Unmanned aerial vehicle intelligent inspection path generation method for building interior based on bim model feature extraction

Abstract

The application discloses a kind of unmanned aerial vehicle intelligent inspection path generation methods for building interior based on BIM model feature extraction, comprising the following steps: extracting spatial geometric feature set, spatial topological feature set and semantic feature set from original BIM model data;Based on the feature set, the basic space graph representing space structure and the semantic enhanced graph representing functional relationship are constructed, and a multi-level semantic graph is generated;Task target and dependency are extracted from inspection task description and multi-level semantic graph, and a task weight matrix is generated;Based on multi-level semantic graph and task weight matrix, an optimized path set is generated;The optimized path set is scored in multiple dimensions and optimized in real time to generate an emergency adjustment strategy;The inspection path is finally verified.This method realizes accurate understanding of building space through multi-level feature extraction and multi-level semantic graph construction, improving the intelligence level and reliability of the inspection path planning.

Description

TECHNICAL FIELD

[0001] The present application relates to a UAV path generation method, in particular a UAV intelligent inspection path generation method for building interiors based on BIM model feature extraction. BACKGROUND

[0002] With the growing demand for intelligent building and facility management, indoor inspection systems based on unmanned aerial vehicles (UAVs) have gradually become an important direction of technological development. Such systems can autonomously complete tasks such as state monitoring, equipment inspection, and safety patrol within buildings, significantly reducing labor costs and improving inspection efficiency and safety. In particular, in complex environments such as large-scale integrated buildings, data centers, and industrial plants, UAV inspection can overcome the problems of high work intensity, high risk, and many dead angles in traditional manual inspection. At the same time, with the widespread application of Building Information Modeling (BIM) technology in building lifecycle management, using the rich spatial information provided by BIM models to guide indoor UAV inspection has become a key technical approach to improving the level of intelligent inspection.

[0003] Current indoor UAV inspection systems mainly use positioning and navigation methods based on laser SLAM and path planning strategies based on pre-set waypoints. In terms of spatial perception, real-time environmental information is mainly obtained through laser radar and visual sensors, and autonomous positioning is achieved through feature point matching and pose estimation. In terms of path planning, classical algorithms such as improved A* and RRT are mainly used, combined with basic collision detection to generate flight trajectories. In terms of task execution, fixed inspection routes and simple obstacle avoidance strategies are usually used. These methods can achieve basic inspection functions in simple indoor environments, but still face many challenges in complex building spaces.

[0004] The existing technology has the following specific problems: First, in terms of spatial feature extraction, existing methods often simplify building spaces into two-dimensional planes or rough three-dimensional grids, failing to fully utilize the component attributes, spatial relationships, and functional information contained in BIM models, resulting in insufficient understanding of complex spatial structures. Second, in terms of spatial representation, there is a lack of effective multi-level semantic modeling methods, which cannot accurately describe the mapping relationship between physical features and functional attributes of spaces, affecting the rationality of path planning. Third, in terms of path generation, existing algorithms mainly consider geometric constraints and distance optimization, and do not adequately consider multi-dimensional objectives such as task requirements, energy consumption, and execution efficiency, making it difficult to generate optimal paths that meet actual inspection requirements. Finally, in terms of path optimization, there is a lack of dynamic adjustment mechanism based on real-time environmental perception, which cannot effectively respond to sudden situations and environmental changes within buildings, affecting the reliability and safety of inspection tasks. These technical problems seriously restrict the actual application effect of indoor UAV inspection systems. SUMMARY

[0005] The application aims to provide a method for generating an intelligent inspection path of a UAV for an interior of a building based on feature extraction of a BIM model, so as to solve at least one technical problem existing in the prior art.

[0006] The technical scheme is a method for generating an intelligent inspection path of a UAV for an interior of a building based on feature extraction of a BIM model, comprising the following steps:

[0007] Step S1: extracting a space boundary point, a voxel state and a local descriptor from pre-stored original BIM model data to generate a space geometric feature set; constructing a space adjacency relationship and a key node set based on the space geometric feature set to generate a space topological feature set; combining the original BIM model data and the space topological feature set to extract a component attribute, a space function and a region importance information to generate a semantic feature set;

[0008] Step S2: constructing a basic space graph representing a space structure based on the space geometric feature set and the space topological feature set; constructing a semantic enhanced graph representing a functional relationship by using the semantic feature set and the basic space graph; and performing multi-level mapping and correlation on the basic space graph and the semantic enhanced graph to generate a multi-level semantic graph;

[0009] Step S3: extracting a task target and a dependency relationship from a pre-stored inspection task description and the multi-level semantic graph to generate a task subset; mapping the task subset to a space structure of the multi-level semantic graph to generate a task space mapping matrix; and calculating a multi-dimensional weight value based on the task subset and the task space mapping matrix to generate a task weight matrix;

[0010] Step S4: generating an initial candidate path set based on the multi-level semantic graph and the task weight matrix; combining the initial candidate path set and a pre-stored space constraint condition set to generate a feasible path set; and performing multi-objective optimization by using the feasible path set and the task weight matrix to generate an optimized path set;

[0011] Step S5: calculating a multi-dimensional score based on the optimized path set and a pre-stored evaluation index set to generate a path score matrix; selecting an optimal path by using the path score matrix and a pre-stored path constraint condition to generate an optimal inspection path; and designing a dynamic adjustment mechanism by combining the optimal inspection path and real-time environment data collected in real time to generate an emergency adjustment strategy;

[0012] Step S6: performing a physical feasibility test based on the optimal inspection path and a pre-stored space constraint condition set to generate a feasibility evaluation result; performing a safety evaluation by combining the optimal inspection path and the emergency adjustment strategy to generate a safety evaluation report; and comprehensively verifying the feasibility evaluation result and the safety evaluation report to generate a verified inspection path.

[0013] The beneficial effect is that, by multi-level feature extraction and multi-level semantic graph construction, accurate understanding of the building space is realized, and the intelligent level and reliability of the inspection path planning are improved. The related technical effects will be described in detail in combination with specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a flowchart of the present application.

[0015] Figure 2 is a flowchart of step S1 of the present application.

[0016] Figure 3 is a flowchart of step S2 of the present application.

[0017] Figure 4 is a flowchart of step S3 of the present application.

[0018] Figure 5 is a flowchart of step S4 of the present application.

[0019] Figure 6 is a flowchart of step S5 of the present application.

[0020] Figure 7 is a flowchart of step S6 of the present application. DETAILED DESCRIPTION

[0021] As shown in Figure 1 , a method for generating an intelligent inspection path of a UAV in a building based on feature extraction of a BIM model is provided, comprising the following steps:

[0022] Step S1, extracting spatial boundary points, voxel states, and local descriptors from pre-stored original BIM model data to generate a spatial geometric feature set; constructing a spatial adjacency relationship and a key node set based on the spatial geometric feature set to generate a spatial topological feature set; combining the original BIM model data and the spatial topological feature set to extract component attributes, spatial functions, and regional importance information to generate a semantic feature set;

[0023] Step S2, constructing a basic space graph representing the spatial structure based on the spatial geometric feature set and the spatial topological feature set; constructing a semantic enhanced graph representing the functional relationship using the semantic feature set and the basic space graph; performing multi-level mapping and correlation between the basic space graph and the semantic enhanced graph to generate a multi-level semantic graph;

[0024] Step S3, extracting task targets and dependency relationships from pre-stored inspection task descriptions and the multi-level semantic graph to generate a task subset; mapping the task subset to the spatial structure of the multi-level semantic graph to generate a task space mapping matrix; calculating multi-dimensional weight values based on the task subset and the task space mapping matrix to generate a task weight matrix;

[0025] Step S4, generating an initial candidate path set based on the multi-level semantic graph and the task weight matrix; generating a feasible path set by combining the initial candidate path set and a pre-stored space constraint condition set; performing multi-objective optimization using the feasible path set and the task weight matrix to generate an optimized path set;

[0026] Step S5, calculating a multi-dimensional score based on the optimized path set and a pre-stored evaluation index set to generate a path score matrix; selecting an optimal path using the path score matrix and a pre-stored path constraint condition to generate an optimal inspection path; designing a dynamic adjustment mechanism by combining the optimal inspection path and real-time environmental data collected in real time to generate an emergency adjustment strategy;

[0027] Step S6, performing physical feasibility verification based on the optimal inspection path and a pre-stored space constraint condition set to generate a feasibility evaluation result; performing safety evaluation by combining the optimal inspection path and the emergency adjustment strategy to generate a safety evaluation report; and performing final verification by integrating the feasibility evaluation result and the safety evaluation report to generate a verified inspection path.

[0028] By innovatively integrating BIM model feature extraction, multi-level semantic graph construction, multi-dimensional path planning, and real-time optimization and adjustment technologies, a complete indoor unmanned aerial vehicle intelligent inspection system is established. The scheme overcomes the problems of incomplete feature extraction, insufficient spatial expression, unreasonable path planning, and poor real-time adaptability in traditional inspection methods. Through multi-level extraction and fusion of features, a comprehensive understanding of the building space is achieved. Through multi-level construction of the semantic graph, the physical and functional relationships of the space are accurately expressed. Through multi-objective optimization and dynamic adjustment of the path, efficient completion of the inspection task is ensured. Through multiple verification and safety evaluation, the reliable operation of the system is ensured. The scheme significantly improves the intelligent level, task completion efficiency, and system reliability of indoor unmanned aerial vehicle inspection, and has important theoretical and practical value.

[0029] According to an aspect of the present application, step S1 is specifically:

[0030] Step S11, reading three-dimensional coordinate information from pre-stored original BIM model data to extract a space boundary point coordinate set; constructing a three-dimensional grid space based on the boundary point coordinate set, and calculating the occupancy state value of each voxel; calculating the local space description value for each boundary point; and combining the boundary point set, voxel state value, and local space description value to generate a space geometric feature set;

[0031] Step S12, calculate the connectivity value between the space regions based on the set of space geometric features, generate a space adjacency matrix; identify the three-dimensional coordinates of feature positions such as doors and corners from the set of space geometric features, generate a set of key nodes; construct a region connectivity graph based on the space adjacency matrix and the set of key nodes; combine the space adjacency matrix, the set of key nodes and the region connectivity graph to generate a set of space topological features;

[0032] Step S13, extract attribute information such as component type and material from the pre-stored original BIM model data to generate a component attribute set; determine the functional type label of each space region based on the region connectivity relationship of the set of space topological features; calculate the importance value of each region; combine the component attribute set, the functional type label and the importance value to generate a semantic feature set.

[0033] By innovatively decomposing the BIM model data into a three-layer structure of a set of space geometric features, a set of space topological features and a semantic feature set, the feature extraction of the interior space of the building is realized in all directions. Among them, the set of space geometric features not only retains the geometric information of the space through the combination of boundary points, voxel states and local descriptors, but also captures the micro features of the space through the local descriptors; the set of space topological features accurately describes the connectivity and key position information of the space through the space adjacency matrix and the key nodes; the semantic feature set combines component attributes, functional labels and importance information, so that the feature extraction contains both physical attributes and functional attributes. This multi-level and multi-dimensional feature extraction method overcomes the limitations of traditional methods that only focus on geometric features or only focus on semantic features, provides rich decision-making basis for subsequent path planning, and significantly improves the efficiency and accuracy of feature extraction and reduces the computational complexity of feature extraction.

[0034] According to one aspect of the present application, step S2 is specifically:

[0035] Step S21, construct a graph node set based on the space boundary information in the set of space geometric features and the set of key nodes in the set of space topological features; establish an edge set between nodes using the space adjacency matrix in the set of space topological features; assign a weight value to each node according to the importance of the space feature; combine the graph node set, the edge set and the node weight value to generate a basic space graph;

[0036] Step S22, construct a semantic node set based on the functional label information in the semantic feature set; establish an edge set between semantic nodes according to the correlation degree of space functions; calculate the weight value of the semantic node using the region importance information in the semantic feature set; combine the semantic node set, the semantic edge set and the semantic weight value to generate a semantic enhanced graph;

[0037] Step S23, establishing a node correspondence relationship matrix based on the basic space graph and the semantic enhanced graph; constructing a cross-layer edge set representing the association relationship of different level nodes; calculating the association strength weight of the cross-layer edge; and combining the basic space graph, the semantic enhanced graph, the node correspondence relationship matrix, the cross-layer edge set and the association strength weight to generate a multi-level semantic graph.

[0038] The innovative multi-level semantic graph construction method realizes the organic unification of spatial structure and functional relationship through the double-layer structure of the basic space graph and the semantic enhanced graph. In the basic space graph, the physical layout and connectivity of the space are accurately expressed through the construction of the node set and the edge set, combined with the allocation of the node weight value. In the semantic enhanced graph, the functional association of the space is effectively described through the construction of the semantic node set and the semantic edge set, combined with the calculation of the semantic weight value. The mapping relationship between the physical space and the functional space is established through the design of the node correspondence relationship matrix and the cross-layer edge set. This multi-level graph structure not only overcomes the problem that the traditional single graph structure is difficult to express the physical and functional relationship at the same time, but also realizes the complementation and enhancement of information through the association between levels, providing a decision basis for subsequent path planning considering both physical constraints and functional requirements. At the same time, through the hierarchical organization of the graph structure, the completeness and accuracy of space expression are greatly improved.

[0039] According to one aspect of the present application, step S3 is specifically:

[0040] Step S31, parsing a task target list from a pre-stored inspection task description; establishing a dependency relationship between tasks based on the semantic relationship of the multi-level semantic graph; extracting a timing constraint condition from the inspection task description; and combining the task target list, the task dependency relationship and the timing constraint condition to generate a task subset;

[0041] Step S32, establishing a correspondence relationship between tasks and space nodes based on the task subset and the multi-level semantic graph; calculating an access frequency value of each space node according to the task requirement; determining an access order constraint of the space node; and combining the task-space correspondence relationship, the access frequency value and the access order constraint to generate a task-space mapping matrix;

[0042] Step S33, calculating a task priority weight value based on the task subset; calculating a time urgency weight value according to the timing constraint in the task subset; calculating a resource consumption weight value based on the task-space mapping matrix; and combining the task priority weight value, the time urgency weight value and the resource consumption weight value to generate a task weight matrix.

[0043] The task decomposition and multi-dimensional weight calculation method are innovatively used to realize accurate modeling and optimal configuration of the inspection task. The complex inspection task is decomposed into a manageable task subset through task target analysis, dependency relationship establishment and time sequence constraint extraction. The task demand and space resource correspondence is established through the construction of the task space mapping matrix. The task priority, time urgency and resource consumption are comprehensively considered through multi-dimensional weight value calculation. The task modeling method overcomes the problem of simple task description in the traditional method. The task planning is more reasonable and efficient through fine task decomposition and multi-dimensional weight calculation. The task space mapping mechanism ensures the feasibility of task execution, and significantly improves the completion quality and efficiency of the inspection task.

[0044] According to an aspect of the present application, step S4 is specifically:

[0045] Step S41, constructing a basic path based on a multi-level semantic graph; performing path extension in combination with a task weight matrix; increasing path diversity through random sampling; and combining the basic path, the extended path and the randomly sampled path to generate an initial candidate path set;

[0046] Step S42, performing geometric feasibility verification based on a pre-stored space constraint condition set and the initial candidate path set; performing dynamic constraint verification on the paths passing the geometric verification; performing energy constraint evaluation on the paths passing the dynamic verification; and combining the paths passing all constraint verifications to generate a feasible path set;

[0047] Step S43, constructing a multi-objective optimization function based on the feasible path set; performing path optimization calculation using the task weight matrix; performing local search improvement on the optimized path; and combining the optimized and improved path to generate an optimized path set.

[0048] The multi-level semantic graph and the task weight matrix are innovatively combined, and a three-stage path generation strategy is adopted to realize efficient and reliable inspection path planning. First, the initial candidate path set with diversity is generated through the combination of basic path generation, path extension and random sampling. Then, the actual feasibility of the path is ensured through multiple verifications of geometric feasibility, dynamic constraint and energy constraint. Finally, the global optimization of the path is realized through multi-objective optimization and local search. The path planning method overcomes the problem that the traditional method is easy to fall into local optimization or ignores actual constraints. Through diversified path generation and multi-level path optimization, the quality of the planning result is ensured, and the efficiency of the planning is improved. Meanwhile, the multiple constraint verifications ensure the actual executability of the path.

[0049] According to an aspect of the present application, step S5 is specifically:

[0050] Step S51, calculating a path length score based on the optimized path set; calculating an energy consumption score according to a pre-stored evaluation index set; evaluating a task completion efficiency score; combining the path length score, the energy consumption score and the task completion efficiency score to generate a path score matrix;

[0051] Step S52, performing fuzzy comprehensive evaluation based on the path score matrix; making multi-objective decision in combination with pre-stored path constraint conditions; performing sensitivity analysis on the evaluation result; selecting and generating an optimal inspection path according to the analysis result;

[0052] Step S53, establishing a state monitoring index based on real-time environment data collected in real time; setting an adjustment trigger condition according to the optimal inspection path; constructing an emergency scheme database; combining the state monitoring index, the adjustment trigger condition and the emergency scheme data to generate an emergency adjustment strategy.

[0053] The method of combining multi-dimensional scoring and dynamic adjustment is innovatively adopted to realize accurate evaluation and real-time optimization of the inspection path. A comprehensive path evaluation system is established through comprehensive scoring of the path length, energy consumption and task completion efficiency. The scientific selection of the optimal path is realized through the method of fuzzy comprehensive evaluation and multi-objective decision. A perfect dynamic adjustment mechanism is established through the design of the state monitoring index, the adjustment trigger condition and the emergency scheme. The evaluation and optimization method overcomes the problems of single evaluation dimension and lack of adaptability in the traditional method, ensures the comprehensiveness and accuracy of the evaluation through multi-dimensional scoring, improves the robustness and adaptability of the system through the dynamic adjustment mechanism, and significantly improves the reliability and adaptability of the inspection system.

[0054] According to an aspect of the present application, step S6 is specifically:

[0055] Step S61, verifying kinematic constraint conditions based on the optimal inspection path; checking collision risks according to a pre-stored set of spatial constraint conditions; evaluating a passing performance index; combining the verification and evaluation results to generate a feasibility evaluation result;

[0056] Step S62, analyzing potential risk points based on the optimal inspection path; evaluating emergency response capability in combination with the emergency adjustment strategy; verifying a safety margin parameter; combining the analysis and evaluation results to generate a safety evaluation report;

[0057] Step S63, performing comprehensive evaluation based on the feasibility evaluation result and the safety evaluation report; performing necessary optimization adjustment on the problems found in the evaluation; confirming a final inspection path scheme; generating a verified inspection path.

[0058] The physical feasibility verification and safety evaluation method is combined innovatively to realize final confirmation and safety guarantee of the inspection path. Through kinematic constraint verification, collision risk checking and passing performance evaluation, the feasibility of the path in the physical layer is ensured. Through potential risk analysis, emergency response capability evaluation and safety margin verification, the safety of the path execution is guaranteed. Through comprehensive evaluation and necessary adjustment, the final optimization of the path scheme is realized. The verification and evaluation method overcomes the problem of ignoring the actual execution risk in the traditional method, and through multi-level feasibility verification and safety evaluation, the actual executability of the path is ensured, the safety of the execution is improved, and through the final comprehensive optimization, the integrity and reliability of the path scheme are further improved.

[0059] According to one aspect of the present application, step S11 is specifically:

[0060] Step S111, read the component geometric information from the pre-stored original BIM model data; perform boundary tracking processing on the geometric information to extract boundary vertices; perform spatial continuity analysis on the boundary vertices; organize and generate a spatial boundary point coordinate set from the processed boundary vertices;

[0061] Wherein, the boundary tracking is specifically: boundary point tracking value BTV(p) = μ1·G(p) + μ2·E(p)·exp(-D(p) / σ) + μ3·C(p); Wherein: G(p) is the gradient intensity; E(p) is the edge continuity; C(p) is the corner response; D(p) is the distance to the known boundary; μ1, μ2, μ3 are weight coefficients and Σ μi = 1; σ is the distance attenuation parameter.

[0062] In another embodiment of the present application, it also includes BIM data standardization, specifically: standardized BIM feature value NBF(x) = β1·F(x) + β2·S(x)·(1+α·Q(x)) + β3·R(x); Wherein: F(x) is the geometric feature value; S(x) is the semantic feature value; Q(x) is the data quality score; R(x) is the relationship completeness; β1, β2, β3 are weight coefficients and Σ βi = 1; α is the quality gain parameter.

[0063] Step S112, determine the spatial range boundary based on the spatial boundary point coordinate set; use an adaptive grid division method to construct a three-dimensional grid space; uniformly sample the three-dimensional grid space to generate a sampling point cloud; calculate the occupancy state value of each voxel based on the sampling point cloud;

[0064] Step S113, determine the calculation region based on the spatial boundary point coordinate set; construct a local coordinate system for each boundary point; calculate the geometric feature in the local coordinate system; convert the geometric feature into a local space description value;

[0065] wherein, the calculation of the local space description value is specifically: local space description value LSV(p) = α1·V(p) + α2·C(p) + α3·A(p) + α4·E(p); wherein: V(p) = Σ(vi·exp(-||p-pi||² / σ²)), V(p) is a voxel occupancy density feature; C(p) = |Σ((pi-p)×(pi+1-p))| / ||pi-p||·||pi+1-p||, C(p) is a curvature feature; A(p) = arccos(n(p)·nref) / π, A(p) is a normal angle feature; E(p) = -Σ(P(θi)·log(P(θi))), E(p) is a local entropy feature; p is a current boundary point; pi is a neighborhood point; vi is a voxel state value; n(p) is a normal vector; nref is a reference direction; P(θi) is a direction histogram probability; α1, α2, α3, α4 are feature weight coefficients, and Σαi=1; σ is a Gaussian kernel parameter.

[0066] In step S114, the space boundary point coordinate set, the occupancy state value and the local space description value are combined as features; the combined features are normalized; and the normalized features are combined to generate a space geometry feature set.

[0067] wherein, the normalization processing is specifically: normalized feature value NF(x) = γ1·tanh(β1(x-μx) / σx) + γ2·(1-exp(-β2·||x-xref||² / δ²)); wherein: x is an original feature value; μx is a feature mean value; σx is a feature standard deviation; xref is a reference feature value; γ1, γ2 are weight coefficients, and γ1+γ2=1; β1, β2 are scaling parameters; δ is a feature distance scale; tanh is a hyperbolic tangent function; this method combines statistical normalization and relative distance normalization, and improves the discrimination of the features.

[0068] The four-stage space geometry feature extraction method is innovatively adopted, and accurate geometry description of the interior space of a building is realized. Through boundary tracking processing and space continuity analysis, the space boundary point coordinate set is accurately extracted; through adaptive grid division and uniform sampling strategy, an accurate three-dimensional grid space model is established and the voxel occupancy state is calculated; through local coordinate system construction and geometry feature calculation, the local space description value of the boundary point is obtained; through feature combination and normalization processing, a standardized space geometry feature set is generated. This feature extraction method overcomes the problems of insufficient accuracy or high computational complexity of traditional methods, and through multi-stage feature extraction and processing, the high-precision acquisition of space geometry features is realized, and the standardization processing of the features ensures the reliability of subsequent processing, and significantly improves the accuracy and computational efficiency of space description.

[0069] According to an aspect of the present application, step S13 is specifically:

[0070] Step S131, read the component basic information from the pre-stored original BIM model data; convert the component basic information into standardized attributes using an attribute mapping table; organize the standardized attributes using a hierarchical classification method; establish an attribute dependency graph according to the correlation between the attributes; perform integrity verification based on the attribute dependency graph; combine the verified attributes to generate a component attribute set;

[0071] In another embodiment of the present application, a space rule library construction process is further included, specifically: rule weight value RW(r) = ζ1·H(r) + ζ2·U(r)·exp(-T(r) / φ) + ζ3·V(r); wherein: H(r) is the historical usage frequency; U(r) is the rule versatility; V(r) is the verification pass rate; T(r) is the time decay factor; ζ1, ζ2, ζ3 are weight coefficients and Σζi=1; φ is the time decay parameter.

[0072] Step S132, establish a space function association graph based on the region connectivity relationship in the space topological feature set; perform preliminary function identification using a pre-stored space function rule library; perform function verification according to the functional constraint relationship of adjacent spaces; calculate the function propagation probability using a space syntax analysis method; optimize the function type based on the function propagation probability; generate a function type label for each space region;

[0073] Wherein, the preliminary function identification is specifically: function identification score FRS(r) = λ1·T(r) + λ2·S(r) + λ3·M(r)·exp(-d(r) / ρ); wherein: T(r)= Σ(wt·ft(r)) is the topological feature score; S(r)= Σ(ws·fs(r)) is the semantic feature score; M(r)= Π(P(r|ri)) is the neighborhood propagation score; d(r) is the topological distance; ft(r) is the topological feature function; fs(r) is the semantic feature function; P(r|ri) is the conditional probability; wt, ws are feature weights; λ1, λ2, λ3 are combination weights; ρ is the distance decay parameter.

[0074] Step S133, read the function type label of each region; calculate the basic importance based on the pre-stored region level rules; calculate the region accessibility index in combination with the space topological feature set; calculate the region influence using a space centrality analysis method; calculate the usage intensity according to the region function density distribution; combine the basic importance, accessibility index, region influence and usage intensity by weighting to generate an importance value;

[0075] Wherein, the importance calculation process is specifically: the region importance value RIV(z) = η1·B(z) + η2·R(z)·(1+κ·C(z)) + η3·D(z); wherein: B(z) = Σ(wb·fb(z)) is the basic importance; R(z) = min(1,Σ(wr·N(z,r) / Nmax)) is the accessibility index; C(z) = |E(z)| / (|V(z)|·(|V(z)|-1)) is the centrality; D(z) = Σ(wd·fd(z)) is the density index; fb(z), fd(z) are evaluation functions; N(z,r) is the number of reachable nodes within the range of r; Nmax is the maximum number of nodes; wb, wr, wd are weight coefficients; η1, η2, η3 are combination weights, and Σηi=1; κ is the centrality gain coefficient.

[0076] Step S134, structurally arrange the component attribute set; map the function type label according to the region code; normalize the importance value; integrate the above data by using the feature fusion algorithm; perform consistency test on the integrated data; and generate the semantic feature set.

[0077] Wherein, the feature fusion algorithm is specifically: the fusion feature value FF(f) = π1·P(f) + π2·L(f)·(1+γ·M(f)) + π3·N(f); wherein: P(f) is the physical feature; L(f) is the logical feature; M(f) is the mapping completeness; N(f) is the correlation degree; π1, π2, π3 are weight coefficients and Σπi=1; γ is the mapping gain parameter.

[0078] The semantic feature extraction method of multi-source information fusion is innovatively used, and the overall semantic description of the building space is realized. Through hierarchical classification and attribute dependency graph construction, the standardized organization of component attributes is realized; through space function association graph and function propagation probability analysis, the function type of the space region is accurately identified; through multi-dimensional index fusion and weighted combination, the region importance is scientifically calculated; through feature fusion and consistency test, the complete semantic feature set is generated. This semantic feature extraction method overcomes the problem of incomplete or weak correlation of semantic information in traditional methods, realizes the overall acquisition of space semantic features through the systematic fusion of multi-source information, and the consistency test of the features ensures the accuracy of the semantic description, greatly improves the effect and reliability of the space function analysis.

[0079] According to one aspect of the present application, step S22 is specifically:

[0080] Step S221, extracting function tag information from the semantic feature set; performing semantic hierarchical processing on the function tag information; performing function normalization processing based on a semantic ontology library; hierarchically organizing the normalized function nodes; performing correlation degree analysis on the organized nodes; and organizing the processed nodes to generate a semantic node set;

[0081] Step S222, reading node pair information from the semantic node set; calculating semantic similarity between nodes based on a pre-stored function correlation rule library; calculating indirect correlation strength using a semantic transmission probability model; calculating comprehensive correlation degree by combining direct correlation and indirect correlation; screening effective connections according to a correlation degree threshold; and generating a semantic edge set from the screened node connection relationships.

[0082] The semantic similarity calculation process specifically includes:

[0083] The semantic similarity SS(i,j) = π1·H(i,j) + π2·O(i,j)·exp(-L(i,j) / τ) + π3·F(i,j); wherein H(i,j) = 1-|hi-hj| / hmax is the hierarchical similarity; O(i,j) = |Oi∩Oj| / |Oi∪Oj| is the ontology overlap degree; F(i,j) = Σ(wf·cos(fi,fj)) is the function similarity; L(i,j) is the shortest path length; hi,hj is the hierarchical depth; Oi,Oj is the ontology set; fi,fj is the function vector; π1,π2,π3 is the weight coefficient; and τ is the path attenuation parameter.

[0084] Step S223, extracting regional importance information based on the semantic feature set; calculating a basic weight by combining the hierarchical levels of the semantic nodes; calculating a structure weight based on the connection degree of the nodes; calculating a position weight using the spatial position relationship; fusing various weights using a multi-criteria weighting method; and generating a weight value of the semantic node.

[0085] The fusing of various weights specifically includes: fusing weight value FW(n) = ϕ1·B(n)·(1+μ·log(1+|En|)) + ϕ2·H(n)·exp(-D(n) / ξ) + ϕ3·P(n); wherein B(n) is the basic weight; H(n) is the hierarchical weight; P(n) is the position weight; |En| is the number of connection edges; D(n) is the distance to an important node; ϕ1,ϕ2,ϕ3 are fusion coefficients and Σϕi=1; μ is a connection degree gain parameter; and ξ is a distance attenuation coefficient.

[0086] Step S224, encoding and identifying the semantic node set; establishing an adjacency list representation for the semantic edge set; mapping the semantic weight values to the corresponding nodes; integrating the node, edge, and weight information using a graph structure organization method; performing consistency verification on the integrated graph structure; and generating a semantic enhanced graph.

[0087] The semantic enhanced graph construction method is innovatively used to realize the deep expression of spatial function relationship. A standard semantic node set is established through semantic layering and semantic ontology mapping; a complete semantic edge set is constructed through semantic similarity calculation and transmission probability model; the importance of the semantic node is determined through multi-criteria weight fusion; and a reliable semantic enhanced graph is generated through graph structure organization and consistency verification. This graph construction method overcomes the problem of insufficient expression of function relationship in traditional methods, realizes the accurate description of spatial function relationship through deep analysis and relationship construction at the semantic level, and ensures the completeness of semantic expression through the standard organization of graph structure, thereby significantly improving the accuracy and practicality of spatial function analysis.

[0088] According to an aspect of the present application, step S31 is specifically:

[0089] Step S311: reading the original task text from the pre-stored inspection task description; performing semantic word segmentation processing by using task parsing rules; identifying task key elements based on a pre-set task template; classifying and organizing the identification results; and organizing the task target list by using the organized task elements.

[0090] Step S312: extracting semantic relationship information from the multi-level semantic graph output by step S2; establishing a task node graph in combination with the task target list; analyzing the sequence between nodes based on a graph traversal algorithm; verifying the sequence rationality by using task logic rules; and generating the dependency relationship between tasks by using the verified sequence relationship.

[0091] Step S313: extracting time requirement information from the pre-stored inspection task description; analyzing the time window constraint based on a time sequence template; analyzing the execution time sequence requirement in combination with the task dependency relationship; detecting the conflict of the time sequence requirement; and generating the time sequence constraint condition by processing the time sequence requirement.

[0092] Step S314: structurally organizing the task target list output by step S311; converting the task dependency relationship output by step S312 into an adjacency matrix representation; normalizing the time sequence constraint condition output by step S313; combining the above data by using a task integration algorithm; performing integrity verification on the combination result; and generating a task subset.

[0093] The task decomposition and dependency relationship construction method is used innovatively to realize fine modeling of the inspection task. The task target elements are accurately identified through semantic word segmentation and template matching. The reliable task dependency relationship is established through graph traversal and logic verification. The reasonable execution constraint is determined through time sequence analysis and conflict detection. The standard task subset is generated through task integration and integrity inspection. The task modeling method overcomes the problem of rough task description or unclear constraint in the traditional method, realizes accurate definition of the inspection task through fine decomposition and systematic organization of the task, and ensures the feasibility of execution through strict verification of the relationship between tasks, thereby greatly improving the scientificity and operability of the task planning.

[0094] According to an aspect of the present application, step S41 is specifically:

[0095] Step S411: Extract the information of nodes and edges from the multi-level semantic graph output in step S2; calculate the shortest path between nodes using an improved Dijkstra algorithm; analyze the reachable path based on node connectivity; generate a path skeleton based on the topological structure of the graph; and organize the generated path skeleton into a basic path.

[0096] The improved Dijkstra algorithm is specifically: path cost PC(u,v) = θ1·L(u,v) + θ2·R(u,v)·(1+ε·S(u,v)) + θ3·W(u,v)·exp(-K(u,v) / ψ); wherein: L(u,v) is the path length; R(u,v) is the risk degree; S(u,v) is the smoothness; W(u,v) is the task weight; K(u,v) is the number of key points; θ1, θ2, θ3 are weight coefficients; ε is a smoothness gain parameter; and ψ is a key point attenuation coefficient.

[0097] Step S412: Read the task weight matrix output in step S3; perform path scoring on the basic path based on the weight value; generate a local mutation path using a path mutation algorithm; generate a combined path using a path recombination method; and generate an extended path from the mutation path and the combined path.

[0098] The path mutation algorithm is specifically: mutation path value MP(p) = χ1·O(p) + χ2·V(p)·exp(-C(p) / φ) + χ3·Q(p); wherein: O(p) is the original path score; V(p) is the mutation amplitude; C(p) is the constraint violation degree; Q(p) is the quality score; χ1, χ2, χ3 are weight coefficients and Σχi=1; and φ is a constraint attenuation parameter.

[0099] Step S413: Establish a sampling space based on the multi-level semantic graph output in step S2; generate sampling points using a hierarchical random sampling strategy; connect the sampling points using a probabilistic route graph method; verify the feasibility of the connected paths; and generate random sampling paths for the paths that pass the verification.

[0100] Step S414: Encode and identify the base paths output in step S411; structure and organize the extended paths output in step S412; normalize the random sampling paths output in step S413; combine the above paths using a path fusion algorithm; remove duplicate paths from the combined path set; and generate an initial candidate path set.

[0101] The diversified path generation strategy is used innovatively to efficiently construct the initial candidate paths. The base path skeleton is generated through the improved Dijkstra algorithm and connectivity analysis; the diversified extended paths are generated through path scoring and variation recombination; the supplementary random paths are obtained through hierarchical random sampling and feasibility verification; and the complete candidate path set is generated through path fusion and duplicate removal. This path generation method overcomes the problems of insufficient path diversity or low quality in traditional methods, and through the combined application of multiple strategies, it realizes the batch generation of high-quality candidate paths. At the same time, the normalization of the paths ensures the efficiency of subsequent optimization, significantly improving the initial quality and convergence speed of the path planning.

[0102] According to one aspect of the present application, step S42 is specifically:

[0103] Step S421: Read the geometric constraint parameters from the pre-stored spatial constraint condition set; extract the path geometric features based on the initial candidate path set output in step S41; verify the clearance using a collision detection algorithm; test the turning radius using a curvature analysis method; verify the vertical spacing based on a spatial subdivision algorithm; and generate a geometric feasible path set for the paths that pass the verification.

[0104] Step S422: Read the path information from the geometric feasible path set; extract the dynamic parameters based on the pre-stored spatial constraint condition set; calculate the speed constraint using a kinematics model; verify the maneuverability using an acceleration analysis method; test the balance stability according to the attitude constraint; and generate a dynamic feasible path set for the paths that pass the verification.

[0105] Wherein, the dynamic constraint value DC(v) = θ1·K(v) + θ2·A(v)·(1+ε·J(v)) + θ3·E(v); wherein: K(v) is the kinematics constraint; A(v) is the acceleration constraint; J(v) is the jerk constraint; E(v) is the energy constraint; θ1, θ2, θ3 are weight coefficients and Σθi=1; ε is the smoothness parameter.

[0106] Step S423: Extract path segments from the set of power feasible paths; read energy consumption parameters based on the pre-stored set of spatial constraints; calculate basic power consumption using the energy consumption model; estimate additional energy consumption in combination with path characteristics; verify the endurance capability using energy balance analysis; generate the set of energy feasible paths for the paths that pass the verification.

[0107] Step S424: Classify and organize the various feasible paths output by steps S421 to S423; perform feasibility level evaluation on the paths; perform path screening based on comprehensive scores; fine-tune the screened paths using path optimization algorithms; perform final verification on the adjusted paths; generate the set of feasible paths.

[0108] The multi-level constraint verification method is innovatively used to realize comprehensive feasibility evaluation of the inspection path. The geometric feasibility of the path is verified through collision detection and curvature analysis; the kinematic feasibility of the path is verified through kinematic model and acceleration analysis; the energy feasibility of the path is verified through energy consumption model and energy balance analysis; the final set of feasible paths is generated through feasibility level evaluation and path fine-tuning. This constraint verification method overcomes the problem of incomplete constraint consideration or insufficient verification in traditional methods, and through strict verification and system optimization in multiple dimensions, the comprehensive guarantee of path feasibility is realized, and the practicality of the verification result is ensured through the fine-tuning mechanism of the path, which significantly improves the success rate and reliability of path execution.

[0109] According to one aspect of the present application, step S43 is specifically:

[0110] Step S431: Extract path feature data from the set of feasible paths output by step S42; construct an objective function based on path length, smoothness and complexity; generate constraint conditions in combination with the task weight matrix output by step S3; establish a sub-objective function using objective decomposition method; combine the objective function and constraint conditions to generate a multi-objective optimization function.

[0111] Step S432: Calculate the initial score of the path based on the multi-objective optimization function; generate a path optimization scheme using genetic algorithm; screen the superior solution using Pareto sorting method; update the optimization scheme in combination with the elite reservation strategy; generate an optimized path set from the optimized scheme.

[0112] Step S433: Extract path node sequences from the optimized path set; select the improvement area based on the variable neighborhood search strategy; optimize the local trajectory using the path smoothing algorithm; adjust the node distribution using the dynamic programming method; generate an improved path set from the improved path.

[0113] Step S434: sorting and classifying the optimized path set output by step S432; scoring and sorting the improved path set output by step S433; integrating the optimized paths using a path merging algorithm; eliminating redundancies of the integrated paths; and combining the processed paths to generate an optimized path set.

[0114] The multi-objective optimization and local search method is used innovatively to realize accurate optimization of the inspection path. A complete optimization function system is established through target decomposition and constraint construction. Global optimization of the path is realized through genetic algorithm and Pareto sorting. Fine adjustment of the local trajectory is completed through variable neighborhood search and path smoothing. The final optimized path set is generated through path merging and redundancy elimination. This path optimization method overcomes the problem of easy falling into local optimum or insufficient optimization of the traditional method. Through the organic combination of global optimization and local adjustment, the quality of the path is significantly improved. The redundancy elimination mechanism ensures the simplicity of the optimization result, greatly improving the efficiency and effect of path planning.

[0115] According to one aspect of the present application, step S51 is specifically:

[0116] Step S511: extracting path geometry data from the optimized path set output by step S4; calculating the length of each path segment using a path segmentation algorithm; calculating the path smoothness value based on curvature analysis; calculating the complexity value in combination with the number of turns; and combining the path length, smoothness value and complexity value to generate a path length score.

[0117] The path length score is specifically: path score PS(r) = ω1·D(r)·(1-σ·T(r)) + ω2·S(r)·exp(-C(r) / ζ) + ω3·E(r); where: D(r) is the distance score; T(r) is the number of turns; S(r) is the smoothness score; C(r) is the complexity; E(r) is the efficiency score; ω1, ω2, ω3 are weight coefficients and Σωi=1; σ is the turn penalty coefficient; ζ is the complexity decay parameter.

[0118] Step S512: extracting energy consumption evaluation criteria based on the pre-stored evaluation index set; extracting motion parameters from the optimized path set; calculating the basic energy consumption value using the energy consumption model; estimating the dynamic energy consumption value according to the path characteristics; and combining the basic energy consumption value and the dynamic energy consumption value to generate an energy consumption score.

[0119] Step S513: extracting the task execution sequence from the optimized path set; obtaining the time evaluation parameters based on the pre-stored evaluation index set; calculating the waiting time value between tasks; analyzing the influence of path overlap on efficiency; and combining the task completion time and path utilization to generate a task completion efficiency score.

[0120] Step S514: normalizing the path length score output by step S511; standardizing the energy consumption score output by step S512; quantitatively mapping the task completion efficiency score output by step S513; integrating the scores by using a score combination algorithm; and generating a path score matrix from the integrated scores.

[0121] The multi-dimensional path scoring method is used innovatively to achieve scientific evaluation of the inspection path. The accurate path length score is calculated through path segmentation and curvature analysis; the reliable energy consumption score is obtained through the energy consumption model and dynamic estimation; the reasonable completion efficiency score is determined through task analysis and efficiency evaluation; and the complete path score matrix is generated through score combination and normalization processing. This scoring method overcomes the problem of single evaluation dimension or unreasonable index setting in the traditional method, and realizes accurate measurement of path quality through comprehensive evaluation of multiple dimensions. At the same time, the standardization processing of the score ensures the comparability of the evaluation result, significantly improves the scientificity and accuracy of the path selection.

[0122] According to one aspect of the present application, step S53 is specifically:

[0123] Step S531: extracting environmental parameters from real-time environmental data collected in real time; processing the environmental parameters by using a multi-sensor data fusion algorithm; calculating the parameter change trend based on a sliding window method; identifying environmental mutations by using an anomaly detection method; and generating state monitoring indicators from the processed environmental indicators.

[0124] The multi-sensor data fusion includes: fusion data value FD(t) = υ1·M(t)·exp(-ΔT / τ) + υ2·N(t)·(1+ρ·V(t)) + υ3·A(t); wherein: M(t) is the main sensor data; N(t) is the auxiliary sensor data; A(t) is the environmental data; ΔT is the time difference; V(t) is the data validity; υ1, υ2, υ3 are weight coefficients and Συi=1; τ is the time decay parameter; and ρ is the validity gain coefficient.

[0125] Step S532: extracting path features from the optimal inspection path output by step S52; determining key monitoring points based on path segmentation analysis; setting parameter warning values by using a threshold calculation method; establishing trigger rules in combination with environmental change features; and generating adjustment trigger conditions from the monitoring points, warning values, and trigger rules.

[0126] In another embodiment of the present application, the data is also updated, and the update frequency is specifically: update frequency value UF(t) = η1·C(t) + η2·V(t)·exp(-D(t) / ρ) + η3·P(t); wherein: C(t) is the calculation complexity; V(t) is the data change rate; P(t) is the priority; D(t) is the delay time; η1, η2, η3 are weight coefficients and Σηi = 1; and ρ is a delay attenuation parameter.

[0127] Step S533: establishing a scheme framework based on a pre-stored emergency plan template; constructing a set of alternative schemes in combination with historical emergency data; establishing an emergency response rule by using a scene matching method; screening effective schemes by using a scheme evaluation model; and generating emergency scheme data from the screened schemes.

[0128] Step S534: classifying the state monitoring indicators output from step S531; performing a rule-based processing on the adjustment trigger conditions output from step S532; performing a scene mapping on the emergency scheme data output from step S533; combining the monitoring, trigger and emergency data by using a strategy integration algorithm; and generating an emergency adjustment strategy from the integrated data.

[0129] The adaptive adjustment strategy is innovatively used to realize the real-time optimization of the inspection path. The reliable state monitoring indicators are established through multi-sensor fusion and anomaly detection. The reasonable adjustment trigger conditions are determined through path analysis and early warning setting. The effective emergency scheme library is constructed through template matching and evaluation screening. The complete emergency adjustment strategy is generated through strategy integration and scene mapping. This adjustment method overcomes the problems of poor adaptability or delayed response of the traditional method. Through the mechanism of real-time monitoring and rapid response, the dynamic optimization of the path is realized. Meanwhile, the system design of the emergency scheme ensures the reliability of the adjustment, and greatly improves the robustness and adaptability of the system.

[0130] According to an aspect of the present application, step S62 is specifically:

[0131] Step S621: extracting trajectory parameters from the optimal inspection path output from step S5; identifying a spatial risk area by using a risk evaluation model; analyzing a collision risk probability based on path characteristics; calculating a passing risk level in combination with environmental constraints; and generating a potential risk point from the risk area, risk probability and risk level.

[0132] Step S622: reading an emergency rule from the emergency adjustment strategy output from step S5; calculating a response time parameter based on rule matching; analyzing disposal efficiency by using a capability evaluation model; verifying scheme reliability in combination with historical data; and generating an emergency response capability value from the response time, disposal efficiency and reliability data.

[0133] The emergency response capability value is specifically: response capability value RC(e) = ψ1·T(e)·exp(-D(e) / δ) + ψ2·P(e)·(1+κ·R(e)) + ψ3·H(e); wherein: T(e) is a response time score; P(e) is disposal efficiency; H(e) is a historical success rate; D(e) is a distance factor; R(e) is resource sufficiency; ψ1, ψ2, ψ3 are weight coefficients and Σψi=1; δ is a distance attenuation parameter; κ is a resource gain coefficient.

[0134] Step S623: extracting key safety parameters based on potential risk points; analyzing spatial margins using a safety distance calculation model; calculating motion margins in combination with kinetic constraints; evaluating residual margins using energy states; and generating safety margin parameters from the spatial margins, motion margins, and residual margins.

[0135] Step S624: classifying the potential risk point data output from step S621; quantitatively evaluating the emergency response capability values output from step S622; threshold analyzing the safety margin parameters output from step S623; integrating the evaluation data using a report generation template; and generating a safety evaluation report from the integrated data.

[0136] Innovatively, a comprehensive safety evaluation method is used to achieve risk prevention and control of the inspection path. Through risk assessment and probability analysis, potential dangerous points are identified; through response time and efficiency evaluation, emergency disposal capability is verified; through multi-dimensional margin analysis, the safety boundary of execution is ensured; through report generation and data integration, system safety evaluation is completed. This evaluation method overcomes the problem of inaccurate risk prediction or inadequate prevention and control measures in traditional methods, and through comprehensive risk assessment and system safety verification, it realizes the safety guarantee of path execution. At the same time, the standardization of the evaluation report ensures the traceability of safety management, significantly improving the safety and reliability of the inspection system.

[0137] In summary, to address the problem of insufficient spatial feature extraction, a three-layer feature extraction architecture (spatial geometric feature set, spatial topological feature set, and semantic feature set) is used to achieve deep analysis of BIM model data. Specifically: precise geometric feature extraction is achieved by combining boundary tracking processing and local spatial description value calculation; complete topological relationships are established through spatial adjacency matrices and key node identification; hierarchical classification and attribute dependency graphs are used to achieve systematic extraction of component attributes, spatial functions, and regional importance. This multi-level feature extraction method ensures comprehensive acquisition and accurate understanding of architectural space features.

[0138] To address the problem of inaccurate spatial representation, a multi-level semantic graph construction method is used to realize the unified representation of physical space and functional space. A double-layer structure of basic space graph and semantic enhanced graph is designed, and an accurate mapping of physical features and functional attributes is established through a node correspondence matrix and a cross-layer edge set. This multi-level graph structure not only accurately represents the physical layout and functional association of the space, but also improves the completeness of the space representation through information complementation between levels.

[0139] To address the problem of unreasonable path generation, a complete path generation system is established. Through task decomposition and multi-dimensional weight calculation, the inspection requirements are accurately modeled; a three-stage strategy of "basic path generation - multi-constraint verification - multi-objective optimization" is adopted, considering multiple dimensions such as geometric constraints, dynamic constraints, energy constraints, etc., to ensure the practicality and superiority of the generated path.

[0140] To address the problem of poor dynamic adaptability, a perfect real-time optimization mechanism is constructed. A state monitoring system based on multi-sensor data fusion is designed, combined with adjustment trigger conditions and emergency scheme database, to realize intelligent dynamic adjustment of the path; through comprehensive safety evaluation and emergency response mechanism, the reliable operation of the system in the face of environmental changes is ensured.

[0141] Through the three-layer feature system, the complete extraction from geometric features to functional semantics is realized. Through the multi-level semantic graph, a unified model of physical space and functional space is established. Through the multi-stage strategy, the rationality and feasibility of path planning are ensured. Through real-time monitoring and dynamic adjustment, the robustness and reliability of the system are improved.

[0142] Not only overcomes the limitations of existing technology, but also significantly improves the overall performance and practical value of indoor unmanned aerial vehicle inspection system. Through the organic combination of feature extraction, space representation, path generation and dynamic optimization, the intelligentization and high efficiency of indoor unmanned aerial vehicle inspection are realized.

[0143] In another embodiment of the present application, the data processing brief process is as follows:

[0144] S1, multi-dimensional feature extraction of BIM model

[0145] S11: geometric feature extraction

[0146] Obtain the original BIM model data, extract the space boundary point set P = {p1, p2,..., pn}, where pi = (xi, yi, zi) represents the three-dimensional coordinates of the boundary point; construct the space voxel matrix V, V(i, j, k) ∈ {0, 1} represents the voxel occupancy state; calculate the local space descriptor L(p) = {l1, l2,..., lm}, where li represents the local geometric feature of point p.

[0147] Generate space geometry feature set G = {P, V, L}

[0148] S12: Topology feature extraction

[0149] Read space geometry feature set G, construct space adjacency matrix A, Aij represents the connectivity of space regions i and j;

[0150] Extract key node set K = {k1, k2,..., kr}, including feature points such as doors and corners;

[0151] Calculate region connectivity graph C, C = (V, E), V is the space region set, E is the connectivity set;

[0152] Generate space topology feature set T = {A, K, C}

[0153] S13: Semantic feature extraction

[0154] Get original BIM model data, space topology feature set T;

[0155] Extract component attribute set M = {m1, m2,..., mt}, including component type, material, etc.

[0156] Construct space function label set F = {f1, f2,..., fu};

[0157] Establish region importance score R = {r1, r2,..., rv};

[0158] Generate semantic feature set S = {M, F, R}.

[0159] S2: Multi-level semantic graph construction

[0160] S21: Basic graph layer construction

[0161] Read space geometry feature set G, space topology feature set T;

[0162] Construct graph node set Vb based on key node set K;

[0163] Establish edge set Eb based on space adjacency matrix A;

[0164] Assign node weight wv based on space feature importance;

[0165] Generate basic space graph Gb = (Vb, Eb, wv);

[0166] S22: Semantic layer construction

[0167] Get semantic feature set S, basic space graph Gb;

[0168] Construct semantic node set Vs, integrate spatial function labels;

[0169] Establish semantic edge set Es, representing function association relationship;

[0170] Calculate semantic weight ws, based on regional importance score;

[0171] Generate semantic enhanced graph Gs=(Vs, Es, ws);

[0172] S23: inter-layer relationship mapping

[0173] Obtain basic spatial graph Gb and semantic enhanced graph Gs;

[0174] Establish inter-layer mapping matrix H, Hij represents the corresponding relationship between nodes in two layers;

[0175] Construct cross-layer edge set Ec, representing the association between different layers;

[0176] Calculate cross-layer weight wc, reflecting the strength of inter-layer relationship;

[0177] Generate multi-level semantic graph G={Gb, Gs, H, Ec, wc};

[0178] S3: inspection task demand analysis

[0179] S31: task decomposition

[0180] Obtain inspection task description and multi-level semantic graph G;

[0181] Parse task objective set O={o1, o2,..., ox};

[0182] Establish task dependency relationship D={d1, d2,..., dy};

[0183] Determine task timing constraint Ct={t1, t2,..., tz};

[0184] Generate task subset T={O, D, Ct};

[0185] S32: space mapping

[0186] Obtain task subset T and multi-level semantic graph G;

[0187] Establish task-space correspondence relationship;

[0188] Calculate spatial access frequency;

[0189] Determine spatial access order constraint;

[0190] Generate task-space mapping matrix M;

[0191] S33: weight calculation

[0192] Obtain task subset T, task space mapping matrix M;

[0193] Calculate task priority weight wp;

[0194] Evaluate time urgency weight wt;

[0195] Estimate resource consumption weight wr;

[0196] Generate task weight matrix W = {wp, wt, wr};

[0197] S4: Adaptive path planning model construction

[0198] S41: Candidate path generation

[0199] Obtain multi-level semantic graph G, task weight matrix W;

[0200] Apply improved A* algorithm to generate basic path;

[0201] Expand path set using probabilistic roadmap method;

[0202] Integrate random sampling to increase diversity;

[0203] Generate initial candidate path set P0;

[0204] S42: Dynamic constraint processing

[0205] Obtain initial candidate path set P0, space constraint condition set C;

[0206] Check geometric feasibility constraints; evaluate dynamic constraints; consider energy constraints; generate feasible path set P1;

[0207] S43: Adaptive optimization

[0208] Obtain feasible path set P1, task weight matrix W;

[0209] Construct multi-objective optimization function; apply improved genetic algorithm optimization; integrate local search strategy; generate optimized path set P2;

[0210] S5: Path optimization and dynamic adjustment

[0211] S51: Path evaluation

[0212] Obtain optimized path set P2, evaluation index set I; calculate path length score sl; evaluate energy consumption score se; estimate task completion efficiency score st; generate path score matrix S = {sl, se, st};

[0213] S52: Optimal path selection

[0214] Obtain the path score matrix S, the path constraint condition C; apply the fuzzy comprehensive evaluation method; consider the multi-objective decision weight; perform the sensitivity analysis; determine the optimal inspection path P*;

[0215] S53: Dynamic adjustment mechanism

[0216] Input: optimal inspection path P*, real-time environment data E Output: emergency adjustment strategy D Processing process: establish a state monitoring mechanism; design a trigger condition threshold; build an alternative scheme library; generate an emergency adjustment strategy D;

[0217] S6: Path feasibility verification

[0218] S61: Physical feasibility test

[0219] Obtain the optimal inspection path P*, the set of spatial constraint conditions C; verify the kinematic constraint; check the collision risk; evaluate the passing performance; generate the feasibility evaluation result R1;

[0220] S62: Safety evaluation

[0221] Obtain the optimal inspection path P*, the emergency adjustment strategy D; analyze the potential risk points; evaluate the emergency response capability; verify the safety margin; generate the safety evaluation report R2;

[0222] S63: Final verification

[0223] Obtain the feasibility evaluation result R1, the safety evaluation report R2; comprehensively evaluate the results; adjust and optimize as necessary; confirm the final scheme; output the verified inspection path Pf.

[0224] The preferred embodiments of the present application are described in detail above, but the present application is not limited to the specific details in the above-described embodiments, and various equivalent transformations can be performed on the technical solutions of the present application within the technical concept range of the present application, and these equivalent transformations all belong to the protection range of the present application.

Claims

1. A method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction, characterized in that, Includes the following steps: Step S1: Extract spatial boundary points, voxel states, and local descriptors from the pre-stored original BIM model data to generate a spatial geometric feature set; construct spatial adjacency relationships and key node sets based on the spatial geometric feature set to generate a spatial topological feature set; combine the original BIM model data and the spatial topological feature set to extract component attributes, spatial functions, and regional importance information to generate a semantic feature set. Step S2: Construct a basic spatial graph representing the spatial structure based on the spatial geometric feature set and the spatial topological feature set; A semantic enhancement graph representing functional relationships is constructed using the semantic feature set and the basic spatial graph; the basic spatial graph and the semantic enhancement graph are mapped and associated at multiple levels to generate a multi-level semantic graph; Step S3: Extract task objectives and dependencies from the pre-stored inspection task description and multi-level semantic graph to generate a task subset; The task subset is mapped onto the spatial structure of a multi-level semantic graph to generate a task space mapping matrix; based on the task subset and the task space mapping matrix, multi-dimensional weight values ​​are calculated to generate a task weight matrix. Step S4: Generate an initial candidate path set based on the multi-level semantic graph and the task weight matrix; combine the initial candidate path set with the pre-stored spatial constraint set to generate a feasible path set; use the feasible path set and the task weight matrix to perform multi-objective optimization and generate an optimized path set. Step S5: Calculate multi-dimensional scores based on the optimized path set and the pre-stored evaluation index set to generate a path score matrix; select the optimal path using the path score matrix and the pre-stored path constraints to generate the optimal inspection path; design a dynamic adjustment mechanism based on the optimal inspection path and the real-time environmental data collected in real time to generate an emergency adjustment strategy. Step S6: Perform physical feasibility verification based on the optimal inspection path and the pre-stored set of spatial constraints, and generate feasibility assessment results; A safety assessment is conducted by combining the optimal inspection path and emergency adjustment strategy, and a safety assessment report is generated. The feasibility assessment results and safety assessment report are combined for final verification to generate the verified inspection path.

2. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S1 is as follows: Step S11: Read the three-dimensional coordinate information from the pre-stored original BIM model data and extract the spatial boundary point coordinate set; construct a three-dimensional grid space based on the boundary point coordinate set and calculate the occupancy status value of each voxel; calculate the local spatial description value for each boundary point; combine the boundary point set, voxel status value and local spatial description value to generate a spatial geometric feature set; Step S12: Calculate the connectivity values ​​between spatial regions based on the spatial geometric feature set to generate a spatial adjacency matrix; identify the three-dimensional coordinates of feature positions such as doors and corners from the spatial geometric feature set to generate a key node set; construct a region connectivity graph based on the spatial adjacency matrix and the key node set; combine the spatial adjacency matrix, the key node set, and the region connectivity graph to generate a spatial topological feature set; Step S13: Extract component type, material and other attribute information from the pre-stored original BIM model data to generate a component attribute set; Based on the regional connectivity of the spatial topological feature set, the functional type label of each spatial region is determined; the importance value of each region is calculated; and the component attribute set, functional type label and importance value are combined to generate a semantic feature set.

3. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S2 is as follows: Step S21: Construct a graph node set based on the spatial boundary information in the spatial geometric feature set and the key node set in the spatial topological feature set; establish an edge set between nodes using the spatial adjacency matrix in the spatial topological feature set; assign a weight value to each node according to the importance of the spatial features; combine the graph node set, edge set, and node weight values ​​to generate a basic spatial graph; Step S22: Construct a semantic node set based on the functional label information in the semantic feature set; establish an edge set between semantic nodes according to the degree of spatial function association; calculate the weight value of the semantic nodes using the regional importance information in the semantic feature set; The semantic node set, semantic edge set, and semantic weight values ​​are combined to generate a semantically enhanced graph; Step S23: Establish a node correspondence matrix based on the basic spatial graph and the semantic enhancement graph; construct a cross-layer edge set representing the association relationship between nodes at different levels; calculate the association strength weight of the cross-layer edges; The basic spatial graph, semantic enhancement graph, node correspondence matrix, cross-layer edge set, and association strength weight are combined to generate a multi-level semantic graph.

4. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S3 is as follows: Step S31: Parse the task target list from the pre-stored inspection task description; establish the dependency relationship between tasks based on the semantic relationship of the multi-level semantic graph; extract the temporal constraint conditions from the inspection task description; The task target list, task dependencies, and timing constraints are combined to generate a task subset; Step S32: Establish the correspondence between tasks and spatial nodes based on task subsets and multi-level semantic graphs; Calculate the access frequency value of each spatial node according to the task requirements; determine the access order constraints of the spatial nodes; combine the task space correspondence, access frequency value and access order constraints to generate a task space mapping matrix; Step S33: Calculate task priority weight values ​​based on task subsets; calculate time urgency weight values ​​based on the time constraints in the task subsets; calculate resource consumption weight values ​​based on the task space mapping matrix; combine the task priority weight values, time urgency weight values, and resource consumption weight values ​​to generate a task weight matrix.

5. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S4 is as follows: Step S41: Construct basic paths based on multi-level semantic graphs; expand paths by combining task weight matrices; increase path diversity through random sampling; combine the basic paths, expanded paths, and randomly sampled paths to generate an initial candidate path set; Step S42: Perform a geometric feasibility check based on the pre-stored set of spatial constraints and the initial candidate path set; Perform dynamic constraint verification on the paths that pass the geometric test; Energy constraint assessments are performed on kinetically validated paths; A feasible path set is generated by combining paths that pass all constraint verifications; Step S43: Construct a multi-objective optimization function based on the feasible path set; The path optimization calculation is performed using the task weight matrix; the optimized path is improved by local search; and the optimized and improved paths are combined to generate an optimized path set.

6. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S5 is as follows: Step S51: Calculate the path length score based on the optimized path set; calculate the energy consumption score based on the pre-stored evaluation index set; evaluate the task completion efficiency score; combine the path length score, energy consumption score, and task completion efficiency score to generate a path scoring matrix; Step S52: Perform fuzzy comprehensive evaluation based on the path scoring matrix; make multi-objective decisions by combining pre-stored path constraints; conduct sensitivity analysis on the evaluation results; select and generate the optimal inspection path based on the analysis results; Step S53: Establish status monitoring indicators based on real-time environmental data collected in real time; set adjustment trigger conditions according to the optimal inspection path; construct an emergency plan database; combine the status monitoring indicators, adjustment trigger conditions and emergency plan data to generate an emergency adjustment strategy.

7. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 1, characterized in that, Step S6 is as follows: Step S61: Verify kinematic constraints based on the optimal inspection path; check collision risks according to the pre-stored set of spatial constraints; evaluate the pass performance indicators; combine the verification and evaluation results to generate a feasibility evaluation result. Step S62: Analyze potential risk points based on the optimal inspection path; Assess emergency response capabilities by incorporating emergency adjustment strategies; Verify the safety margin parameters; The analysis and evaluation results are combined to generate a security assessment report; Step S63: Conduct a comprehensive evaluation based on the feasibility assessment results and safety assessment report; make necessary optimizations and adjustments to address the issues identified in the assessment. Confirm the final inspection route plan; generate the verified inspection route.

8. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 2, characterized in that, Step S11 is as follows: Step S111: Read the component geometric information from the pre-stored original BIM model data; perform boundary tracing processing on the geometric information to extract boundary vertices; perform spatial continuity analysis on the boundary vertices; organize the processed boundary vertices to generate a set of spatial boundary point coordinates; Step S112: Determine the spatial boundary based on the set of spatial boundary point coordinates; construct a three-dimensional grid space using an adaptive mesh generation method; The three-dimensional grid space is uniformly sampled to generate a sampling point cloud; The occupancy status value of each voxel is calculated based on the sampling points; Step S113: Determine the computational region based on the set of spatial boundary point coordinates; construct a local coordinate system for each boundary point; calculate geometric features in the local coordinate system; convert the geometric features into local spatial description values; Step S114: Combine the spatial boundary point coordinate set, occupancy state value and local spatial description value into features; normalize the combined features; and generate a spatial geometric feature set by combining the normalized features.

9. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 2, characterized in that, Step S13 is as follows: Step S131: Read the basic information of components from the pre-stored original BIM model data; convert the basic information of components into standardized attributes using the attribute mapping table; A hierarchical classification method is used to organize standardized attributes; Establish an attribute dependency graph based on the relationships between attributes; perform integrity verification based on the attribute dependency graph; combine the verified attributes to generate a component attribute set; Step S132: Establish a spatial function association graph based on the regional connectivity relationships in the spatial topological feature set; perform preliminary function identification using a pre-stored spatial function rule base; and verify the function based on the functional constraint relationships between adjacent spaces. The spatial parsing method is used to calculate the function propagation probability; the function type is optimized based on the function propagation probability; and a function type label is generated for each spatial region. Step S133: Read the functional type label of each region; calculate the basic importance based on the pre-stored region level rules; calculate the region accessibility index by combining the spatial topological feature set; calculate the region influence using the spatial centrality analysis method; calculate the usage intensity according to the region functional density distribution; and generate an importance value by weighting and combining the basic importance, accessibility index, region influence, and usage intensity. Step S134: Organize the component attribute set into a structured manner; map the function type labels according to the region code; The importance values ​​were normalized. The above data are integrated using a feature fusion algorithm; the consistency of the integrated data is checked; and a semantic feature set is generated.

10. The method for generating intelligent UAV inspection paths for building interiors based on BIM model feature extraction according to claim 3, characterized in that, Step S22 is as follows: Step S221: Extract functional tag information from the semantic feature set; perform semantic layering processing on the functional tag information; perform functional normalization processing based on the semantic ontology library; organize the normalized functional nodes hierarchically; perform correlation analysis on the organized nodes; and organize the processed nodes to generate a semantic node set. Step S222: Read node pair information from the semantic node set; calculate the semantic similarity between nodes based on the pre-stored functional association rule base; Calculate the strength of indirect associations using a semantic transitivity probability model; Calculate the overall correlation degree by combining direct and indirect associations; filter valid connections based on the correlation degree threshold; generate a semantic edge set from the filtered node connection relationships; Step S223: Extract region importance information based on semantic feature set; Calculate the basic weights by combining the hierarchical levels of semantic nodes; Calculate structural weights based on node connectivity. Calculate position weights using spatial relationships; A multi-criteria weighting method is used to integrate the weights. Generate the weight values ​​of semantic nodes; Step S224: Encode and identify the semantic node set; Represent the semantic edge set using an adjacency list; Map semantic weight values ​​to corresponding nodes; A graph structure organization method is used to integrate node, edge, and weight information; consistency verification is performed on the integrated graph structure. Generate a semantically enhanced graph.

Images