Indoor reconstruction space identification and classification method for old building

Abstract

The invention relates to an old building indoor reconstruction space identification and classification method, and the method comprises the following steps: constructing a space degradation factor model which comprises a sealing degree, a use trace feature, and a furniture residual layout hidden variable; judging whether the space is subjected to use degradation or not by utilizing internal texture feature recognition and abnormal layout mode detection of the space including wall peeling and pipeline exposure; a fuzzy function label mechanism is introduced, and the space is allowed to have multiple purposes in the recognition stage; constructing a spatial use evolution graph which comprises historical use nodes, current use nodes and potential use nodes; establishing a semantic graph neural network SGNN, and taking spatial physical attributes + a connection relationship + a use evolution trajectory as input; space purpose reasoning and transformation direction prediction are carried out, wherein a certain storage room is reasoned as a convertible study room or a laundry room; each space node has the application evolution credibility, and assists a decision maker in screening the transformation direction.

Description

Technical Field

[0001] The present invention relates to a method for identifying and classifying renovated spaces, specifically a method for identifying and classifying interior renovated spaces in old buildings. Background Art

[0002] Currently, in the practice of interior renovation of old buildings, the methods for space identification and classification still mainly rely on manual experience, assisted by static drawings, and single-dimensional judgment, with a large number of deficiencies in terms of technology and logic. Especially when facing old space environments with mixed functions, complex structures, and missing historical information, the limitations of traditional methods are more prominent. First of all, the current mainstream space identification methods mostly rely on original building drawings and simple measurement tools, or judge function division through two-dimensional CAD drawings. However, during the use of old buildings, there are often multiple non-standard usage behavior changes, such as a bedroom being changed into a storage room, a balcony being enclosed and becoming an extension of the kitchen, and a corridor being opened up and becoming a work area. These space evolutions do not leave traces in the drawings, resulting in a serious deviation between the identification result and the actual function. At the same time, some buildings even lack original drawings, and it is almost impossible to carry out effective identification relying on traditional methods;

[0003] Secondly, the current classification method highly relies on manual experience. Even with the assistance of a BIM platform or house surveying and mapping software, its core classification logic is still limited to the extensive judgment of "space size + name label", unable to truly reflect whether the space conforms to the current functional attributes, and even less able to identify the states of "mixed functions", "degraded uses", or "ambiguous spaces", resulting in renovation plans often being based on misjudgments, and problems such as inapplicable spaces, structural conflicts, and functional overlaps occurring after renovation; Moreover, existing identification methods almost ignore the semantic relationships and temporal evolution trajectories between spaces. The functions of spaces in a building do not exist in isolation. Their functions are often affected by the uses of adjacent spaces, access paths, lighting orientations, and structural couplings. For example, whether a small room can be converted into a study does not only depend on its size, but also on whether it is close to a quiet area, well-ventilated, and has psychological independence. Traditional identification methods ignore such semantic dimensions and space network attributes and are difficult to establish a systematic renovation logic;

[0004] In addition, the traditional method seriously lacks the integrated analysis of structural feasibility and functional value. Existing evaluations mostly judge the structure and use separately. That is, structural engineers are responsible for judging whether a wall can be demolished, and designers judge whether the use should be changed. There is a lack of a unified and integrated quantitative standard for the transformability of space, resulting in frequent mismatches in practice such as "wanting to change but unable to change" and "able to change but not worth changing". There is also a lack of a mechanism to evaluate the credibility of the evolution of space from its original use to a new use. Even with certain experience support, it is difficult to systematically evaluate each space in large and complex buildings one by one, with extremely low efficiency. For another example, current technical means have not effectively introduced non-structural information such as the historical usage trajectory, usage marks, changes in furniture layout, and degree of equipment aging between spaces. Although this information is not engineering-oriented, it is an important basis for judging the actual usage status and transformation value of space. For example, whether there are electrical fixing holes on the wall, whether there are settlement marks on the ground, and whether the corner is chronically damp. These may all reflect the actual use and problem risks of the space in the past period, but traditional identification methods cannot model, analyze, or use them for subsequent classification and design decisions. A more prominent problem is that after the identification and classification are completed, traditional methods cannot form a traceable and verifiable renovation decision-making chain. Design solutions mostly rely on designers' intuition to judge how to adjust the space, lacking the logical closed-loop between the previous identification results and the subsequent design actions, and it is also difficult to provide multi-scheme comparison and credibility ranking, resulting in the design results being easily affected by individual preferences and experience levels and lacking a unified quality judgment standard. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for identifying and classifying the interior renovation space of old buildings, so as to solve some of the drawbacks and deficiencies pointed out in the background technology.

[0006] The technical solutions adopted by the present invention to solve its above technical problems include the following steps:

[0007] S1. Adopt an indoor structure-function fuzzy identification mechanism:

[0008] S1.1. Construct a space degradation factor model, including the latent variables of the degree of enclosure, usage mark characteristics, and residual furniture layout;

[0009] S1.2. Use the internal texture feature recognition of the space, including wall peeling and pipeline exposure, and abnormal layout pattern detection to judge whether the space has experienced usage degradation; and introduce a fuzzy function label mechanism to allow multiple usage tendencies of the space during the identification stage;

[0010] S2. Construct an indoor usage inference of an evolutionary space semantic network:

[0011] S2.1. Construct a spatial use evolution map, including historical use, current use, and potential use nodes; establish a semantic graph neural network SGNN, with spatial physical attributes + connection relationships + usage evolution trajectories as inputs;

[0012] S2.2. Conduct use reasoning and renovation direction prediction for spaces, including inferring that a storage room can be transformed into a study or a laundry room; each spatial node is attached with a credibility of use evolution to assist decision-makers in screening renovation directions;

[0013] S3. Adopt a spatial renovation adaptation index through coupling analysis:

[0014] S3.1. Construct a structural constraint model, calculated based on load-bearing wall layouts, structural nodes, and beam-slab structure parameters; at the same time, call the evolution potential scores in the spatial semantic map to form a dual-index evaluation matrix;

[0015] S3.2. Introduce a spatial renovation adaptation index SPI, generated through a non-linear fusion algorithm of structural bearing capacity × functional evolution potential; the adaptation index SPI provides three types of renovation suggestions for designers, namely priority renovation areas, areas that should be carefully handled, and areas to be reserved or abandoned;

[0016] S4. Adopt a spatial adaptability enhancement and reconstruction strategy: Based on the spatial adaptation index and use prediction results, design a spatial reconstruction strategy generator; the strategy types include: space release type, function replacement type, transitional connection type, and micro-intervention optimization type.

[0017] Furthermore, the indoor use reasoning process for constructing an evolutionary spatial semantic network includes:

[0018] Based on the historical drawings, on-site scanning data, and spatial usage traces of old buildings, model each indoor space unit, including rooms, corridors, and storage rooms, as nodes in a graph structure; each node is assigned a set of attribute vectors, covering geometric dimensions, physical states, usage traces, and historical uses, and analyze the physical connections, access paths, and shared functional elements with adjacent spaces to establish edge connection relationships between spaces; the dependence strength between nodes is represented by a connection weight function:

[0019]

[0020] Among them:

[0021] f r (i, j) represents the functional connection weight between spatial nodes i and j; γ ij represents the structural coupling strength between nodes i and j; δ ij represents the functional interaction frequency between the two nodes; d ijis the geometric center distance between two spaces, used to measure the physical proximity; α is an adjustment coefficient that controls the attenuation degree of the influence of distance in weight calculation.

[0022] Furthermore, the indoor use inference process for constructing an evolutionary spatial semantic network includes:

[0023] Based on node attributes and connection weights, construct a spatial use evolution map, record the original design use, current state, and potential transformable uses of each space, and organize the use evolution process in a time series manner; the map is used as input to enter a graph neural network model for training to learn the migration trend of spatial uses during the actual use evolution process. The prediction expression of use evolution potential is as follows:

[0024] P u (i,k) = ∫0 T θ i (t)·μ ik (t)dt

[0025] Where:

[0026] P u (i,k) represents the cumulative potential score for the evolution of space i to use type k; T is the evolution time range; θ i (t) represents the functional stability function of space i at time t; μ ik (t) represents the adaptive evolution rate function of space i in the direction of use k; it reflects the degree of fit between structure, location, usage traces, etc. and the target use.

[0027] Furthermore, the indoor use inference process for constructing an evolutionary spatial semantic network includes: introducing a use evolution credibility scoring function to measure the reliability and construction adaptability of each use evolution suggestion, and the calculation is as follows:

[0028]

[0029] Where:

[0030] C(i,k) is the credibility score for the evolution of space i to use k; N is the total number of attribute types of the spaces participating in the evaluation, including lighting, ventilation, floor height, and area; ψ n (i,k) represents the fitness evaluation score of the nth attribute for space i under use k; w n is the importance weight factor of attribute n, determined by the user or design task parameters.

[0031] Furthermore, the method for constructing the spatial transformation adaptation index of the coupling analysis:

[0032] Based on the three core elements of the load-bearing wall identification density, beam-column joint structure, and floor response characteristics, a structural constraint factor model is constructed for each indoor space; the inhibitory strength of the restrictive factors on the space transformability is quantified, and the structural constraint index S(i) of space i is defined as:

[0033]

[0034] Where:

[0035] η p represents the influence weight of the p-th type of structural factor including load-bearing walls, shear walls, and beam-column layouts on the space transformation restriction; ξ ip is the actual strength performance of space i on this factor, including wall thickness and layout density; ρ in the denominator i represents the structural complexity adjustment coefficient of the space, which is used to balance the deviation tendency in the calculation of spaces with many nodes and dense structures.

[0036] Furthermore, the method for constructing the space transformation adaptation index of the coupling analysis:

[0037] Introduce a semantic-level space use evolution potential evaluation model to measure the functional feasibility and sustainability of the space evolving from the existing state to a new use; the potential score depends on the static attributes of the space itself such as size, shape, lighting, and ventilation, and also includes the functional linkage with the surrounding spaces and the dynamic factors of the long-term use trajectory; the functional evolution potential scoring function F(i) of space i is defined as follows:

[0038]

[0039] Where:

[0040] φ i (t) represents the use elasticity state of space i at time t, which is used to describe the trend of functional reconstruction under the current use; ω iq (t) is the response value of the q-th type of evolution factor including adjacent space use support degree, connection strength, and structural openness on space i, reflecting the support degree of the environment for evolution; λ q is the influence weight of the q-th type of evolution factor, indicating the emphasis on the influence degree in different design scenarios; and the integration time period T represents the observation period of the functional evolution trend.

[0041] Furthermore, the method for constructing the space transformation adaptation index of the coupling analysis:

[0042] After completing the two-dimensional modeling of the structural constraint S(i) and the functional potential F(i), a unified evaluation index, namely the space transformation adaptation index SPI, is generated through non-linear fusion of the two types of indicators; the index takes the coupling of the functional potential and the structural constraint as the core evaluation logic and is defined as:

[0043]

[0044] Wherein:

[0045] SPI(i) is the adaptation index of space i, which is used to evaluate whether the space is suitable for indoor functional transformation; σ is the structure suppression coefficient, which is used to control the weakening degree of the structure limitation on the overall index; S(i) is the aforementioned structure limitation value, and max(S) is the maximum value of the structure limitation in the entire building space, which is used for normalization; F(i) is the functional evolution potential score of the space, is the average potential score of all spaces; β is the functional potential strengthening factor, which adjusts the priority score growth amplitude of high-potential spaces.

[0046] Further, according to the SPI values of each space, the internal space of the building is divided into three categories: those with SPI values higher than the set upper threshold are marked as priority transformation areas; spaces with SPI values in the middle gray area are classified as cautious treatment areas according to the fluctuation degree of the structure and function scores; spaces with SPI values far lower than the average level, indicating that the structure is immovable and the evolution value is low, are classified into reserved or abandoned areas.

[0047] The method for identifying and classifying indoor renovation spaces of old buildings proposed by the present invention aims at the prominent problems in the current renovation process of old buildings, such as "fuzzy space functions, complex structure limitations, unclear renovation directions, and design decisions relying on manual experience", and provides a systematic identification and classification method based on data-driven, semantic understanding and structural function integration, which has the following

[0048] Beneficial effects:

[0049] By constructing a spatial semantic map and introducing a graph neural network model, the spatial recognition not only stays at the geometric form level, but also delves deeper into the understanding of the spatial function evolution logic and the intelligent reasoning of the usage trend, thus greatly improving the cognitive accuracy of the spatial attributes and potential in the early stage of renovation;

[0050] Through the modeling of structure limitation factors and the integral evaluation of functional potential, the spatial renovation adaptation index (SPI) is creatively proposed, realizing the non-linear integration of structural transformability and functional evolution value, making the spatial classification no longer an empirical judgment, but a scientific decision based on evidence and quantifiable ranking; it supports the output of the reliability score of the usage evolution and the renovation recommendation level for each space, providing multi-dimensional intelligent references of "whether to renovate, what to renovate, and where to renovate first" for designers, and significantly reducing the risks of repeated adjustment, misjudgment and resource waste of the renovation plan. Description of the Drawings

[0051] Figure 1 is the flow chart of the method for identifying and classifying indoor renovation spaces of old buildings of the present invention.

[0052] Figure 2 Flowchart of the indoor usage inference process for constructing an evolutionary spatial semantic network of the present invention.

[0053] Figure 3 Flowchart for constructing the spatial transformation adaptation index for coupling analysis of the present invention. Detailed implementation manners

[0054] The following makes a detailed description of the specific implementation manners of the present invention with reference to the accompanying drawings.

[0055] In combination with Figure 1The process. The core starting point of the method for identifying and classifying the interior renovation space of old buildings lies in its accurate understanding and determination of the current state of the interior space. Especially for old buildings with serious functional mixing, use changes, or long-term degraded use, their spatial attributes often appear fuzzy, chaotic, and do not conform to the original design standards. Therefore, this method first proposes a fuzzy recognition mechanism for indoor structural functions as the pre-entry of the entire renovation recognition system. This mechanism includes two core steps. The first step, S1.1, is to construct a spatial degradation factor model. This model does not rely on the traditional method of classification only by area or bay, but deeply explores those hidden variables that reflect the actual state of the space but are not visible in conventional drawings. Specifically, it includes the degree of enclosure of the space, that is, indicators such as whether there is an access passage, whether the ventilation path is blocked, and whether the windows are closed for a long time; secondly, the recognition of use trace characteristics, such as whether there are modified socket lines, serious aging of wall coatings, or traces such as smoke stains and mildew; furthermore, the residual furniture layout. This kind of information infers the previous use purpose and frequency of the space by identifying the distribution of the remaining furniture in the space, furniture indentations, and the distribution of ground wear areas. These hidden variables together constitute the "degradation factor map" of this space; on this basis, enter the second step, S1.2, that is, use the indoor texture image recognition and layout pattern analysis method to identify whether there are use degradation characteristics in the space. This process is mainly judged based on visible damage and non-standard structural layouts. For example, by identifying the surface texture information of the space such as the peeling area of the wall, the exposed pipeline area, and the non-enclosed ceiling area, it is inferred that the space may have been in a state of long-term lack of maintenance or use function transformation. At the same time, combined with the analysis of abnormal space structures, such as non-orthogonal partition walls, secondary installed or modified door and window openings, and non-standard layout elements such as narrow passages, these data features are comprehensively judged through algorithms to identify whether the space has deviated from its design function and entered a degraded state; after completing this recognition process, this method introduces a fuzzy function label mechanism, that is, allows each space to be assigned multiple use tendencies during the function classification stage, rather than being forcibly defined as a single use type. For example, a space can be marked as "storage tendency + secondary living tendency" at the same time. This multi-label strategy effectively avoids the misjudgment problem caused by fuzzy use in the traditional recognition process and provides a more flexible semantic basis for the subsequent functional evolution path analysis and renovation strategy generation.

[0056] Enter the second stage, namely the indoor use inference for constructing an evolutionary spatial semantic network. The goal of this stage is to achieve a deep understanding of spatial uses and an intelligent prediction of renovation directions. Its core steps are divided into two parts. The first step, S2.1, is to construct an evolutionary map of spatial uses. Each spatial node in the map not only retains the current use status but also includes its original design function and potential renovation function labels. This triple semantic annotation enables each space to evolve from a static definition to a dynamic one. The nodes in the map are connected by physical connectivity, structural dependency relationships, and historical use migration paths to form connection edges, reflecting the functional coupling and usage logic between spaces. Based on this map, the method further establishes a semantic graph neural network (SGNN) model. This model takes the multi-dimensional attribute vectors of spatial nodes as input, including physical parameters such as the geometric size, bay ratio, orientation characteristics, ventilation and lighting conditions of the space. At the same time, it introduces the connection relationship characteristics between spaces, such as whether there is a direct door connection, whether the load-bearing structure is shared, and whether it is connected to the same traffic flow line. On this basis, the usage evolution trajectory is added, that is, the evolution record of the spatial function at different times, including state transition information such as being opened up, enclosed, replaced, merged, etc. Through training, this SGNN model can learn the functional evolution logic between spaces and the potential patterns of use migration.

[0057] The second step, S2.2, is to use the above-trained model for spatial use inference and renovation direction prediction. When the state parameters of a current spatial node are input, the system can automatically output the possible evolution path and corresponding use suggestions for this space based on the context information of its map. For example, a node currently used as a storage room can be inferred to be convertible into a study or a laundry room in the future. This kind of inference is not based on rule matching of individual spaces but on finding the evolution trends of similar spaces in the overall map and performing semantic propagation to achieve use judgment based on the global network context. At the same time, to improve the interpretability and design practicality of the inference results, each spatial node is also accompanied by a use evolution credibility score, which measures the structural fitness, functional rationality, and semantic consistency of its migration from the current state to a certain target use.

[0058] Enter the third stage, that is, use the spatial transformation adaptation index of coupling analysis to conduct the final transformation feasibility judgment and priority ranking. The core of this stage lies in coupling and calculating the two originally independent evaluation dimensions of "whether the structure can be transformed" and "whether the function is worth transforming" to form a unified quantitative standard for transformation decision-making. First, in step S3.1, construct a structural constraint model. Based on the building structure system, for each indoor space, extract key structural parameters such as the distribution of load-bearing walls, the density of beam-column joints, the thickness of the floor slab, the allowable opening degree, and the structural vibration response, and calculate the structural bearing risk level of this space according to these data, so as to clarify the difficulty and risk of structural intervention. At the same time, the system synchronously calls the evolution potential score in the spatial semantic map constructed in the previous stage, that is, to what extent the space has the possibility of functional transformation at the semantic level and whether it can be reasonably evolved into other uses, such as living space, working space, service space, etc. Two groups of indicators - the structural constraint value and the functional evolution potential value - are constructed into a dual-index evaluation matrix to identify the relationship between "transformability" and "should-be-transformedness" of each space in the early stage of design.

[0059] Then, in step S3.2, introduce the spatial transformation adaptation index SPI. This index is generated through a non-linear fusion algorithm. The fusion method is not a simple weighting, but by multiplying the structural bearing transformability and the functional use evolution potential and then non-linearly scaling to form a comprehensive score, which is used to express the dual satisfaction degree of "structural feasibility + functional value" of a certain space in actual transformation. The higher the SPI index, the more likely the space is to be physically intervened and the more worthy of functional replacement, thus becoming a priority transformation object. On the contrary, the lower the SPI, it may mean that the structure cannot be moved or the benefit after transformation is too small, and it should be included in the reserved or abandoned area. By sorting and grading the SPI values of all indoor spaces in the whole building, the system will automatically generate three types of transformation recommendation areas, namely, the "priority transformation area", that is, the space with low structural constraints, high functional potential, and should be given priority to consider the input of design resources; the "area that should be handled with caution", that is, there are uncertainties in structural or functional indicators, and it needs to be comprehensively evaluated in combination with the design intention; the "reserved or abandoned area", that is, the structure is severely restricted or the functional evolution value is low, and it is not suitable for large-scale intervention.

[0060] Enter the fourth stage, that is, adopt the spatial adaptability enhancement and reconstruction strategy to achieve a natural transition from recognition and judgment to design actions. This stage no longer just stays at the analysis level of whether the space "can be transformed", but further proposes a systematic strategy output mechanism of "how to transform and what to transform into". Its core is to establish a spatial reconstruction strategy generator. This generator takes the spatial transformation adaptation index (SPI) generated in the previous stage and the usage prediction results obtained from semantic network reasoning as inputs, automatically matches the transformation methods and intervention depths of each space, and generates corresponding design strategies. The system combines the preset transformation target types with the actual feasibility and proposes four types of spatial reconstruction strategies. The first type is the space release type, mainly targeting those spaces that are currently occupied by functions or overly divided, such as storage rooms, mezzanines, enclosed balconies, etc. By means of structural demolition, merging, interface simplification, etc., the space potential is released to achieve area reconstruction and enhanced sense of openness. The second type is the function replacement type, which is applicable to units where the function is no longer applicable but the spatial form is good. For example, if it was originally a kitchen but lacks flue or pipeline support, it can be replaced with a study, work area or leisure corner. Based on the usage prediction results, the system recommends feasible alternative uses and proposes necessary structural and equipment modification suggestions. The third type is the transitional connection type, which deals with the problems of space fragmentation and unsmooth moving lines. For example, by increasing openings, setting up transitional buffer zones, guiding paving, etc., the functional connection and flow relationship between spaces are strengthened, so that the originally scattered space combination is transformed into a continuous living scene.

[0061] The fourth type is the micro-intervention optimization type, which is applicable to spaces where the structure is immovable, the function is basically reasonable but there are problems in details such as insufficient lighting, poor sound insulation, and insufficient storage. Such strategies emphasize improving the usage performance through prefabricated components, small-scale furniture reorganization or interface update without affecting the structure and main function. The entire strategy generator has an adaptive ability, can judge the strength of the intervention strategy according to the SPI value, and combines the directional suggestions given by the usage prediction to finally achieve the intelligent leap of the space from being transformable to how to transform, providing designers with scientific, efficient and implementable transformation strategy solutions, effectively enhancing the adaptability, flexibility and modern function matching ability of the interior space of old buildings.

[0062] Example 1:

[0063] A certain residential building is an urban apartment built in the 1980s. It has a structure of two bedrooms and one living room with a usable area of 58 square meters. In recent years, due to the outdated functional layout and serious spatial fragmentation, there has been a significant deviation in the actual use of the space by residents. For example, the original dining room was converted into a storage room, and the secondary bedroom was converted into a home work area, resulting in a chaotic overall space function. There is a large deviation between the original design drawings and the actual use. Therefore, before the renovation, intelligent recognition and usage prediction need to be carried out based on the spatial evolution logic. First, during the construction of the graph structure, according to the on-site laser scanning point cloud and historical building drawings, the system divides this indoor space into 6 nodes, including the master bedroom A, the secondary bedroom B, the living room C, the kitchen D, the corridor E, and the utility room F. The attribute vectors of each node are extracted. For example, the area of room A is 11.5㎡, the window opening rate is 80%, the peeling area of the wall is 2㎡, and the usage traces indicate a sleeping space (bed, bedside table). While the area of the utility room F is only 2.4㎡, there are no windows, no heating interfaces, and there are old hooks and water stains on the wall, indicating that it was once used for stacking water buckets and household sundries. The connection relationship between nodes is also encoded as graph edges. For example, there is a main door directly connecting A and C, then δ AC = 1.0, indicating a high frequency of functional interaction between the two spaces. At the same time, they share one side of the load-bearing wall. Let the structural coupling strength γ AC = 0.8, the central distance d between the two rooms AC = 3.6 meters. The adjustment coefficient α controls the neighborhood sensitivity. Empirically, the value range is set to [0.5, 2.5]. In this project, α = 1.2 is selected and substituted into the connection weight function to obtain:

[0064]

[0065] It shows that there is a strong linkage between the master bedroom A and the living room C in terms of structure and function, and it is suitable to jointly consider the optimization of space use. Taking the utility room F and the kitchen D as an example, assume that they share a wall but have no functional connection (no door). Let γ FD = 0.6, δ FD = 0.2, the central distance d FD = 2.2 meters. Still substituting α = 1.2, we get:

[0066]

[0067] This value is much lower than the connection weight between A and C, indicating that although the utility room and the kitchen are physically close, their functional linkage is poor, and direct functional replacement is not recommended; then the system constructs a spatial semantic graph of all nodes and connection edge weights and inputs it into the semantic graph neural network SGNN. The SGNN model structure includes three layers of node aggregation units. The first layer aggregates geometric and physical features, the second layer aggregates historical usage trajectories and adjacent node semantics, and the third layer predicts the usage evolution path. After data training, the model outputs potential usage suggestions and their credibility scores for each node. The final reasoning result for node F is: ① Laundry The original uses of ① the kitchen (credibility 0.78), ② the pet washing room (credibility 0.63), and ③ the storage room (credibility 0.59) are retained, indicating that they are most likely to evolve into functional service spaces. The system recommends converting F into a laundry area and sharing the water supply and drainage system with the kitchen to achieve dry and wet zoning optimization. At the same time, the use recommendations for node B (second bedroom) are family study (credibility 0.81) and children's room (credibility 0.66). Considering its characteristics of being close to the sun, moderate in size, and adjacent to the master bedroom, it is recommended to convert it into a family workspace and retain the temporary sleeping function to leave room for multi-functional use.

[0068] Next, a "spatial use evolution map" is constructed based on node attributes and connection weights, and the trend of future space use evolution is predicted based on the map. First, based on the spatial semantic map, the system introduces time series data for each node to record its complete change process from original design to current use and then to the expected function. Taking the second bedroom B as an example, its original design purpose is "living space" and its current use status is "mixed work and storage space". At the same time, residual traces of desks, printers, bookshelves, some old clothes and children's toys are recorded in historical on-site sampling. Based on this, the system will identify its potential The evolutionary path is set as: residence → multi-function → study or children's activity room, and the path is marked in the graph as the evolutionary trajectory. The graph uses time t as the vertical axis, and the use state of each node is encoded in stage order. The main evolutionary direction of node B in the graph is k = study. The graph is then sent as input to the SGNN (graph neural network) model for training. The model structure includes a function embedding layer, a spatial coupling layer, and a use propagation layer. The input includes geometric parameters (such as a room height of 2.5m and an area of 9.2㎡), structural information (whether it is against a load-bearing wall or a window), and connection weight values (such as B-Cf r =1.68, B-Ef r =1.51), the model also introduces the usage evolution potential scoring function P u (i, k) to calculate the development possibility of space in a specific use direction, which is expressed as:

[0069] P u (i,k)=∫0 T θ i (t) μ ik(t)dt

[0070] Among them, space B corresponds to i = B, the target use is k = study, and the evolution time range T is set to 5 years, which means the prediction analysis of the use migration in the next five years, θ B (t) represents the functional stability of the space at time t. The system is fitted according to the frequency of its historical usage changes. Since the usage status of space B has been relatively stable in the past three years, the value range θ B (t)∈[0.75,0.9], indicating that the usage structure is relatively stable; while the adaptive evolution rate function μ Bk (t) is calculated based on multiple factors, including: whether it has the natural lighting required for office functions (the window area of B is 2.1 m2, accounting for 32% of the wall, and the lighting is sufficient); whether it is close to a quiet area (B is next to the master bedroom A, not a traffic node, and has little interference); structural adaptability (no load-bearing wall can be dismantled and modified), so the comprehensive fitting is μ Bk (t)∈[0.6,0.8], substituting these data into the function for approximate integral estimation, it can be simplified to replace the continuous value of the integral interval with the average value, and obtain:

[0071]

[0072] This result indicates that the second bedroom B has the potential to evolve into a study room of medium to high level in the next five years. The system also scores other paths, such as B→children's room P u =2.15, B→Storage room P u =1.42, and finally the maximum potential path is used as the main use recommendation, and the result is: "study room (main use) + auxiliary sleeping function (optional design)". In addition, the utility room F is also evaluated because it is currently highly closed (no window), has obvious water marks (the fitting of usage marks indicates frequent use), is close to the kitchen D, and has a wall with an openable door. Assume that its θ F (t)∈[0.55,0.7],μ Fk (t) (used in laundry room) is set to [0.7, 0.85] according to pipeline adaptability and ventilation modification possibility, and the average value is calculated to obtain:

[0073] P u (F, laundry room) ≈ 5 0.63 0.77 = 2.43

[0074] This result is higher than F as the potential of storage room P u =1.66, so the system recommends that F should be connected to the kitchen and converted into a functional laundry area, forming a closed loop with the previous design plan.

[0075] In this embodiment, the feasibility of each spatial node in the direction of potential renovation use is finally quantified and evaluated. The core of this process is to comprehensively consider the adaptation degree of various spatial attribute dimensions, such as daylighting, ventilation, floor height, area, etc., in a specific use scenario, and make a weighted judgment in combination with the priorities of the design task or user concerns, so as to improve the accuracy and engineering implementability of the renovation suggestions. Here, still taking node B (originally a secondary bedroom, currently a mixed office space) as an example, the optimal evolution direction deduced in the previous step is a study (use k = study). Here, the credibility of this use will be calculated. Let the total number of attributes participating in the evaluation be N = 4, which are: n1 - daylighting adaptation degree, n2 - ventilation condition, n3 - floor height acceptability, n4 - area utilization efficiency;

[0076] The actual data is as follows: The window-wall ratio of space B is 32%, and the daylighting index is relatively high. After being scored by the model, ψ1(B, study) = 0.82; in terms of ventilation, it is a single-sided window, but facing north, and the cross-ventilation is weak, with a score of ψ2 = 0.65; the floor height is 2.5m, slightly lower than the ideal value of 2.7m for a study, with a medium adaptability, ψ3 = 0.72; the area is 9.2㎡, and desks and bookshelves can be arranged without pressure, with a score of ψ4 = 0.88. In the design task, the owner pays more attention to daylighting and usable area. Therefore, the weights of each attribute are set as: w1 = 0.35, w2 = 0.2, w3 = 0.15, w4 = 0.3, where the sum of the weight coefficients is required to be 1, that is, ∑w n = 1.0, and substitute it into the expression of the use evolution credibility scoring function proposed by the present invention:

[0077]

[0078] Finally, the credibility score of this space as a study is 0.789, indicating that it has high functional adaptability and implementation feasibility. The critical recommendation standard for the credibility score is: C < 0.6 is low credibility, and renovation is not recommended; 0.6 ≤ C < 0.75 is medium credibility, and careful evaluation is required; C ≥ 0.75 is high credibility, and it is recommended as the priority renovation direction. The score in this example meets the high credibility category. The system includes this plan in the first priority of the renovation design suggestion set. At the same time, the plan that infers the utility room F as a laundry room is also scored in the same way. Let the attribute scores be ψ1 = 0.42 (no window, poor daylighting), ψ2 = 0.78 (near the pipeline, low cost for ventilation modification), ψ3 = 0.76 (the floor height of 2.45m is basically usable), ψ4 = 0.62 (the area is only 2.4㎡, and the operating space is narrow). At the same time, it is assumed that the design party pays more attention to ventilation and floor height adaptability in the laundry room scenario, and the weights are set as w1 = 0.2, w2 = 0.35, w3 = 0.3, w4 = 0.15, and substitute them into the function to get:

[0079]

[0080] The credibility is 0.678, belonging to the medium credibility level, indicating that this space can be converted into a laundry room. However, there are limitations such as narrow space and insufficient lighting. Auxiliary lighting and space storage modules should be introduced in the design to improve the usage experience. In addition, this score can also be used for parallel comparison of multiple solutions. For example, if F is changed to a drying and storage room, the score is only 0.55. The system can automatically screen out the optimal solution. The overall process realizes the "data-driven + semantic understanding + structurally feasible" usage evolution credibility assessment through a clear weight setting and attribute scoring system, with engineering logic, general adaptability, and real executability, further reflecting the intelligent judgment ability and practical implementation value of the present invention in the recommendation of renovation plans for old buildings.

[0081] Embodiment 2:

[0082] Continuing with the previous 1980s old residential house of 58㎡ in Embodiment 1 as the test scenario, the current stage enters the key decision-making part, that is, quantitatively calculating the physical transformability of each space through the structural constraint factor model, and finally integrating the functional evolution potential to form a renovation priority ranking. The key point of this step is to establish the structural constraint index S(i) to evaluate the degree of inhibition of the structural conditions on space renovation, so as to assist the calculation of the subsequent renovation adaptation index SPI.

[0083] The present invention proposes to set three structural factors, namely the load-bearing wall identification density, the beam-column joint structure, and the floor response characteristics, as the main variables, that is, p = 1, 2, 3, and the specific definitions are as follows: p1 is the layout intensity of the load-bearing wall, measured according to the wall thickness and the distribution quantity. The proportion of the load-bearing wall per unit area exceeding 40% is high density; p2 is the complexity of the beam-column joint structure, counting the number of beam-column intersections in the unit space; p3 is the risk of floor opening and deflection, reflecting the adaptability of the floor to renovation loads or openings. In this example, Secondary Bedroom B is selected for detailed analysis, and its structural attributes are: the layout proportion of the load-bearing wall is 38%, and its actual performance value ξ B1 = 0.38, there are 2 main beams and 4 nodes in the node, and the structural complexity is medium ξ B2 = 0.6. The floor is a traditional precast slab with a thickness of 12 cm, a support span of 3.8 meters, and a medium and low anti-interference performance, and is evaluated as ξ B3 = 0.72. For unified calculation, the present invention sets the weight η p of each structural factor to have a standard value range of [0.2, 0.6]. In this scenario, the designer believes that the load-bearing wall restriction is the most significant, and the set weights are η1 = 0.5, η2 = 0.3, η3 = 0.2. At the same time, to avoid misjudging the space with many structural nodes but not interfering with the renovation as a high-risk space, the structural complexity adjustment factor ρ i is introduced, and its value range is recommended to be [0.1, 1.5]. Space B is of medium level, and ρ B = 0.7 is substituted into the structural constraint index calculation formula as follows:

[0084]

[0085] The result shows that the structural constraint index of secondary bedroom B is 0.302, belonging to the low to medium structural constraint level (this method suggests considering S(i) < 0.4 as low constraint, 0.4 - 0.6 as medium constraint, and > 0.6 as high constraint), indicating that this space has a certain degree of possibility of structural intervention. For example, non-load-bearing partition walls can be demolished or moderately opened up to expand functions. Subsequently, this index and the previously calculated potential score for use evolution F(B, study) = 2.87 are sent into the calculation process of the subsequent renovation adaptation index SPI for integrated evaluation. The system establishes a data support basis through a dual-channel model of structural factors and functional factors, effectively improving the objectivity and rationality of the judgment basis.

[0086] This embodiment aims to measure the feasibility and sustainability of the space when evolving into a new functional state in the future. Its core lies in not only considering the physical static characteristics of the space itself, such as area, shape, lighting, ventilation, etc., but also introducing dynamic factors such as the functional linkage relationship with adjacent spaces and the long-term use traces of the space. Furthermore, by simulating the evolution trend of functional adaptability through the time dimension, a comprehensive judgment on whether the space function "can be renovated" and "is worth renovating" is formed.

[0087] In this example, the calculation of the potential score for use evolution of space B (9.2 ㎡) is carried out, and the target use is a study. First, the integral time interval T = 5 years is set, that is, the model simulates the evolution trend of the space in the next 5 years; the use elasticity function φ B (t) represents its potential trend for function conversion in the current state. Based on the fact that its current state has been semi-converted into an office use (with a desk, socket installation, and stable lighting), after model fitting, φ B (t) = 0.55 + 0.05t is set, indicating a linear growth trend with the elasticity increasing year by year and reaching 0.8 in the 5th year; at the same time, the evolution factors include 3 categories, that is, n = 3, which are: ω B1 (t) - the support degree of adjacent spaces (such as whether it is adjacent to a quiet master bedroom), ω B2 (t) - the connection strength (whether there is a clear passage), ω B3 (t) - the structural openness (whether there is a sense of enclosure, structural barriers);

[0088] The actual value-taking situations are as follows: ω B1 (t) = 0.72 is stable because the wall between the master bedroom and the secondary bedroom is a non-load-bearing wall with little interference; ω B2 (t) = 0.65 is stable because the independent door of space B is connected to the aisle smoothly; ω B3ψ(t) = 0.68 + 0.02t, indicating that as the light decoration gradually removes the shielding wall, the openness of the structure increases to 0.78 year by year, and the influence weight λ q In terms of, according to the design objective preference, it is defined that the study design pays more attention to the adjacent functional environment and the openness of the structure. Let λ1 = 0.4, λ2 = 0.25, λ3 = 0.35, and the sum of the weights is 1. Substitute into the potential scoring function:

[0089] F(B) = ∫0 5 [φ B (t) + 0.4·ω B1 (t) + 0.25·ω B2 (t) + 0.35·ω B3 (t)]dt

[0090] Substitute the values. Since ω B1 (t) and ω B2 (t) are constants, the integral expression can be simplified:

[0091] F(B) = ∫0 5 [0.55 + 0.05t + 0.4·0.72 + 0.25·0.65 + 0.35·(0.68 + 0.02t)]dt

[0092] Simplify to get:

[0093] F(B) = ∫0 5 [0.55 + 0.05t + 0.288 + 0.1625 + 0.238 + 0.007t]dt = ∫0 5 [1.2385 + 0.057t]dt

[0094] Continue to integrate:

[0095]

[0096] Therefore, the functional evolution potential score of space B is FB≈6.91, indicating that this space has a strong trend of use evolution and functional adaptability within 5 years. If the scoring range is set as: 0–3 for low evolution potential, 3–6 for medium, and above 6 for high potential area, then space B belongs to the "high evolution potential space", which is worthy of the designer to give priority to incorporating into the functional reconstruction plan for resource allocation and layout innovation. This scoring result will subsequently be used together with the structural constraint index S(B) = 0.302 for the non-linear fusion of the renovation adaptation index SPI to form the priority ranking of space renovation.

[0097] Construct a Spatial Transformation Adaptation Index (SPI) to uniformly evaluate whether each space is worthy and capable of being preferentially transformed. Its definition logic not only considers the physical intervention boundary brought by structural limitations but also fully integrates the usage potential represented by the functional evolution trend, ultimately forming a comprehensive quantitative indicator to measure "should it be transformed, is it worth transforming, and can the transformation produce results". The formula is defined as follows:

[0098]

[0099] The specific meanings of each variable and their values in this project are as follows: Space i is node B (secondary bedroom). Given that its structural limitation value S(B) = 0.302, after the system scans all space nodes, it is found that the maximum structural limitation value appears at node D in the kitchen. Due to the dense load-bearing walls and overlapping beams and columns, it is calculated that S max = 0.684. Therefore, the normalized structural limitation ratio of space B is S(B) / max(S) = 0.302 / 0.684 ≈ 0.441. The system defines the value range of the structural suppression coefficient σ as [0.5, 1.5], which represents the weight of the impact of structural limitations on the transformation priority. If the structure is more immovable, the score should be significantly reduced. In this case, the design team sets σ = 1.0 to represent neutral suppression for safety. The previous functional potential score of space B is F(B) = 6.91. According to the statistical average of the functional potential of 6 nodes in the whole house, F = 4.86 is obtained. This value is mainly affected by low-potential nodes such as the kitchen (F = 2.9) and the utility room (F = 3.4). Therefore, the functional performance of the secondary bedroom B is significantly better than the average. In addition, a functional enhancement factor β is introduced to amplify the stretching effect of functional potential in the high-score range. Its recommended value range is [0.8, 1.5]. When emphasizing functional evolution in the design, a larger value is taken. Here, β = 1.2 is selected and substituted into the formula to obtain:

[0100]

[0101] Finally, the transformation adaptation index SPI of the secondary bedroom B is obtained as 0.874, belonging to the high adaptation level. The SPI partition thresholds set by this method are as follows: SPI≥0.75 is the priority transformation area, SPI∈0.45,0.75 is the area for cautious handling, and SPI<0.45 is the area for retention or abandonment. According to this standard, space B is automatically divided into the priority transformation area. The system combines its evolution path and determines it as "a study use with high confidence", and suggests a light-structure function reconstruction for sound insulation optimization, lighting enhancement, and interface improvement. At the same time, a multi-strategy design plan is generated. Meanwhile, to verify the rationality of the adaptation index classification ability, the SPI value of the kitchen D node is calculated again. It is known that its structure limitation index S(D)=0.684 (i.e., the maximum value), and the function potential score is F(D)=2.9. After normalization, the structure ratio is 1, and the function ratio is 2.9 / 4.86≈0.597. If σ = 1.0 and β = 1.2 are also set and substituted into the formula, we get:

[0102] SPI(D)=(1 - 1.0·1.0)·(0.597) 1.2 =0·0.528 = 0

[0103] It shows that the kitchen D has insufficient transformation potential under the premise that the structure cannot be intervened. The system automatically divides it into the retention area and only suggests non-structural optimizations such as ventilation and lighting, and does not include it in the structural reconstruction strategy list. For another example, substituting the utility room F (S = 0.44, F = 3.4) gives SPIF≈0.59, which is the intermediate value. The system automatically classifies it as the "area for cautious handling", and it is necessary to comprehensively evaluate whether it is worth intervening by combining space fine-tuning strategies and structural optimizations. Through the above data and examples, the practicality of the SPI index of the present invention in complex old spaces is verified. It organically combines structural load analysis, semantic evolution reasoning, and non-linear decision-making mechanisms to form an efficient, accurate, and engineering-executable space transformation partition guidance model, providing an extensible and intelligent space evaluation tool for the transformation of a large number of old residential, office, and public building spaces in urban renewal.

[0104] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. An indoor renovation space identification and classification method for old buildings, characterized in that It includes the following steps: S1. Adopt an indoor structural function fuzzy recognition mechanism: S1.

1. Construct a space degradation factor model, including implicit variables such as the degree of enclosure, usage trace characteristics, and residual furniture layout; S1.

2. Use the internal texture features of the space, including wall peeling and pipeline exposure, for recognition, and detect abnormal layout patterns to determine whether the space has experienced usage degradation; and introduce a fuzzy function label mechanism to allow multiple usage tendencies of the space during the recognition stage; S2. Construct an indoor usage inference of an evolutionary space semantic network: S2.

1. Construct a space usage evolution map, including historical usage, current usage, and potential usage nodes; establish a semantic graph neural network SGNN, with the spatial physical attributes + connection relationships + usage evolution trajectories as inputs; S2.

2. Conduct usage inference and transformation direction prediction for the space, including inferring that a storage room can be transformed into a study or a laundry room; each space node is attached with a usage evolution credibility to assist decision-makers in screening transformation directions; S3. Adopt a space transformation adaptation index through coupling analysis: S3.

1. Construct a structural constraint model, calculated based on the layout of load-bearing walls, structural nodes, and beam-slab structure parameters; at the same time, call the evolution potential score in the spatial semantic map to form a dual-index evaluation matrix; S3.

2. Introduce a space transformation adaptation index SPI, generated through a non-linear fusion algorithm of structural bearing capacity × functional evolution potential; the adaptation index SPI provides three types of transformation suggestions for designers, namely priority transformation areas, areas that should be carefully handled, and areas to be reserved or abandoned; S4. Adopt a spatial adaptability enhancement reconstruction strategy: Based on the spatial adaptation index and usage prediction results, design a spatial reconstruction strategy generator; The strategy types include: space release type, function replacement type, transitional connection type, and micro-intervention optimization type.

2. The method for identifying and classifying the interior renovation space of old buildings according to claim 1, wherein The indoor usage inference process of constructing an evolutionary space semantic network includes: According to the historical drawings, on-site scanning data, and spatial usage traces of old buildings, model each indoor space unit, including rooms, corridors, and storage rooms, as nodes in a graph structure; each node is assigned a set of attribute vectors, covering geometric dimensions, physical states, usage traces, and historical usage, and analyze the physical connections, access paths, and shared functional elements with adjacent spaces to establish edge connection relationships between spaces.

3. The method for identifying and classifying the interior renovation space of old buildings according to claim 2, wherein The indoor usage inference process of constructing an evolutionary space semantic network includes: Based on node attributes and connection weights, construct a space usage evolution map, record the original design usage, current state, and potential transformable usage of each space, and organize the usage evolution process in a time series manner; the map is used as an input to enter the graph neural network model for training to learn the migration trend of space usage during the actual usage evolution process.

4. The method for identifying and classifying the interior renovation space of old buildings according to claim 3, wherein The indoor usage inference process of constructing an evolutionary space semantic network includes: Introduce a usage evolution credibility scoring function to measure the reliability and construction adaptability of each usage evolution suggestion, and the calculation is as follows: Where: C(i,k) is the credibility score for the evolution of space i towards use k; N is the total number of attribute types of the spaces participating in the evaluation, including lighting, ventilation, floor height, and area; ψ n (i,k) represents the fitness evaluation score of the nth attribute for space i under use k; w n is the importance weight factor of attribute n, determined by the user or design task parameters.

5. The method for identifying and classifying the interior renovation space of old buildings according to claim 1, wherein The construction method of the space transformation adaptation index through coupling analysis: Based on the three core elements of the bearing wall identification density, beam-column joint structure, and floor response characteristics, a structural constraint factor model is constructed for each indoor space; the inhibition intensity of the restrictive factors on the spatial transformability is quantified, and the structural constraint index S(i) of space i is defined as: Where: η p represents the influence weight of the structural factors including load-bearing walls, shear walls, and beam-column layouts in the p-th category on the spatial transformation restrictions; ξ ip is the actual strength performance of the space i on this factor, including wall thickness and layout density; ρ in the denominator i represents the structural complexity adjustment coefficient of the space, which is used to balance the deviation tendency in the calculation of spaces with many nodes and dense structures.

6. The method for identifying and classifying the interior renovation space of old buildings according to claim 5, wherein The method for constructing the spatial transformation adaptation index of the coupling analysis: Introduce a spatial use evolution potential evaluation model at the semantic level to measure the functional feasibility and sustainability of the space evolving from the existing state to a new use; the potential score depends on the static attributes of the space itself such as size, shape, lighting, and ventilation, and also includes the functional linkage with the surrounding space and the dynamic factors of the long-term use trajectory; the functional evolution potential scoring function F(i) of space i is defined as follows: Where: φ i (t) represents the elastic state of the use of space i at time t, which is used to describe the trend of work reconstruction under the current use; ω iq (t) is the response value of the q-th type of evolution factor including adjacent space use support degree, connection strength, and structural openness on space i, reflecting the degree of support of the environment for evolution; λ q is the influence weight of the q-th type of evolution factor, indicating the emphasis on the influence degree in different design scenarios; and the integration time period T represents the observation period of the functional evolution trend.

7. The method for identifying and classifying the interior renovation space of old buildings according to claim 6, wherein The method for constructing the spatial transformation adaptation index of the coupling analysis: After completing the two-dimensional modeling of the structural constraint S(i) and the functional potential F(i), a unified evaluation index, that is, the spatial transformation adaptation index SPI, is generated through non-linear fusion of the two types of indicators; the index takes the coupling of the functional potential and the structural constraint as the core evaluation logic and is defined as: Where: SPI(i) is the adaptation index of space i, used to evaluate whether the space is suitable for indoor functional transformation; σ is the structural suppression coefficient, used to control the weakening degree of the structural limitation on the overall index; S(i) is the aforementioned structural limitation value, and max(S) is the maximum value of the structural limitation in the entire building space, used for normalization; F(i) is the functional evolution potential score of the space, which is the average potential score of all spaces; β is the functional potential enhancement factor, which adjusts the priority score growth amplitude of high-potential spaces.

8. The method for identifying and classifying the interior renovation space of old buildings according to claim 7, wherein According to the SPI values of each space, the internal space of the building is divided into three categories: those with SPI values higher than the set upper threshold are marked as priority transformation areas; Spaces with SPI values in the middle gray area are classified as areas for cautious handling according to the fluctuation degree of the structural and functional scores; Spaces with SPI values far lower than the average level, indicating spaces with immovable structures and low evolution value, are classified into the reserved or abandoned areas.

Images