the group labels and a candidate rectangular size pool is generated for each functional group. The building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions and the atomic region combinations that satisfy connectivity are enumerated for each functional group. Based on constraint planning, non-overlapping spatial assignment is performed on each functional group to generate a group-level layout scheme.

[0056] Based on the aforementioned cluster-level layout scheme, room-level space allocation is performed within the division area of ​​each functional cluster based on full coverage and column grid accommodation constraints to complete the subdivision of rooms within the cluster. Service space candidate positions are generated near obstacle neighborhoods and corridor intersection nodes and placed into service spaces to obtain a preliminary layout scheme.

[0057] The preliminary layout scheme is scored globally, and problem areas are identified based on circulation distance, column conflicts, and morphological quality. The problem areas are then locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

[0058] In one optional implementation, shape transition lines, main corridor edge lines, and column grid axes are selected from the basic cutting line set as priority cutting lines and sorted by weight to form a weighted priority cutting line set, including:

[0059] Extract the coordinates of each vertex of the building's outer contour and calculate the turning angle of the adjacent side at each vertex to determine the concavity / convexity attribute. Locate all concave vertices and generate candidate segmentation lines that pass through the concave vertices and are parallel to the adjacent contour sides. Identify the candidate segmentation lines that can completely divide the concave area into independent rectangular areas as shape transition lines and assign them the highest weight.

[0060] Extract continuous line segments that are consistent with the direction of the corridor along both sides of the main corridor, merge the continuous line segments into the edge line of the main corridor and assign them the second highest weight;

[0061] The column direction axis is formed by sequentially connecting the center points of the same column along each column direction and sequentially connecting the center points of the same row along each row direction. The column direction axis and the row direction axis are then merged into the column grid axis and assigned a third weight.

[0062] The morphological transition line is inserted into the basic cutting line set with the highest priority, the main corridor edge line is inserted into the basic cutting line set with the second highest priority, and the column grid axis is inserted into the basic cutting line set with the third priority. This allows the high-weight priority cutting line to obtain priority division weight during the discretization process. The three types of priority cutting lines are then added to the basic cutting line set in a weighted overlapping manner to form the weighted priority cutting line set.

[0063] Extracting the vertex coordinates of the building's outer contour is the starting point for generating shape transition lines. Vertex traversal is performed on the outer contour polygon of the target building plan, sequentially obtaining each vertex. Two-dimensional coordinates And calculate the interior angle formed by the two adjacent sides at that vertex. Specifically, with vertices Centered on, take its predecessor vertex and successor vertex Construct vectors respectively and The angle between the two vectors is calculated using the inverse cosine function, thus obtaining the turning angle at the vertex. .when When, the vertex is determined to be a convex vertex; when When this occurs, the vertex is determined to be a concave vertex. The existence of a concave vertex means that the building plane has undergone concave deformation at that location, which is a direct indication of irregular planar shape.

[0064] For all locations identified as concave vertices, candidate segmentation lines are generated that pass through the vertex and are parallel to its adjacent contour edges. Since there are two adjacent contour edges, each concave vertex can generate at most two candidate segmentation lines with different directions. The extension range of the candidate segmentation lines is bounded by the bounding box of the building plan, ensuring that the line segments can completely traverse the planar region. Among all candidate segmentation lines, those that can completely cut the concave region into independent rectangular sub-regions—that is, each sub-region after cutting satisfies the condition that its four sides are parallel to the coordinate axes and that it has no concave corners—are identified as morphological transition lines. Morphological transition lines directly determine whether the building plan can be regularly decomposed into rectangular atomic regions, fundamentally affecting the quality of subsequent discretization; therefore, they are assigned the highest weight. .

[0065] Extracting the edge lines of the main corridor relies on the geometric recognition of the corridor space. In the functional semantic annotation of the building plan, the corridor area typically exists as a slender rectangle or a polygonal line, with its width significantly smaller than its length. Line segment sequences consistent with the corridor's direction are extracted along both sides of the main corridor's boundary. The corridor's direction is determined by the major axis of the smallest bounding rectangle of the corridor area. The extracted line segment sequences are merged according to spatial continuity: if the projection distance between two adjacent line segments in the direction is less than a set merging threshold... Furthermore, the offset of the two line segments in the normal direction is less than the tolerance. Then, they are merged into a single continuous line segment. The merged line segment becomes the edge line of the main corridor and is assigned a secondary weight. ,in During the discretization process, the integrity of the corridor space is prioritized to prevent room dividing lines from crossing the corridor boundary and causing the corridor space to break up.

[0066] The generation of the column grid lines is based on the building structure column position data. The center point coordinates of all columns in the building plan are clustered according to the column direction of the coordinates (usually 1000-10000). (coordinates) and row direction (usually) Perform one-dimensional clustering on each coordinate, and cluster the columns whose coordinate differences are less than the column grid tolerance. The columns are grouped into the same column or row. For the center point of all columns in the same column, according to... Connect the coordinates sequentially to generate column direction axes; for all column center points in the same row, according to... Connecting the coordinates sequentially generates row axes. The column axes, together with the row axes, form the column grid axis set. The column grid axes reflect the geometric laws of the building structure system. Spatial division along the column grid axes ensures that room boundaries are aligned with structural columns, preventing columns from appearing inside rooms and causing usability obstacles. The column grid axes are assigned a third weight. ,in .

[0067] Three priority cutting lines are inserted into the basic cutting line set in order of their weight hierarchy. The basic cutting line set consists of all candidate cutting lines generated during the uniform grid discretization of the building plan, covering all possible horizontal and vertical division positions within the plan. Shape transition lines are inserted with the highest priority, meaning that during the discretization process, any cutting operation at the location of a shape transition line must be executed first and cannot be replaced or skipped by other cutting lines. The main corridor edge line is inserted with the second highest priority to ensure that the corridor boundary is accurately preserved in the discretization result. The column grid axis is inserted with the third highest priority to ensure that the structural axis is reflected in the spatial division. After the three priority cutting lines are superimposed into the basic cutting line set, a weighted priority cutting line set is formed.

[0068] The weighted priority cutting lines are clustered, each carrying its corresponding weight value. In the subsequent constraint-aware discretization process, the cutting order is arranged from highest to lowest weight, with higher-weighted cutting lines receiving priority. When two cutting lines conflict or are mutually exclusive in space, the lower-weighted cutting line yields to the higher-weighted one, ensuring that the integrity of morphological transition lines and corridor boundaries remains undisturbed. This mechanism enables the discretization results to achieve high quality in three dimensions: geometric regularity, corridor spatial integrity, and structural axis alignment, providing a reliable geometric basis for subsequent atomic region division of functional clusters and room-level spatial allocation.

[0069] In real-world engineering scenarios, protected buildings often have irregular planar shapes, numerous concave vertices, complex corridor layouts, and uneven column grid spacing. By employing the identification and weighting mechanism of the three priority cutting lines mentioned above, the structural and spatial semantics of the building plan can be effectively encoded into the discretization process. This ensures that the final generated atomic regions are as regular in shape as possible and aligned with existing building elements on their boundaries, thereby reducing the complexity of subsequent constraint planning solutions and improving the feasibility and quality of the layout scheme. (Merging Threshold) and tolerance The value can be adaptively adjusted according to the corridor width and column grid accuracy of the specific building to adapt to building plan data of different scales and accuracy.

[0070] In one alternative implementation, candidate segmentation lines capable of completely dividing the concave region into independent rectangular regions are identified as shape transition lines, including:

[0071] Extract all vertices of the building's planar outline and number each vertex in a clockwise direction, recording the coordinates of each vertex. For each vertex, calculate the cross product sign formed by the previous outline edge vector and the next outline edge vector, and determine whether the current vertex is a convex vertex or a concave vertex based on the sign of the cross product sign.

[0072] Traverse all concave vertices and, starting from the current concave vertex, draw rays along the extension directions of the two contour edges adjacent to that concave vertex. The line connecting the contour point where each ray first intersects the outer contour of the building plane and the concave vertex is taken as the candidate dividing line.

[0073] The building plan is divided into two sub-regions by each candidate dividing line. The rectangular integrity of each sub-region is calculated as the ratio of the area of ​​the sub-region to the area of ​​its smallest bounding rectangle. The weighted average of the rectangular integrity of the two sub-regions is then taken. At the same time, the ratio of the sum of the areas of the two sub-regions to the total area of ​​the building plan is calculated as the area utilization rate. The weighted average of the rectangular integrity and the area utilization rate are used as a comprehensive index to evaluate the dividing effect of the candidate dividing line.

[0074] Sort all candidate dividing lines according to the comprehensive index from high to low, and determine the candidate dividing line corresponding to the maximum value of the comprehensive index as the morphological turning line.

[0075] Before performing constrained perceptual discretization on the target building's plan, it is necessary to accurately identify candidate dividing lines, i.e., morphological transition lines, that can completely divide the concave area into independent rectangular areas. The plan shape of protected buildings often presents an irregular concave outline due to historical evolution. Such outlines cannot be directly covered by rectangular atomic areas. It is necessary to introduce dividing lines to decompose the concave area into several rectangular sub-areas in order to provide a regular operational basis for the subsequent functional space layout.

[0076] Extract all vertices of the building's plan outline and number them sequentially in a clockwise direction. Let the vertices be... The coordinates of the vertices are ,in The x-axis is... The vertical axis is numbered from... to , This represents the total number of vertices on the outer contour. For each vertex... Construct a vector from the previous contour edge to that vertex. and the vector pointing from that vertex to the next contour edge. Calculate the cross product of these two vectors. ,Right now Under the agreed-upon clockwise direction, if Then the vertex It is a convex vertex; if Then the vertex If it is a concave vertex; Then the vertex These are collinear points and do not participate in the generation of subsequent dividing lines. By calculating the cross product sign for each vertex, the complete set of concave vertices on the outer contour can be marked. These concave vertices are the root cause of irregular shapes.

[0077] Iterate through all vertices marked as concave, and for each concave vertex, draw a ray along the extension directions of its two adjacent contour edges. Specifically, for a concave vertex... The direction vector of its forward adjacent edge is Along this direction from the vertex The ray originates from the point of origin; its direction vector towards the adjacent side is... Similarly, from the vertex Draw a ray along this direction. Intersect each ray with every edge of the building's outer contour. Take the first intersection point along the ray direction that is closest to the concave vertex as the ray's endpoint, denoted as . and These correspond to the forward and backward rays, respectively. The concave vertex... and The connection lines and concave vertices and Each line connecting the points is added to the candidate set as a candidate dividing line. If a ray fails to find a valid intersection point within the outer contour of the building plane (e.g., the ray direction is parallel to all contour edges or extends outside the plane), then a candidate dividing line in that direction is not generated. After the above traversal, at most two candidate dividing lines can be generated for each concave vertex, and all candidate dividing lines together constitute the candidate set to be evaluated.

[0078] For each candidate dividing line in the candidate set, the building plan is divided into two sub-regions using that dividing line as the boundary. and Calculate sub-regions area and the area of ​​its minimum circumscribed rectangle Define subregions The rectangle completeness is Similarly, calculate sub-regions Rectangular completeness ,in sub-region area, sub-region The area of ​​the minimum bounding rectangle. The range of values ​​for rectangle integrity is... The closer the value is This indicates that the closer the sub-region is to a rectangle, the more suitable it is for subsequent tiling operations on rectangular atomic regions. A weighted average of the rectangular integrity of the two sub-regions is taken, with the weights determined according to the proportion of each sub-region's area to the sum of the areas of the two sub-regions. That is, the weighted average rectangular integrity is... Using area proportion as a weight can avoid the distortion of evaluation results due to the extremely small area of ​​a certain sub-region.

[0079] At the same time, calculate the sum of the areas of the two sub-regions. Total floor area of ​​the building The ratio of this ratio to the area utilization rate of the candidate dividing line. Area utilization rate reflects whether there are any issues with area omissions or overlapping calculations during the geometric division of candidate dividing lines. Ideally, Should approach This indicates that the dividing line completely divides the building plan into two sub-regions without omission or overlap. For some candidate dividing lines that traverse obstacle areas within the building plan, their area utilization rate is lower than... This indicates the existence of residual regions that cannot be covered by two sub-regions, and such dividing lines will be penalized in the evaluation.

[0080] Weighted average rectangle completeness With area utilization rate By performing a weighted summation, a comprehensive index for the candidate dividing line is obtained. ,in and These are the weighting coefficients for rectangular integrity and area utilization, respectively, satisfying... , , In practical applications, for buildings with complex floor plans and deep concave shapes, the efficiency can be appropriately increased. The value of makes the evaluation results focus more on the degree of rectangularity of the sub-region; while for cases where the planar shape is relatively regular and the main focus is on the integrity of the segmentation, the value can be appropriately increased. The value of . Comprehensive index The range of values ​​is The larger the value, the better the segmentation effect of the candidate dividing line.

[0081] All candidate dividing lines in the candidate set are determined according to comprehensive indicators. Sort the candidates by comprehensive index from highest to lowest, and select the candidate dividing line corresponding to the maximum value of the comprehensive index as the final determined shape turning line. This shape turning line is assigned the highest weight in the subsequent construction of the weighted priority cutting line set. Prioritize participation in the constraint perception discretization of the building plan to ensure that concave areas can be preferentially divided into regular rectangular sub-regions, providing a reliable geometric basis for the atomic region division of functional groups and the generation of candidate rectangular size pools. If multiple candidate dividing lines have the same comprehensive index and are all at the maximum value, then the shortest candidate dividing line is selected according to its length from shortest to longest to reduce unnecessary cutting of the internal space of the building plan and protect the integrity of the original spatial pattern of the building.

[0082] In one alternative implementation, the building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions, and combinations of atomic regions satisfying connectivity are enumerated for each functional cluster, including:

[0083] Using the weighted priority cutting line set as the dividing boundary, the discretized building plane is successively divided along the extension direction of each priority cutting line, so that the area between two adjacent priority cutting lines is divided into rectangular grid units. Each rectangular grid unit is defined as an atomic region and all atomic regions are non-overlapping and completely cover the entire building plane.

[0084] An adjacency graph of atomic regions is established with each atomic region as the basic operation unit. Each node in the graph represents an atomic region and is connected to the nodes corresponding to adjacent atomic regions by an edge. The length of the shared boundary between the two atomic regions connected by each edge is recorded as the adjacency weight.

[0085] Based on the area requirements of each functional group, the adjacency graph of the atomic regions is used as the search space. Starting from the initial atomic region corresponding to the location of each functional group, all connected subgraphs including the initial atomic region are enumerated by breadth-first traversal. The set of atomic regions contained in each connected subgraph is taken as a connected region combination. Connected region combinations whose absolute difference between the total area of ​​the connected region combination and the target area of ​​the corresponding functional group is within the preset tolerance range are selected as candidate assignment combinations for the functional group.

[0086] After constructing the weighted priority cutting line set, the discretized building plan needs to be further decomposed into the smallest rectangular units that are finest-grained, non-overlapping, and capable of completely covering the entire plan, to support subsequent functional space assignment. Specifically, the building plan is divided sequentially along its extension direction, using each priority cutting line in the weighted priority cutting line set as the dividing boundary. The division order is from highest to lowest weight: first along the shape transition line, then along the main corridor edge line, and finally along the column grid axis. Each division operation bisects the currently incompletely divided area along the extension direction of the corresponding cutting line, until the area between two adjacent priority cutting lines in the same direction is divided into rectangular grid units. Since the priority cutting lines themselves originate from shape transitions, corridor edges, and column grid axes, their extension direction is naturally aligned with the building structure direction. Therefore, each rectangular grid unit obtained from the division maintains a consistent geometric shape with the building structure, without producing oblique or irregular residual areas. Each rectangular grid cell is defined as an atomic region. All atomic regions do not overlap with each other and completely cover the entire building plane in space, leaving no unassigned blank areas.

[0087] After dividing the atomic regions, it is necessary to establish the adjacency relationships between them to support subsequent enumeration of connected regions. An adjacency graph is constructed using each atomic region as a node. Any two atomic regions sharing a spatial boundary are connected by an edge; that is, two atomic regions are considered adjacent only if they share a common line segment geometrically (not just corner points). For each edge in the graph, the length of the line segment representing the shared boundary between the two connected atomic regions is recorded as the adjacency weight of that edge. The length of the shared boundary is denoted as... ,in and These are the numbers of two adjacent atomic regions. A longer shared boundary indicates a closer spatial connection between the two atomic regions. In subsequent combination enumerations, atomic regions with longer shared boundaries can be preferentially included in the same functional group to ensure the integrity and accessibility of the internal space of the functional group. Adjacency weight. It is also used in the subsequent generation stage of corridor intersection nodes and service space candidate sites, as a basis for judging the strength of spatial connection.

[0088] After the adjacency graph of atomic regions is established, candidate assignment combinations are enumerated for each functional group. Each functional group has a pre-resolved target area requirement, denoted as the... The target area for each functional cluster is The preset tolerance is The enumeration process uses breadth-first search as its core search strategy. Starting from the initial atomic region corresponding to each functional group, it gradually expands through the adjacency graph of atomic regions, enumerating all connected subgraphs that include the initial atomic region. The selection of the initial atomic region is determined based on the positional constraints of the functional group. For example, functional groups that need to be adjacent to the outer wall preferentially select atomic regions located at the edge of the building plan as the starting node, while functional groups that need to be located inside select atomic regions far from the outer wall as the starting node.

[0089] The breadth-first traversal process is as follows: The initial atomic region is added to the initial candidate set. Using this candidate set as the root node, the traversal expands layer by layer. During each expansion, all unincluded adjacent atomic regions of all nodes in the current candidate set are added sequentially, forming a new candidate set. After each expansion, the total area of ​​the atomic regions covered by the current candidate set is calculated and denoted as . Its calculation method is the sum of the areas of all atomic regions in the candidate set. When satisfy When the current candidate set contains a set of atomic regions, it is recorded as a valid connected region combination and added to the candidate assignment combination list of that functional group. If the area of ​​the current expansion path exceeds the allowed range, further expansion on that path is terminated to avoid invalid searches. Through this mechanism, the enumeration process can efficiently find all combinations of connected regions that satisfy the area constraints within a limited search space, avoiding redundant calculations for combinations with obviously inconsistent areas.

[0090] In real-world engineering scenarios, protected building plans often contain irregular areas, such as recessed areas or irregularly shaped boundaries resulting from the preservation of historical components. For these areas, the resulting atomic regions vary significantly in size, leading to a substantial increase in the number of expansion layers and branches in the breadth-first traversal. To control the enumeration scale, an upper limit is set on the number of candidate assignment combinations for each functional group. When the number of recorded candidate assignment combinations reaches When doing so, prioritize retaining shared boundary weights. Larger connected region combinations are selected by prioritizing candidate combinations with stronger internal connectivity and discarding those with weaker internal connectivity, thereby controlling the enumeration scale while ensuring solution quality.

[0091] After the candidate assignment combination list is generated, each functional group holds a set of candidate atomic regions that satisfy area constraints and ensure internal connectivity. These candidate sets will serve as input to the constraint programming solver for subsequent group-level non-overlapping spatial assignment. Constraint programming must ensure that there is no overlap of atomic regions between the candidate assignment combinations selected by different functional groups; that is, the intersection of the atomic region sets contained in the final assignment combinations of any two functional groups is an empty set. This constraint, together with the connectivity constraint of the candidate assignment combinations, constitutes the core constraint system for solving group-level layout schemes, ensuring that the final generated group-level layout schemes are spatially non-overlapping and non-omitted, and that the assigned regions of each functional group remain topologically connected.

[0092] Area tolerance of candidate assignment combinations The setup is typically adjusted dynamically based on the type and size of the functional clusters. For functional clusters such as experimental operation areas, which require high area accuracy, Choose the smaller value to ensure a strict match between the space area and the experimental equipment layout requirements; for functional groups such as auxiliary storage areas where area requirements are relatively relaxed, ... The constraints can be appropriately relaxed to increase the number of candidate combinations and improve the flexibility of solving constraint programming. This can be achieved by... The differentiated settings can ensure the accuracy of the area of ​​key functional spaces while providing sufficient search freedom for the generation of the overall layout scheme, thereby obtaining a higher quality cluster-level layout scheme in the constrained planning stage.

[0093] In one optional implementation, non-overlapping spatial assignments are performed on each of the functional clusters based on constraint planning, including:

[0094] Using each candidate assignment combination of each functional group as a Boolean decision variable, construct a non-overlapping constraint that each atomic region is occupied by at most one functional group and a mutual exclusion constraint that each functional group has one and only one candidate assignment combination selected, and construct a cardinality constraint that the number of atomic regions contained in the selected candidate assignment combination must be within a preset range.

[0095] The absolute value of the difference between the actual total area of ​​the selected candidate assignment combination and the target area of ​​the corresponding functional group is used as the area deviation penalty term, and the difference between the total length of the boundary line of the selected candidate assignment combination and the perimeter of the square with the same area is used as the morphological density reward term. The weighted sum of the area deviation penalty term and the morphological density reward term is used as the objective function.

[0096] Using the non-overlapping constraints, the mutual exclusion constraints, and the cardinal constraints as constraints, and minimizing the objective function as the optimization direction, the values ​​of each Boolean decision variable are solved through constraint programming to minimize the objective function. The set of atomic regions corresponding to the selected candidate assignment combinations in each functional group is taken as the occupied area of ​​the functional group on the building plan. The set of occupied areas of all functional groups constitutes the cluster-level layout scheme.

[0097] like Figure 2 As shown, the method includes:

[0098] After enumerating the candidate assignment combinations for functional clusters, it is necessary to use constraint programming to allocate each functional cluster to non-overlapping atomic regions on the building plan, thereby forming a cluster-level layout scheme. To this end, a Boolean decision variable is introduced for each candidate assignment combination of each functional cluster: if a candidate assignment combination is selected, the corresponding Boolean decision variable takes the value 1; otherwise, it takes the value 0. Let the... The functional groups were enumerated to obtain There are candidate assignment combinations, corresponding to Boolean decision variables denoted as . ,in Index of candidate assignment combinations. The Boolean decision variables of all functional groups constitute a complete decision space. The constraint programming solver will search in this space for the combination of variable values ​​that satisfies all constraints and minimizes the objective function.

[0099] The non-overlapping constraint requires that each atomic region in the building plan can be occupied by at most one functional group, meaning that two or more functional groups are not allowed to be assigned to the same atomic region simultaneously. For any atomic region Iterate through all functional groups and their candidate assignment combinations, and collect all regions containing atomic regions. The Boolean decision variables corresponding to the candidate assignment combinations are summed, and the summation value is required to be no more than 1. This constraint is mathematically expressed as: for each atomic region All candidate assignment combinations that satisfy the condition "contain atomic regions" "Conditional The sum is less than or equal to 1. This constraint ensures the physical feasibility of the cluster-level layout scheme in space and prevents overlapping and occupancy of space.

[0100] The mutual exclusion constraint requires that each functional group has exactly one candidate assignment combination selected, meaning that the sum of the Boolean decision variables corresponding to all candidate assignment combinations within the same functional group is exactly equal to 1. For the Each functional group has a mutual exclusion constraint expressed as follows: This constraint ensures that each functional cluster occupies a defined area on the building plan, preventing situations where a functional cluster is left unallocated or where the same functional cluster occupies multiple different areas simultaneously. Mutually exclusive constraints and non-overlapping constraints together form the basis for the feasibility of the constrained planning model.

[0101] Constraint programming solutions employ the following three types of constraints: The mathematical expression for the unique choice constraint for each cluster is: Where k is the cluster number and j is the candidate number for that cluster. The non-overlapping constraint for candidate clusters is expressed in grid cells as follows: This constraint holds for all atomic regions 'a', ensuring that any grid cell is covered by at most one cluster candidate. The upper and lower bound constraints on the number of connected partitions are expressed as: , where (l_r) and (u_r) are the lower and upper bounds of the number of clusters in the connected partition r, respectively.

[0102] In the cluster spatial layout stage, in order to sort and filter the candidate options in the connectivity candidate set, an internal cost function for each candidate option is defined:

[0103]

[0104] in, Let S be the area of ​​candidate partition S. For the target area of ​​group k, Let S be the perimeter of the candidate partition. The cost function consists of two terms: the first term penalizes the deviation between the area of ​​the candidate partition and the target area; the second term encourages the candidate partition to become more compact by providing an isotropic reward (i.e., an amplified area-to-perimeter ratio).

[0105] The cardinality constraint further restricts the number of atomic regions contained in the selected candidate assignment combination to within a preset range. Specifically, for the first... Each functional cluster pre-defines a lower bound on the number of atomic regions based on its target area requirements and the average area of ​​atomic regions in the building plan. and the Upper Realm For the first The first functional group Let there be n candidate assignment combinations, and let the number of atomic regions they contain be n. The cardinality constraint requires that the selected candidate assignment combination satisfy... In a constrained programming model, this constraint is equivalent to: for all conditions satisfying... or The candidate assignment combinations are forced to have their corresponding Boolean decision variables. This allows for the direct elimination of non-compliant candidate combinations during the search phase, effectively compressing the solution space and improving solution efficiency.

[0106] The objective function is composed of a weighted average of an area deviation penalty term and a morphological density reward term. The area deviation penalty term measures the degree of deviation between the actual total area of ​​the selected candidate assignment combination and the target area of ​​the corresponding functional group: Let the first term be... The target area requirement for each functional cluster is: , No. The actual total area of ​​the candidate assignment combinations is Then the area deviation penalty term for Since the absolute value term needs to be linearized in mixed-integer programming, an auxiliary variable can be introduced. Make and This transforms the area deviation penalty term into an equivalent linear form for solving the problem.

[0107] The shape density bonus measures whether the shape of the selected candidate assignment combination is close to a square, thereby encouraging functional groups to occupy areas with regular shapes that are easy to subdivide in the room. The specific calculation method is as follows: for the first... The first functional group The total length of the union boundary of the atomic regions covered by each of the candidate assignment combinations is denoted as . The difference between the perimeter of a square and the perimeter of a square with the same area is denoted as . Since a square has the largest area when its perimeter is fixed (equivalently, its perimeter is the smallest when its area is fixed),... A smaller value indicates a denser pattern. Pattern Density Bonus Item for The smaller the value, the more compact and regular the overall layout.

[0108] The area deviation penalty and the morphological compactness reward are weighted by coefficients. and We perform a weighted summation to obtain the objective function. ,in The extent to which control area deviation affects the optimization direction. The degree to which the density of the control form affects the optimization direction. In the scenario of renovating preserved buildings, the functional area requirements are usually subject to strong rigid constraints. The value is generally higher than This ensures that area matching accuracy takes precedence over morphological compactness. Both weighting coefficients can be adjusted during implementation based on specific project requirements.

[0109] Using non-overlapping constraints, mutual exclusion constraints, and cardinal constraints as constraints, and with the objective function... Minimization is the optimization direction; the constrained programming solver is invoked to apply the solution to all Boolean decision variables. The solution is then performed. During the solution process, the constraint programming solver continuously compresses the variable domain using constraint propagation techniques: whenever a certain constraint propagation technique is applied... When the value is set to 1, the Boolean decision variables corresponding to all other candidate assignment combinations that share the atomic region with this candidate assignment combination are immediately forced to be set to 0, thereby triggering further constraint propagation and significantly reducing the remaining search space. For large building plans, a solution time limit can be set. If the optimal solution cannot be found within the time limit, the currently known optimal feasible solution is returned as an approximate result.

[0110] After the solution is completed, for each functional group Find its corresponding selected candidate assignment combination (i.e. The combination of functional groups (the one containing the given elements) is used to extract all atomic regions contained within the combination. The collection of these atomic regions is then used as the area occupied by that functional group on the building plan. The occupied areas of all functional groups are summarized to obtain a group-level layout scheme. This scheme describes the spatial division of each functional group on the building plan at the granularity of atomic regions, ensuring no spatial overlap between groups and that the occupied area of ​​each functional group meets the requirements of connectivity, area matching, and form compactness. This provides a reliable spatial foundation for subsequent room-level space allocation and service space placement.

[0111] In one alternative implementation, room-level space allocation is performed within the defined areas of each functional cluster, and service spaces are placed near obstacle neighborhoods and corridor intersections, including:

[0112] Within the division area of ​​each functional group, the area requirements and rectangular shape requirements of each room are used as input parameters. The hard constraint is full coverage, that is, the sum of the rectangular areas of each room must be equal to the area of ​​the division area of ​​the functional group. The spatial constraint is column grid accommodation, that is, when the minimum area of ​​a room is greater than three times the area of ​​a column grid unit, the room is allowed to accommodate columns. By enumerating the candidate rectangular size combinations of each room and verifying whether each combination meets the full coverage constraint and column grid accommodation constraint, the combination that meets all the constraints is taken as the room-level subdivision scheme.

[0113] The union of the preset neighborhood range of each obstacle and the preset neighborhood range of the corridor intersection node is taken as the service space candidate area. Multiple candidate placement positions are generated in the candidate area at preset intervals, and the target placement position is determined by the minimum distance between each candidate placement position and the boundary of the adjacent assigned room as the evaluation index.

[0114] After placing the service space at the target placement location, it is detected whether there is an overlap or gap between the boundary of the service space and the boundary of the allocated room. If there is an overlap or gap, the boundaries of the remaining allocated rooms are kept unchanged, and the functional groups involved in the overlapping or gap area are reassigned to room-level spaces separately.

[0115] After finalizing the cluster-level layout plan, the designated areas for each functional cluster need to be further refined into specific room units, while avoiding fixed and buffer zones. For functional clusters that contain or are adjacent to protected elements, the fixed protection zone is first deducted from the designated area of ​​that functional cluster, and the buffer zone is marked as a restricted layout area. Then, room-level space allocation is performed within the remaining allocable area. Room-level space allocation uses the area requirements and rectangular shape requirements of each room as input parameters. The area requirements are derived from the functional requirements analysis phase, specifying a minimum usable area for each type of room. The rectangular shape requirement stipulates that the room plan must be rectangular to ensure space usability and ease of subsequent renovation and modification. In the scenario of renovating protected buildings, the rectangular room shape also helps to avoid conflicts with the original structural system.

[0116] Before room-level space allocation, the protected element areas in the target building plan are explicitly locked. Typically, protected elements include non-removable walls, historical components, special structural parts, and other architectural relics that need to be preserved in situ. Protected fixed zones are generated based on the geometric boundaries of the protected elements, and protected buffer zones can be set around them according to preset requirements. The protected fixed zones are locked areas that cannot be occupied during subsequent layout solving; their boundaries, locations, and areas are not included in optimization adjustments. Protected buffer zones are not used to arrange any laboratory rooms or equipment rooms; they can only be used for corridor functions.

[0117] The full coverage constraint, a hard constraint for room-level subdivision, requires that the sum of the rectangular areas of all rooms within the same functional cluster division area equals the total area of ​​that functional cluster division area. This means that no blank areas are allowed that are not covered by any room, and rooms are not allowed to overlap. When enumerating candidate rectangular size combinations, several candidate width and length pairing schemes are pre-generated for each room. The generation of candidate sizes is determined based on the room area requirements and the unit size defined by the building discretization. The enumeration process fills each room sequentially from left to right and from top to bottom. For each candidate size combination, it is verified whether the current combination can completely cover the entire area without exceeding the boundary of the functional cluster division area. If a candidate combination has unfillable remaining fragmented areas during the assembly process, the combination is eliminated; if all room candidate size combinations can be seamlessly assembled and the total area is exactly equal to the area of ​​the functional cluster division area, the combination passes the full coverage constraint verification.

[0118] Column grid accommodation constraints, as spatial constraints, are used to control the relationship between rooms and structural columns. When the minimum area of ​​a room is greater than three times the area of ​​a column grid unit, the room is allowed to accommodate columns, meaning columns can fall inside the room without having to be on the room boundary. When the room area is small and does not meet the above conditions, the room layout is required to place columns as far as possible on the room boundary or corners to avoid columns appearing in the center of the room's usable space and affecting its functionality. When validating candidate size combinations, the coordinates of columns within the coverage area of ​​each candidate rectangular room region are extracted, the number of column grid units falling inside the room is counted, and the ratio of the room area to the column grid unit area is used to determine whether the column grid accommodation constraint is satisfied. Candidate size combinations that simultaneously satisfy both the full coverage constraint and the column grid accommodation constraint are recorded as valid room-level subdivision schemes. If multiple valid schemes exist, the optimal scheme is selected based on the criterion of minimizing the total deviation between the room area and the target area.

[0119] The placement of service spaces is conducted using a candidate location generation and evaluation method. Service spaces typically include auxiliary functional units such as restrooms, storage rooms, and stairwell anterooms. Their location is closely related to the distribution of obstacles (such as structural walls, equipment shafts, and fire pipes) and the location of corridor intersections. The union of the preset neighborhood range of each obstacle and the preset neighborhood range of the corridor intersection is used to obtain the candidate service space area. The preset neighborhood range is usually determined according to building design codes and the usage habits of experimental functional spaces. For example, a buffer zone extending 1.5m to 2.0m outward from the outer contour of the obstacle is used as the neighborhood range, and a circular area with a radius of 2.0m to 3.0m centered on the corridor intersection is used as the node's neighborhood range.

[0120] Within the candidate service space area, multiple candidate placement locations are generated at preset intervals. For each candidate placement location, the minimum distance between that location and the boundary of the adjacent assigned room is calculated, and this minimum distance is used as an evaluation metric. The smaller the minimum distance, the closer the connection between the service space and the existing room, which helps reduce unnecessary corridor area and improve the overall layout compactness. The candidate placement location with the optimal evaluation metric is determined as the target placement location, and the service space is placed at that location. The rectangular dimensions of the service space are preset according to its functional type and regulatory requirements; for example, the minimum size of a bathroom is usually no less than 1.8m × 2.0m, and the minimum size of a storage room is usually no less than 1.5m × 1.5m.

[0121] After the service space is placed, the relationship between the service space boundary and the boundary of the assigned room needs to be checked. If the service space boundary overlaps with the boundary of an assigned room, it means that the service space occupies part of the original room area, and the affected area needs to be corrected. If there is a gap, it means that there is an uncovered blank area between the service space and the adjacent room, which also needs to be corrected. The detection method uses rectangular area intersection calculation. The intersection area of ​​the service space rectangle and each adjacent assigned room rectangle is calculated one by one. If the intersection area is greater than a preset threshold, it is judged as an overlap. If the gap width between the service space and the adjacent room is greater than a preset minimum gap threshold, it is judged as a gap.

[0122] When overlaps or gaps are detected, the boundaries of the remaining unaffected allocated rooms are preserved, and only the functional groups involved in the overlapping or gap areas are reassigned at the room level. During reassignment, the areas occupied by service spaces are deducted or supplemented from the functional group's division area to obtain the corrected effective area of ​​the functional group. Then, the candidate rectangle size enumeration and constraint verification process is re-executed using the corrected effective area as input until all rooms in the functional group meet the full coverage constraint and column grid accommodation constraint. This local reassignment mechanism avoids global re-layout caused by the placement of service spaces, effectively reducing computational overhead, while ensuring the spatial integrity and seamless coverage within the functional group's division area in the final layout scheme. After the above room-level space allocation and service space placement process, a preliminary layout scheme containing all room units and service space units is formed, providing a foundation for subsequent global scoring and iterative optimization.

[0123] In one optional implementation, the preliminary layout scheme is globally scored, and problem areas are identified for local release and re-solution, including:

[0124] The sum of the shortest path lengths between each functional group and between each functional group and the entrance / exit is used as the circulation distance score; the cumulative value of the number of columns contained in a single room in all rooms is used as the column conflict score; the arithmetic mean of the ratio of the area of ​​each room to the area of ​​its smallest bounding rectangle is used as the morphological quality score; and the weighted sum of the above three scores is used as the global score.

[0125] The contribution of each functional group to the circulation distance score and morphological quality score in the global score is decomposed from the global score. Functional groups with a contribution exceeding a preset threshold are marked as problem areas and the room subdivision results in the problem areas are released. The room subdivision results of functional groups in non-problem areas are retained as locked layouts.

[0126] The problem area after release is taken as the solution space. The locked layout is taken as an immutable boundary condition. The solution space is redistributed at the room level based on the full coverage partitioning constraint and the column grid accommodation constraint. The acceptance criterion is whether to retain the current rediscovery result based on whether the global score after rediscovery is improved compared with the previous round of global score. The iterative process of global score, problem area identification and local rediscovery is repeated until the global score is not improved in multiple consecutive rounds of iteration. Then the iteration is terminated and the final layout scheme is output.

[0127] The circulation distance score reflects the ease of access between functional clusters and between functional clusters and building entrances / exits. Specifically, a circulation map is constructed on the layout plan, where nodes correspond to the geometric centroid and entrance / exit locations of each functional cluster, and edge weights are the Euclidean distances between adjacent accessible areas. A shortest path search is performed on each pair of nodes, and the total circulation distance is obtained by summing the shortest path lengths between all node pairs. . The smaller the value, the more compact the movement between functional groups and between them and entrances / exits, resulting in higher traffic efficiency.

[0128] Column conflict score measures the degree of interference between room divisions and the existing column grid. For each room in the layout, the number of columns contained within its interior (i.e., the area enclosed by the room, excluding the boundaries) is counted; the total number of columns in all rooms is then summed to obtain the total column conflict count. . The larger the value, the more serious the conflict between room division and column grid, which will affect structural safety and indoor usable space. Therefore, it is included as a penalty item in the overall score.

[0129] The morphological quality score reflects the regularity of the floor plan shape of each room. For each room, the ratio of its actual area to the area of ​​its smallest bounding rectangle is calculated; this ratio is called the room's rectangular integrity. The morphological quality score is obtained by taking the arithmetic mean of the rectangular integrity scores of all rooms. . The closer the value is to 1, the more regular the shape of each room is, which is more conducive to furniture arrangement and usage efficiency.

[0130] The above three sub-scores are combined into a global score. The calculation method is as follows: ,in , , These are the weighting coefficients for traffic distance penalty, column conflict penalty, and form quality reward, respectively. All three are positive numbers and can be adjusted according to the project's protection requirements and functional needs. The larger the value, the higher the overall quality of the layout.

[0131] A more detailed macro-level scoring mechanism was adopted in the cluster-level layout phase, and the scoring function was defined as:

[0132] .

[0133] The shape score measures the rectangular integrity and aspect ratio rationality of each group partition; For location scoring, for zones designated as "remote clusters", an increase in the Euclidean distance between their geometric center and the overall building plan centroid is awarded. As a compactness penalty, it applies to clusters that are too scattered or elongated. This is an edge penalty term, which comprehensively measures the alignment quality of cluster partition boundaries. Its calculation method is as follows: ,in, The misalignment length between the boundary of the cluster zone and the corner point of the building outline. This refers to the misalignment length between the cluster zone boundary and the corridor edge line. The irregular length of the common boundary between adjacent clusters.

[0134] For reward items adjacent to the corridor, there is a common boundary between the reward cluster area and the corridor. This is a connectivity bonus, rewarding the spatial connectivity quality of atomic regions within a cluster partition.

[0135] After obtaining the global score, it is necessary to further decompose the contribution of each functional group to the circulation distance score and the morphological quality score from the global score, in order to accurately identify the groups with weak quality. For the circulation distance contribution, the sum of all shortest path lengths related to a certain functional group (i.e., the sum of the shortest paths of all node pairs involved in that group's nodes) is used as the percentage... The proportion is defined as the contribution of the movement distance of the group. Regarding the contribution of morphological quality, the average rectangular integrity of all rooms within a functional cluster is compared with the global average. The absolute value of the difference is defined as the contribution of the morphological quality deviation of the cluster. .

[0136] Will Exceeding the preset threshold for contribution of travel distance ,or Exceeding the preset threshold for contribution of morphological quality deviation Functional clusters are marked as problem regions. The logic for identifying problem regions reflects the idea of ​​precise local intervention: only clusters with significantly higher contribution are included in the re-solution scope, thereby avoiding unnecessary disturbance to regions that have already reached a relatively optimal state.

[0137] For functional clusters marked as problem areas, their current room subdivision results are released, meaning the fixed state of all room boundaries within the cluster is revoked, restoring them to continuous regions to be assigned. Meanwhile, functional clusters outside the problem areas retain their existing room subdivision results, participating in subsequent solutions as locked layouts. Room boundaries in locked layouts are introduced as immutable boundary conditions during the re-solution phase, ensuring that the results of optimized regions are not destroyed, while providing explicit spatial constraints for the re-solution of problem areas.

[0138] Using the released problem region as the solution space, room-level space allocation is re-executed under the immutable boundary constraints of the locked layout. During the re-solution, the full coverage partitioning constraint (all atomic regions within the problem region must be assigned to a room, with no unassigned regions) and the column grid containment constraint (room boundaries must not separate columns within the same column grid span into different rooms to ensure the integrity of the structural logic) are still followed. Under this constraint framework, the constraint programming solver regenerates room subdivision schemes for the problem region and merges them with the locked layout to form a new round of complete layout schemes.

[0139] After the new layout scheme is generated, the global score is recalculated to obtain a new global score. .by As an acceptance criterion: if the global score of the new solution is higher than that of the previous round, the current re-solution result is accepted, the old solution is replaced with the new solution, and the global score record is updated; if the global score of the new solution fails to improve, the current re-solution result is discarded, and the previous solution is retained. This acceptance criterion ensures the monotonic improvement of the iterative process and prevents the layout quality from degrading during iteration.

[0140] In each iteration, the scope of problem region identification changes as the layout quality improves: as previously marked problem regions improve, a new round of scoring decomposition can identify other clusters as new problem regions, thus driving the iteration to advance towards different local regions. This dynamic problem region identification mechanism allows the optimization process to gradually spread across the entire plane, avoiding getting trapped in local optima.

[0141] The iteration termination condition is: no improvement in the global score is produced in multiple consecutive iterations, that is, for any consecutive iterations... Each iteration has . The preset number of iterations for termination is typically 2 to 5, which can be adjusted based on building size and computational resources. When the termination condition is met, the iteration process ends, and the currently retained optimal layout scheme is output as the final layout scheme. The final layout scheme comprehensively reflects the optimization results in three dimensions: circulation efficiency, column grid compatibility, and room form regularity, meeting the comprehensive quality requirements for the layout of experimental functional spaces in the renovation of protected buildings.

[0142] A second aspect of the present invention provides a constraint-optimized experimental functional space layout system for the renovation of protected buildings, comprising:

[0143] The discretization unit is used to generate a basic cutting line set by discretizing the acquired target building plan data through a grid, and to select the shape turning line, the main corridor edge line and the column grid axis as priority cutting lines. The weighted priority cutting line set is sorted according to weight and used to perform constraint-aware discretization of the building plan.

[0144] The cluster assignment unit is used to analyze the functional requirements of the discretized building plan, group the functional requirements into functional clusters according to the cluster label, generate a candidate rectangular size pool for each functional cluster, divide the building plan into non-overlapping and tessellated minimum rectangular atomic regions, enumerate the atomic region combinations that satisfy connectivity for each functional cluster, and generate a cluster-level layout scheme by non-overlapping spatial assignment of each functional cluster based on constraint planning.

[0145] The room allocation unit is used to perform room-level space allocation within the division area of ​​each functional group based on the group-level layout scheme and full coverage and column grid accommodation constraints to complete the subdivision of rooms within the group. It generates service space candidate positions near obstacle neighborhoods and corridor intersection nodes and places them into service spaces to obtain a preliminary layout scheme.

[0146] The iterative optimization unit is used to perform a global score on the preliminary layout scheme and identify problem areas based on the distance of the movement line, the conflict of the column positions and the shape quality. The problem areas are locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

[0147] A third aspect of the present invention provides an electronic device, comprising:

[0148] processor;

[0149] Memory used to store processor-executable instructions;

[0150] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0151] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0152] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0153] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the layout of experimental functional spaces based on building renewal constraints, characterized in that, include: The acquired target building plan data is discretized by grid to generate a basic cutting line set. From this set, the shape turning line, the main corridor edge line and the column grid axis are selected as priority cutting lines. The weighted priority cutting line set is sorted by weight and used to perform constraint perception discretization on the building plan. Functional requirements are analyzed on the discretized building plan. The functional requirements are grouped into functional groups according to the group labels and a candidate rectangular size pool is generated for each functional group. The building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions and the atomic region combinations that satisfy connectivity are enumerated for each functional group. Based on constraint planning, non-overlapping spatial assignment is performed on each functional group to generate a group-level layout scheme. Based on the cluster-level layout scheme, within the division area of ​​each functional cluster, the protection fixed area corresponding to the protection element is regarded as an unoccupiable locked area. Based on the full coverage and column grid accommodation constraints, room-level space allocation is carried out to complete the subdivision of rooms within the cluster, so that the laboratory functional rooms are arranged around the outside of the protection fixed area. Service space candidate positions are generated near the intersection of the protection fixed area and the corridor and placed into the service space to obtain the preliminary layout scheme. The preliminary layout scheme is scored globally, and problem areas are identified based on circulation distance, column conflicts, and morphological quality. The problem areas are then locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

2. The method according to claim 1, characterized in that, From the basic cutting line set, shape transition lines, main corridor edge lines, and column grid axes are selected as priority cutting lines and sorted by weight to form a weighted priority cutting line set, including: Extract the coordinates of each vertex of the building's outer contour and calculate the turning angle of the adjacent side at each vertex to determine the concavity / convexity attribute. Locate all concave vertices and generate candidate segmentation lines that pass through the concave vertices and are parallel to the adjacent contour sides. Identify the candidate segmentation lines that can completely divide the concave area into independent rectangular areas as shape transition lines and assign them the highest weight. Extract continuous line segments that are consistent with the direction of the corridor along both sides of the main corridor, merge the continuous line segments into the edge line of the main corridor and assign them the second highest weight; The column direction axis is formed by sequentially connecting the center points of the same column along each column direction and sequentially connecting the center points of the same row along each row direction. The column direction axis and the row direction axis are then merged into the column grid axis and assigned a third weight. The morphological transition line is inserted into the basic cutting line set with the highest priority, the main corridor edge line is inserted into the basic cutting line set with the second highest priority, and the column grid axis is inserted into the basic cutting line set with the third priority. This allows the high-weight priority cutting line to obtain priority division weight during the discretization process. The three types of priority cutting lines are then added to the basic cutting line set in a weighted overlapping manner to form the weighted priority cutting line set.

3. The method according to claim 2, characterized in that, Candidate segmentation lines that can completely divide a concave region into independent rectangular regions are identified as shape transition lines, including: Extract all vertices of the building's planar outline and number each vertex in a clockwise direction, recording the coordinates of each vertex. For each vertex, calculate the cross product sign formed by the previous outline edge vector and the next outline edge vector, and determine whether the current vertex is a convex vertex or a concave vertex based on the sign of the cross product sign. Traverse all concave vertices and, starting from the current concave vertex, draw rays along the extension directions of the two contour edges adjacent to that concave vertex. The line connecting the contour point where each ray first intersects the outer contour of the building plane and the concave vertex is taken as the candidate dividing line. The building plan is divided into two sub-regions by each candidate dividing line. The rectangular integrity of each sub-region is calculated as the ratio of the area of ​​the sub-region to the area of ​​its smallest bounding rectangle. The weighted average of the rectangular integrity of the two sub-regions is then taken. At the same time, the ratio of the sum of the areas of the two sub-regions to the total area of ​​the building plan is calculated as the area utilization rate. The weighted average of the rectangular integrity and the area utilization rate are used as a comprehensive index to evaluate the dividing effect of the candidate dividing line. Sort all candidate dividing lines according to the comprehensive index from high to low, and determine the candidate dividing line corresponding to the maximum value of the comprehensive index as the morphological turning line.

4. The method according to claim 1, characterized in that, The building plan is divided into non-overlapping and tessellated minimum rectangular atomic regions, and combinations of atomic regions satisfying connectivity are enumerated for each functional cluster, including: Using the weighted priority cutting line set as the dividing boundary, the discretized building plane is successively divided along the extension direction of each priority cutting line, so that the area between two adjacent priority cutting lines is divided into rectangular grid units. Each rectangular grid unit is defined as an atomic region and all atomic regions are non-overlapping and completely cover the entire building plane. An adjacency graph of atomic regions is established with each atomic region as the basic operation unit. Each node in the graph represents an atomic region and is connected to the nodes corresponding to adjacent atomic regions by an edge. The length of the shared boundary between the two atomic regions connected by each edge is recorded as the adjacency weight. Based on the area requirements of each functional group, the adjacency graph of the atomic regions is used as the search space. Starting from the initial atomic region corresponding to the location of each functional group, all connected subgraphs including the initial atomic region are enumerated by breadth-first traversal. The set of atomic regions contained in each connected subgraph is taken as a connected region combination. Connected region combinations whose absolute difference between the total area of ​​the connected region combination and the target area of ​​the corresponding functional group is within the preset tolerance range are selected as candidate assignment combinations for the functional group.

5. The method according to claim 4, characterized in that, Based on constraint programming, non-overlapping spatial assignments are performed on each of the aforementioned functional clusters, including: Using each candidate assignment combination of each functional group as a Boolean decision variable, construct a non-overlapping constraint that each atomic region is occupied by at most one functional group and a mutual exclusion constraint that each functional group has one and only one candidate assignment combination selected, and construct a cardinality constraint that the number of atomic regions contained in the selected candidate assignment combination must be within a preset range. The absolute value of the difference between the actual total area of ​​the selected candidate assignment combination and the target area of ​​the corresponding functional group is used as the area deviation penalty term, and the difference between the total length of the boundary line of the selected candidate assignment combination and the perimeter of the square with the same area is used as the morphological density reward term. The weighted sum of the area deviation penalty term and the morphological density reward term is used as the objective function. Using the non-overlapping constraints, the mutual exclusion constraints, and the cardinal constraints as constraints, and minimizing the objective function as the optimization direction, the values ​​of each Boolean decision variable are solved through constraint programming to minimize the objective function. The set of atomic regions corresponding to the selected candidate assignment combinations in each functional group is taken as the occupied area of ​​the functional group on the building plan. The set of occupied areas of all functional groups constitutes the cluster-level layout scheme.

6. The method according to claim 1, characterized in that, Room-level space allocation is carried out within the defined areas of each functional cluster, and service spaces are placed near obstacle neighborhoods and corridor intersections, including: Within the division area of ​​each functional group, the area requirements and rectangular shape requirements of each room are used as input parameters. The hard constraint is full coverage, that is, the sum of the rectangular areas of each room must be equal to the area of ​​the division area of ​​the functional group. The spatial constraint is column grid accommodation, that is, when the minimum area of ​​a room is greater than three times the area of ​​a column grid unit, the room is allowed to accommodate columns. By enumerating the candidate rectangular size combinations of each room and verifying whether each combination meets the full coverage constraint and column grid accommodation constraint, the combination that meets all the constraints is taken as the room-level subdivision scheme. The union of the preset neighborhood range of each obstacle and the preset neighborhood range of the corridor intersection node is taken as the service space candidate area. Multiple candidate placement positions are generated in the candidate area at preset intervals, and the target placement position is determined by the minimum distance between each candidate placement position and the boundary of the adjacent assigned room as the evaluation index. After placing the service space at the target placement location, it is detected whether there is an overlap or gap between the boundary of the service space and the boundary of the allocated room. If there is an overlap or gap, the boundaries of the remaining allocated rooms are kept unchanged, and the functional groups involved in the overlapping or gap area are reassigned to room-level spaces separately.

7. The method according to claim 1, characterized in that, The preliminary layout scheme is globally scored, and problem areas are identified for local release and re-solution, including: The sum of the shortest path lengths between each functional group and between each functional group and the entrance / exit is used as the circulation distance score; the cumulative value of the number of columns contained in a single room in all rooms is used as the column conflict score; the arithmetic mean of the ratio of the area of ​​each room to the area of ​​its smallest bounding rectangle is used as the morphological quality score; and the weighted sum of the above three scores is used as the global score. The contribution of each functional group to the circulation distance score and morphological quality score in the global score is decomposed from the global score. Functional groups with a contribution exceeding a preset threshold are marked as problem areas and the room subdivision results in the problem areas are released. The room subdivision results of functional groups in non-problem areas are retained as locked layouts. The problem area after release is taken as the solution space. The locked layout is taken as an immutable boundary condition. The solution space is redistributed at the room level based on the full coverage partitioning constraint and the column grid accommodation constraint. The acceptance criterion is whether to retain the current rediscovery result based on whether the global score after rediscovery is improved compared with the previous round of global score. The iterative process of global score, problem area identification and local rediscovery is repeated until the global score is not improved in multiple consecutive rounds of iteration. Then the iteration is terminated and the final layout scheme is output.

8. An experimental functional space layout system based on building renewal constraints, used to implement the method as described in any one of claims 1-7, characterized in that, include: The discretization unit is used to generate a basic cutting line set by discretizing the acquired target building plan data through a grid, and to select the shape turning line, the main corridor edge line and the column grid axis as priority cutting lines. The weighted priority cutting line set is sorted according to weight and used to perform constraint-aware discretization of the building plan. The cluster assignment unit is used to analyze the functional requirements of the discretized building plan, group the functional requirements into functional clusters according to the cluster label, generate a candidate rectangular size pool for each functional cluster, divide the building plan into non-overlapping and tessellated minimum rectangular atomic regions, enumerate the atomic region combinations that satisfy connectivity for each functional cluster, and generate a cluster-level layout scheme by non-overlapping spatial assignment of each functional cluster based on constraint planning. The room allocation unit is used to perform room-level space allocation within the division area of ​​each functional group based on the group-level layout scheme and full coverage and column grid accommodation constraints to complete the subdivision of rooms within the group. It generates service space candidate positions near obstacle neighborhoods and corridor intersection nodes and places them into service spaces to obtain a preliminary layout scheme. The iterative optimization unit is used to perform a global score on the preliminary layout scheme and identify problem areas based on the distance of the movement line, the conflict of the column positions and the shape quality. The problem areas are locally released and re-solved based on constraint programming. The final layout scheme is output through iterative optimization.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images