A Dynamic Monitoring Method and System Based on BIM Spatial Topology and Construction Sound Matching

Abstract

This invention provides a dynamic monitoring method and system based on BIM spatial topology and construction sound matching, relating to the field of dynamic monitoring of building construction. The invention constructs a BIM spatial topology model and an initial construction acoustic association library, parses BIM model data and establishes spatial topological relationships; acquires construction acoustic events and locates sound sources, mapping the sound source locations to the BIM coordinate system; filters existing BIM candidate components based on the BIM spatial topology; calculates the matching results of candidate objects based on the uncertainty state of events and dynamic evidence fusion, dynamically adjusting the reliability of the evidence; generates temporary acoustic objects, virtual components, temporary work units, or sound source probability volume nodes for events with insufficient explanation, outputs the matching results, and updates the dynamic construction acoustic association library based on feedback. This application can improve the accuracy of object attribution judgment between construction acoustic events and BIM components, enhance the interpretability of results, improve project adaptability, express temporary construction objects, and improve matching stability.

Description

Technical Field

[0001] This invention relates to the field of building construction technology, and more specifically, to a dynamic monitoring method and system based on BIM spatial topology and construction sound matching. Background Technology

[0002] With the upgrading of digital and intelligent management at construction sites, construction behavior recognition technology based on audio perception has become an important research direction in the field of intelligent construction. Various operations at construction sites, such as drilling, cutting, grinding, and demolition, generate characteristic acoustic signals. By collecting, preprocessing, extracting features, classifying and identifying, and locating sound sources from the site's audio, acoustic event information such as the type, location, duration, and intensity of construction sounds can be effectively obtained, providing fundamental data support for construction behavior recognition, construction status assessment, and abnormal operation investigation.

[0003] Existing construction audio recognition methods typically focus on sound category determination, that is, determining which construction sound category the current sound belongs to based on audio characteristics or acoustic models. When microphone arrays or multiple acoustic acquisition nodes are deployed on-site, time difference of arrival, sound intensity difference, or sound source localization algorithms can also be used to estimate the approximate location of the sound. Through these methods, acoustic event information such as "what kind of construction sound occurred" and "where the sound approximately came from" can be obtained.

[0004] However, under complex construction site conditions, existing technologies have many core shortcomings, making it difficult to meet the needs of precise construction supervision:

[0005] First, the semantic level of sound recognition is superficial. It can only distinguish the categories of basic construction sounds, but cannot determine the specific construction operation object corresponding to the acoustic signal. The same type of sound can correspond to multiple construction scenarios and components, which easily leads to problems such as ambiguity in construction semantics and unclear attribution of operation objects.

[0006] Secondly, the sound source matching logic is too simplistic, relying excessively on the geometric distance between the sound source point and the building information model (BIM) components to complete the matching. It does not consider the positioning errors caused by sound reflection, obstruction, and attenuation at the construction site. In densely populated areas, it is easy to mismatch acoustic events to components that are close in distance but have unreasonable construction semantics, resulting in insufficient matching accuracy and reliability.

[0007] Third, the BIM model has low utilization. It only calls basic geometric information such as component coordinates and numbers, and does not fully explore the core spatial topological relationships such as component ownership, adjacency, connectivity, and structural attachment. It cannot use spatial constraints to correct matching errors in scenarios such as cross-space sound transmission and component occlusion, which can easily lead to incorrect object ownership.

[0008] Fourth, the scene adaptability is limited. Existing technology assumes that all construction acoustic events can be matched with BIM entity components, which cannot be adapted to a large number of unmodeled temporary work objects such as temporary processing tables, mobile devices, and temporary work areas on the construction site. Forcing such acoustic signals to match existing BIM components will result in matching results that are seriously inconsistent with the actual working conditions on site, greatly reducing the recognition accuracy.

[0009] Fifth, the acoustic association mechanism is rigid, relying on fixed rules and human experience to establish the correspondence between sound, components and construction behavior. The association weight cannot be dynamically updated, making it difficult to adapt to the differentiated scenarios of different engineering projects, construction stages and construction methods, resulting in poor project universality and adaptability.

[0010] Sixth, there is a lack of a comprehensive uncertainty processing mechanism. For complex and uncertain scenarios such as confusion in audio recognition categories, low accuracy of sound source localization, and competition among multiple candidate components, there is no unified comprehensive evaluation system. It is impossible to integrate multi-dimensional information such as recognition confidence, topological relationship, construction stage, and historical matching data, resulting in weak stability and anti-interference of matching results.

[0011] In view of the above, this application is hereby submitted. Summary of the Invention

[0012] This invention aims to provide a dynamic monitoring method and system based on BIM spatial topology and construction sound matching, in order to solve the problems of low accuracy in matching construction sound source objects, poor scene adaptability, and insufficient ability to handle uncertainties in complex working conditions that are common in existing technologies.

[0013] Specifically, the existing technology has the following technical problems: First, the results of construction audio recognition mainly remain at the level of sound category, making it difficult to directly determine the specific construction object corresponding to the sound. The sound of an impact drill may correspond to various construction behaviors such as drilling holes in walls, opening holes in floor slabs, fixing brackets, or installing equipment foundations. Relying solely on sound category to determine construction behavior can easily lead to unclear construction semantics. Secondly, when combining the sound source localization results with the BIM model, matching is usually performed according to the spatial distance between the sound source point and the component, and the component closest to the sound source point is taken as the target object. However, the components are densely distributed on the construction site, and there may be multiple BIM components such as walls, beams, columns, floor slabs, pipes, cable trays, doors and windows near the sound source. The sound will also be affected by reflection, obstruction, attenuation and localization error during the propagation process, resulting in a deviation between the estimated location of the sound source and the actual work object. Matching based solely on geometric distance is prone to classifying the sound into components that are close in distance but have unreasonable construction semantics. Third, existing technologies mainly utilize BIM models by focusing on the geometric location or component numbering of components, lacking comprehensive utilization of BIM spatial topology relationships. They fail to use information such as the ownership relationship between components and rooms, the adjacency relationship between components, the connectivity relationship between spaces, and the attachment relationship between pipelines or cable trays and the main structure for spatial constraints. Fourth, existing technologies typically assume that construction sounds can be matched to existing physical components in the BIM model. However, actual construction sites contain a large number of temporary processing tables, material storage areas, mobile equipment, temporary supports, unfinished walls, temporary installation areas, and other objects that are not explicitly modeled in the BIM model. If such sounds are forcibly matched to existing BIM components, it will lead to inconsistencies between construction acoustic events and actual on-site objects. Fifth, the correspondence between construction sounds and components largely relies on fixed rules or human experience. Under different engineering projects, different construction stages, and different construction methods, the component objects corresponding to the same sound category may be different. The association cannot be updated based on the on-site matching results and the results of manual verification, making it difficult for the system to adapt to the actual construction characteristics of different projects. Sixth, existing technologies lack a unified comprehensive evaluation mechanism for low-confidence sound recognition results, low-precision sound source localization results, and multi-candidate component scenarios. They cannot effectively integrate information such as sound confidence, location confidence, component attributes, construction stage, spatial topology, and historical feedback, resulting in insufficient stability of matching results.

[0014] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: A dynamic monitoring method based on BIM spatial topology and construction sound matching includes: S1. Construct a BIM spatial topology model and an initial construction acoustic association library. Analyze and standardize the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map. S2, collect the original audio signals from the construction site, perform preprocessing, segmentation and construction sound recognition, and then merge them to form construction acoustic events. Then, perform sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a set of construction acoustic events. S3. Based on the set of construction acoustic events and the BIM spatial topology map, determine the spatial retrieval range. At the same time, combine the initial construction acoustic association library to perform preliminary screening of candidate component types to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, calculate the minimum distance from the sound source point of the existing BIM entity component to the component surface, generate event-related topology tags, and obtain the existing BIM candidate component set. S4. Based on the set of construction acoustic events and the existing set of BIM candidate components, the uncertainty state of the event is constructed from sound recognition, sound source localization, and candidate component competition. The reliability of various types of evidence is calculated to obtain dynamic evidence weights. Then, the dynamic evidence is weighted and fused to calculate the dynamic explanatory ability of candidate components for construction acoustic events. At the same time, the candidate competitive advantage adaptive matching threshold is combined to make a comprehensive judgment and output reliable matching results, candidate non-unique results, or events with insufficient global explanation. S5. Based on the global under-explained events, calculate the joint probability contribution of the events to any point in space to delineate candidate dynamic acoustic regions, calculate the dynamic correlation between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence, generate dynamic acoustic topology nodes under the condition that the node confidence meets a preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multiple topology associations between topology nodes and original nodes, and update to obtain the extended BIM space topology map; S6. Based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, establish a matching record between construction acoustic events and dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal levels of construction behavior, and dynamically update the initial construction acoustic association library to achieve dynamic intelligent supervision of the construction site.

[0015] This invention also provides a dynamic monitoring system based on BIM spatial topology and construction sound matching, comprising: The BIM spatial topology generation unit is used to construct a BIM spatial topology model and an initial construction acoustic association library. It performs analysis and standardization processing on the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map. The construction acoustic event set generation unit is used to collect the original audio signals at the construction site, perform preprocessing, segmentation and construction sound recognition, and then merge them to form construction acoustic events. The unit also performs sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a construction acoustic event set. The BIM candidate component screening unit is used to determine the spatial retrieval range based on the construction acoustic event set and the BIM spatial topology map, and at the same time, to perform preliminary screening of candidate component types in conjunction with the initial construction acoustic association library to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, it calculates the minimum distance from the sound source point of the existing BIM entity component to the component surface, generates event-related topology tags, and obtains the existing BIM candidate component set. The component acoustic event matching unit is used to calculate the reliability of various types of evidence to obtain dynamic evidence weights based on the construction acoustic event set and the existing BIM candidate component set, from sound recognition, sound source localization, and candidate component competition to construct the event uncertainty state. Then, it calculates the dynamic explanatory ability of candidate components for construction acoustic events by weighted fusion of dynamic evidence, and combines the candidate competitive advantage adaptive matching threshold for comprehensive judgment, and outputs reliable matching results, candidate non-unique results, or events with insufficient global explanation. The BIM spatial topology map extension unit is used to calculate the joint probability contribution of events to any point in space based on the global under-explained events, so as to delineate candidate dynamic acoustic regions, calculate the dynamic correlation degree between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence degree, generate dynamic acoustic topology nodes under the condition that the node confidence degree meets the preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multiple types of topology associations between topology nodes and original nodes, and update to obtain the extended BIM spatial topology map; The dynamic update unit is used to establish a matching record between construction acoustic events and dynamic acoustic topology nodes based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal level of construction behavior, and dynamically update the initial construction acoustic association library to realize dynamic intelligent supervision of the construction site.

[0016] In summary, compared with the prior art, the present invention has the following beneficial effects: First, improve the accuracy of object matching: This invention avoids incorrect attribution caused by relying solely on geometric distance for matching by comprehensively judging BIM spatial topology constraints and dynamic construction acoustic association library, thereby improving the accuracy of object attribution judgment between construction acoustic events and BIM components.

[0017] Second, it enhances the interpretability of the results: By constructing an event uncertainty state and dynamic evidence weights, this invention dynamically adjusts the reliability of each piece of evidence based on factors such as sound recognition uncertainty, sound source localization uncertainty, and the degree of competition among candidate components, thereby making the matching results more interpretable.

[0018] Third, improve project adaptability: This invention continuously updates the dynamic construction acoustic association library and adjusts the association weights of the construction acoustic association library according to the specific matching results and feedback information of the project, so that the system can gradually adapt to the actual construction characteristics of different projects.

[0019] Fourth, representing temporary construction objects: This invention generates dynamic acoustic topology nodes to incorporate temporary processing tables, material storage areas, mobile equipment, temporary supports, unfinished components, or temporary work areas that are not explicitly represented in the BIM model into the BIM spatial topology relationship, thus solving the problem that the BIM model only contains formal design components and is difficult to represent temporary construction objects on site.

[0020] Fifth, improve matching stability: By integrating multiple pieces of evidence and making dynamic weight adjustments, this invention can still obtain stable matching results in scenarios with low confidence in sound recognition, low precision in sound source localization, and multiple candidate components. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of a dynamic monitoring method based on BIM spatial topology and construction sound matching provided in Example 1.

[0023] Figure 2 This is a flowchart illustrating a dynamic monitoring method based on BIM spatial topology and construction sound matching, as provided in Example 1.

[0024] Figure 3 This is a schematic diagram of spatial topology matching between construction acoustic events and BIM candidate objects provided in Example 1.

[0025] Figure 4 This is a schematic diagram of a dynamic monitoring system based on BIM spatial topology and construction sound matching provided in Example 2.

[0026] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0028] Example 1 Embodiment 1 of the present invention provides a dynamic monitoring method based on BIM spatial topology and construction sound matching, which can be implemented by a dynamic monitoring device based on BIM spatial topology and construction sound matching (hereinafter referred to as dynamic monitoring device), specifically, executed by one or more processors within the dynamic monitoring device.

[0029] In this embodiment, the dynamic monitoring device may be an electronic device equipped with a processor. The processor carries a computer program for the dynamic monitoring method based on BIM spatial topology and construction sound matching, and the computer program can be executed. Such devices include computers, smartphones, smart tablets, workstations, etc., which are not limited here.

[0030] like Figure 1 , Figure 2 As shown, a dynamic monitoring method based on BIM spatial topology and construction sound matching includes steps S1 to S6.

[0031] S1. Construct a BIM spatial topology model and an initial construction acoustic association library. Analyze and standardize the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map.

[0032] Step S1.1: Obtain the BIM spatial topology model and perform data standardization. Use Autodesk Revit to create or receive a BIM spatial topology model, and export the BIM spatial topology model as an IFC format file. Use the IFC parsing module to read data such as building component entities, spatial entities, floor entities, spatial structural relationships, material relationships, attribute sets, and geometric representation information from the IFC format file to obtain BIM data.

[0033] Specifically, BIM data includes at least: ① building component entities; ② spatial unit entities; ③ floor entities; ④ component material information; ⑤ component attribute set information; ⑥ component local coordinate system information; ⑦ component geometric expression information; ⑧ spatial inclusion relationships; ⑨ spatial boundary relationships; and ⑩ connection or attachment relationships between components.

[0034] Among them, building component entities include walls, beams, columns, slabs, doors and windows, pipes, cable trays, equipment foundations, supports, or other construction objects; spatial unit entities include rooms, corridors, floor zones, construction areas, or other spatial ranges used to determine the attribution of construction activities.

[0035] If the IFC format file does not display information on the construction area or construction stage, then the construction area and construction stage information will be supplemented based on the component's custom attributes, external construction progress data, floor range, room boundaries, or preset construction zoning rules.

[0036] Step S1.2: Standardize BIM data. To avoid data inconsistencies caused by different BIM modeling software or different IFC export methods, the parsed data is standardized. This standardization process includes at least: ① unifying length units to meters; ② unifying angle units to radians or degrees; ③ converting component local coordinates to BIM global coordinates; ④ converting component category names to unified category codes; ⑤ converting material names to unified material codes; ⑥ establishing a mapping relationship between component numbers, IFC global identifiers, and internal calculation numbers; and ⑦ completing or normalizing missing or non-standard component attributes.

[0037] For the component's local coordinates in the IFC file, the coordinate transformation relationship of the parent spatial structure where the component is located is read level by level to convert the component's local coordinates into BIM global coordinates. Let the component... The local coordinates of the point are The transformation matrix from the component's local coordinate system to the BIM global coordinate system is: Then the global expression for the coordinates of the component points is: ;in, , This represents the global coordinate transformation matrix of the project. Represents the floor coordinate transformation matrix. This represents the transformation matrix for the local placement of a component.

[0038] After standardization, the standardized basic attribute records of the components are obtained, then the i-th component... The basic attribute record is represented as follows: ; in, Indicates the internal calculation number. This represents the IFC global identifier. Indicates the component type. Indicates the material of the component. Indicates the floor number. Indicates the spatial unit to which it belongs. Indicates the construction area. Indicates the construction phase. It represents the geometric representation information of the component.

[0039] Through the above processing, the spatial position and geometric expression of each component in the unified BIM coordinate system are obtained, which provides a basis for subsequent calculation of the distance between the sound source point and the component, the occlusion relationship and the spatial belonging relationship.

[0040] S1.3: Component geometry is converted into surface patch models. For each BIM component, its geometric representation information is parsed and converted into a computable surface mesh model. For extruded bodies, Boolean bodies, boundary representations, or mapped geometries in the IFC file, their geometric contours, extrusion directions, spatial positioning, and Boolean operation relationships are first parsed. For components with door / window openings or reserved holes, the component geometry is processed by hole deduction based on the subtraction relationship between the component and the opening object, resulting in a component geometric shape consistent with the actual BIM model.

[0041] Subsequently, the surface of the component is discretized into a set of triangular facets. For the i-th component... Its surface mesh model is represented as: ; in, For the set of vertices on the surface of the component, It is a collection of triangular facets on the surface of the component.

[0042] The set of triangular facets can be represented as: ; in, Indicates the first The first component A triangular facet.

[0043] Each triangular facet It should include at least: ① coordinates of the three vertices; ② normal vector of the face; ③ area of ​​the face; ④ component number; ⑤ orientation of the component surface where the face is located; ⑥ a mark indicating whether it belongs to a candidate workable surface.

[0044] By processing the surface triangular facets, we can avoid representing the entire component solely by the component's center point, the center point of the bounding box, or the component's axis, thereby improving the accuracy of subsequent calculations of the distance from the sound source point to the component's surface.

[0045] After surface patching, the component's geometric data set is represented as follows: ,in, This represents the set of surface patch models for all components. This represents the total number of surface patch models.

[0046] S1.4: Establish spatial ownership relationships and BIM spatial topology relationships. Based on the spatial structural relationships, spatial boundary relationships, and component geometric data obtained from IFC analysis, establish spatial topology relationships between components, spatial units, floors, and construction areas.

[0047] First, determine the floor, spatial unit, and construction area to which the component belongs based on the spatial inclusion relationships in the IFC file. If the component Belongs to spatial unit Then establish: .

[0048] If the IFC file lacks the containment relationship between components and spatial elements, it is supplemented by determining the relationship based on the geometric position of the components and the boundaries of the spatial elements. Let the component... geometric center point for: ;in, For the set of vertices on the surface of the component, Indicates the number of vertices. Represents the three-dimensional coordinates of a vertex.

[0049] The spatial attribution relationship is determined by judging the geometric center point of the component. Does it fall within a spatial unit? Within the boundary, if Located in space unit Within the spatial boundary, the components will Belonging to spatial unit .

[0050] For cross-space components, calculate the component With spatial units The overlap ratio of the components The formula for calculating the overlap ratio of the primary home space is: ; in, Representing components With spatial units The overlap ratio; Representing components Located in space unit Internal volume; Representing components The total volume.

[0051] The spatial unit with the largest overlap ratio is taken as the main space to which the component belongs, and the remaining overlapping spaces are taken as the associated spaces of the component.

[0052] Then, establish the adjacency relationships between components (inter-component adjacency relationships). For two components... and The minimum distance between component surfaces is calculated based on its surface patch model, and the expression is: ; in, This represents the shortest distance between two triangular faces. If the following conditions are met:

[0053] Then create components With components The adjacency relationship between them, where, This is a preset threshold for the distance between adjacent components.

[0054] Subsequently, based on spatial boundaries, doorways, corridors, open boundaries, or spatial connections, the connectivity relationships between spatial units are established. If spatial units... With spatial units If a connected structure exists, then establish: .

[0055] For components such as pipes, cable trays, supports, and equipment foundations, if there is an IFC connection, support relationship, or attachment relationship between them and the main components such as walls, beams, slabs, and columns, or if the surface distance between them is less than a preset attachment threshold, then an attachment relationship is established: ,but Indicates the attachment component. Indicates the main component.

[0056] The final BIM spatial topology map is generated: ,in, It is a set of topological nodes, including component nodes, spatial unit nodes, floor nodes, and construction area nodes; It is a set of spatial topological edges, which include inclusion, belonging, adjacency, connectivity, boundary, and attachment relationships.

[0057] S1.5: Establish an initial construction acoustic association library. Based on construction method knowledge, historical construction audio samples, manual annotation results, and construction task rules, establish an initial construction acoustic association library.

[0058] The initial construction acoustic association database is used to record the correspondence between component attributes, construction sound categories, and construction behaviors. Each association record includes at least: ① component type; ② component material; ③ component specialty; ④ component construction stage; ⑤ sound category; ⑥ possible construction behaviors; ⑦ initial association weight; ⑧ abnormal or prohibited conditions.

[0059] Represented as: ; in, Indicates the first The component type of the associated record, Indicates the first The component materials of the associated records, Indicates the first The associated record indicates the component's construction stage or the specialty it belongs to. Indicates the sound category, This indicates potential construction activity. Indicates the initial association weight. Indicates an abnormal or prohibited condition.

[0060] The initial association weight Determined based on rule weights, sample statistical weights, and manual annotation weights: ; in, This indicates the assignment of values ​​to the construction method rules. This represents the statistical results of historical construction audio samples. This indicates the results of manual annotation. , , Let be the weighting coefficient, and satisfy: .

[0061] For example, the initial construction acoustic association library includes: (Concrete wall, concrete, electromechanical installation stage, impact drill sound, drilling or bracket fixing, 0.90); (Pipeline, metal, electromechanical installation stage, cutting sound, pipe processing or installation adjustment, 0.86); (Steel beam, steel material, steel structure construction stage, grinding sound, pre-welding treatment or surface finishing, 0.88).

[0062] The initial construction acoustic association library is represented as follows: ; This indicates the number of associated records in the initial construction acoustic association library.

[0063] S1.6: Form a spatial-acoustic foundation dataset. Unify the standardized component data, spatial unit data, component surface patch models, BIM spatial topology diagrams, and initial construction acoustic association library to form a spatial-acoustic foundation dataset.

[0064] Spatial-Acoustic Fundamentals Dataset Represented as: ; in, Represents a set of components. Represents a set of spatial units. Represents a set of floors. Indicates a collection of construction areas. This represents a set of surface patch models of a component. Represents the BIM spatial topology map. This represents the initial construction acoustic association library.

[0065] S2: Collect the original audio signals from the construction site, perform preprocessing, segmentation, and construction sound recognition, and then merge them to form construction acoustic events. Perform sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a set of construction acoustic events.

[0066] S2.1: Deploy acoustic acquisition units and establish sensor coordinate information. Acoustic acquisition units, including microphone arrays or multiple distributed microphone nodes, are deployed at the construction site.

[0067] For each acoustic acquisition unit, a sensor attribute record is established. The sensor attribute record includes at least: ① sensor number; ② sensor type; ③ sampling frequency; ④ installation location; ⑤ installation height; ⑥ floor; ⑦ spatial unit; ⑧ sensor local coordinates; ⑨ sensor time synchronization information.

[0068] For the The locations of the acoustic acquisition units are indicated as follows: ; in, Indicates the first In the array, the first The location coordinates of each microphone.

[0069] Time synchronization processing is performed on multiple acoustic acquisition units to ensure that audio signals acquired from different acquisition channels have a unified timestamp. Time synchronization methods include network time synchronization, hardware-triggered synchronization, or time correction based on synchronization pulse signals.

[0070] S2.2: Acquire audio from the construction site and segment the audio data. Audio from the construction site is continuously acquired using an acoustic acquisition unit to obtain the raw audio signal.

[0071] For the The original audio signal acquired by the acoustic acquisition channels is represented as follows: ,in, Indicates the sampling time.

[0072] The original audio signal is subjected to uniform sampling rate conversion, amplitude normalization, and noise suppression to obtain a cleaned audio signal:

[0073] The purified audio signal is divided into sliding segments according to a preset time window length and window step size. Let the window length be... Window step size is Then the first Each audio segment is represented as: ; in, Indicates the first The first acquisition channel An audio clip, This indicates the start time of the audio segment.

[0074] For each audio segment, record its start time, end time, acquisition channel, and acoustic acquisition unit number.

[0075] S2.3: Perform construction sound category identification on audio segments. Extract acoustic feature vectors from the segmented audio segments and input them into the construction sound recognition model to obtain the construction sound category identification results.

[0076] The construction sound recognition model can be a convolutional neural network, support vector machine, or other supervised classification model. The model outputs a sound category probability vector, representing the probability that the audio segment belongs to drilling, cutting, grinding, knocking, breaking, moving, or non-construction sound. The system determines the sound category and recognition confidence level based on the probability vector. If the recognition confidence level is lower than a preset threshold, the audio segment is marked as a low-confidence acoustic segment. If the sound category is non-construction sound, it does not proceed to subsequent matching steps.

[0077] The acoustic feature vectors include short-time energy, zero-crossing rate, spectral centroid, spectral bandwidth, Mel frequency cepstral coefficients, Mel spectrogram or time-spectral graph features, etc.

[0078] The acoustic feature vector is input into the construction sound recognition model, and the output sound category probability vector is represented as follows: ; in, Indicates the first A probability vector of the sound category of each audio segment; Indicates the first The audio segment belongs to the first The probability of construction-like noises.

[0079] The construction sound categories include at least drilling sounds, cutting sounds, grinding sounds, knocking sounds, breaking sounds, handling sounds, and non-construction sounds.

[0080] Determine the first The audio segments are categorized as follows: ; Corresponding recognition confidence level for: ; This indicates taking the maximum value.

[0081] If the confidence level is identified If the confidence level is less than the preset recognition threshold, the audio segment is marked as a low-confidence acoustic segment; if... If the sound is not from construction, the subsequent sound source object matching step will not be performed.

[0082] S2.4: Forming construction acoustic events. Audio segments with the same sound category, continuous time, and meeting the confidence requirements within a continuous time window are merged to form construction acoustic events.

[0083] Let adjacent audio segments and satisfy: And the time interval between the two is less than the preset event merging threshold. If so, they are merged into the same construction acoustic event.

[0084] After merging, the construction acoustic events are obtained, then the first... Construction acoustics incident Represented as: ; in, Indicates an acoustic event The acoustic event number, Indicates an acoustic event sound categories, Indicates an acoustic event Event recognition confidence, Indicates an acoustic event The start time of the event, Indicates an acoustic event The duration of the event, Indicates an acoustic event The intensity of the event sound, Indicates an acoustic event The corresponding set of audio segments.

[0085] Among them, event identification confidence It can be obtained from the mean or weighted mean of the confidence scores of the merged segments: ; in, Indicates an acoustic event The number of audio segments included.

[0086] Event sound intensity It can be represented by the average sound energy over the duration of the event: ; in, Indicates the first The sound energy of an audio clip.

[0087] S2.5: Localize the sound sources of construction acoustic events. Localize the sound sources of existing construction acoustic events to obtain their positions in the site coordinate system.

[0088] When using a microphone array or multiple distributed microphone nodes, the sound source location is calculated based on the arrival time difference between different acquisition channels of the microphone array, and the sound source coordinates are solved using global least squares.

[0089] Let the first The microphone and the first The time difference between the arrival times of the same acoustic event received by the microphones is The speed of sound in air is The location of the sound source in the on-site coordinate system is Then we have: ; For a combination of multiple microphones, the sound source location can be determined by minimizing the time difference of arrival error, as expressed by: ; in, For acoustic events The corresponding three-dimensional coordinates of the sound source; For the microphone pairs involved in positioning; , The first , The location coordinates of each microphone; The speed of sound in air; For the first , The time difference of arrival of each microphone; Let be the three-dimensional coordinates of the sound source to be solved; Represents Euclidean spatial distance; It is a least squares operator.

[0090] If the sound source localization conditions are insufficient, the spatial range of the acoustic event is determined based on the spatial unit where the acoustic acquisition unit is located, the sound intensity attenuation, the location of the event within multiple time windows, and the sensor coverage area. The center point of the area or the probability center point is used as the estimated value of the sound source location.

[0091] A location reliability record is established for the positioning results. The location reliability is related to the number of sensors, signal-to-noise ratio, time difference estimation stability, and positioning residual.

[0092] S2.6: Map the sound source location to the BIM coordinate system.

[0093] The location of the construction acoustic event in the on-site coordinate system is transformed to the BIM coordinate system to obtain the location of the sound source in the BIM coordinate system.

[0094] Construction acoustic incident The location in the on-site coordinate system is The corresponding BIM coordinates are The rigid transformation relationship from the site coordinate system to the BIM coordinate system is expressed as follows: ; in, Represents the rotation matrix. This represents the translation vector.

[0095] The rotation matrix and translation vector can be determined through the correspondence between field control points, BIM reference points, measurement points, or sensor installation points.

[0096] If there is a scale difference between the site coordinate system and the BIM coordinate system, a similarity transformation is used to represent it as follows: ; in, Represents the scale factor.

[0097] After completing the coordinate transformation, the construction acoustic events will be... The location of the sound source is spatially matched with the BIM spatial topology map formed in S1 to determine the floor, spatial unit, and construction area to which the acoustic event belongs.

[0098] If the BIM coordinates of the sound source point Located in space unit Within the spatial boundary, then: .

[0099] If the sound source is located near multiple spatial boundaries, the candidate spatial unit set is determined based on the sound source location reliability, the minimum distance from the boundary, and the spatial connectivity.

[0100] S2.7: Output a set of construction acoustic events. After construction sound recognition, event merging, sound source localization, and BIM spatial mapping, a set of construction acoustic events is formed, with each construction acoustic event... The records include: ; in, Indicates the acoustic event number. Indicates the sound category, Represents the probability vector of sound categories. Indicates the confidence level of identification. Indicates the start time of the event. Indicates the duration of the event. Indicates the intensity of the sound of the event. This indicates the location of the sound source in the BIM coordinate system. Indicates the spatial unit to which it belongs. Indicates the construction area. This indicates the reliability of sound source localization.

[0101] Sound source localization reliability The formula for determining the accuracy of microphone array positioning is based on a comprehensive analysis of relevant peak sharpness, latency residual consistency, and continuous frame positioning stability during the positioning process. ; in, Indicates an acoustic event The corresponding sound source localization reliability; Indicates the peak sharpness related to GCC-PHAT. This indicates the consistency between the actual delay and the theoretical delay calculated from the location point. This indicates the spatial stability of consecutive short-time localization results within the same acoustic event. , , They are respectively , , The weighting coefficients, and .

[0102] The value of is generally in the range of: , The closer the value is to 1, the more reliable the positioning. The closer it is to 0, the less reliable the positioning is.

[0103] It should be noted that the relevant peak resolution It involves calculating the cross-correlation function between each pair of channels by using the GCC-PHAT generalized cross-correlation algorithm on multiple audio signals acquired by a microphone array, extracting the ratio of the amplitude of the main peak to the amplitude of the secondary peak of the cross-correlation curve, and then normalizing the result. The higher the main peak is relative to the secondary peak, the sharper the waveform. The closer the value is to 1, the higher the clarity of the source-related peak.

[0104] Delay Consistency First, the measured signal delay of each microphone pair is obtained through the GCC-PHAT peak position. Then, based on the current three-dimensional positioning coordinates of the sound source, the theoretical geometric delay corresponding to each microphone pair in the array is calculated in reverse. The mean square error between the measured delay and the theoretical delay is obtained, and then mapped to the [0,1] interval through inverse normalization. The smaller the deviation between the measured delay and the theoretical delay, the better. The closer a value is to 1, the better the consistency of latency matching.

[0105] Short-time positioning spatial stability By constituting the same acoustic event Each continuous short-time audio segment was independently used to locate the sound source, resulting in a set of continuous short-time three-dimensional coordinate sequences. The variance of the spatial Euclidean distance between all localization points within this coordinate sequence was calculated, and the variance was inversely normalized to obtain... The more concentrated the distribution of continuous short-term positioning points, the smaller the spatial fluctuation. The closer it is to 1, the stronger the spatial stability of the short-term positioning result.

[0106] This results in a set of construction acoustic events that includes the construction acoustic events, the location of the sound source in the BIM coordinate system, the spatial unit to which it belongs, the construction area to which it belongs, and the location confidence of the sound source.

[0107] S3. Based on the set of construction acoustic events and the BIM spatial topology map, determine the spatial retrieval range. At the same time, combine the initial construction acoustic association library to perform preliminary screening of candidate component types to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, calculate the minimum distance from the sound source point of the existing BIM entity component to the component surface, generate event-related topology tags, and obtain the existing BIM candidate component set.

[0108] S3.1: Read construction acoustic events and determine the spatial retrieval range. Read the set of construction acoustic events output in step S2, for the... A construction acoustic event was identified, and its event log was retrieved. The spatial retrieval radius is determined based on the sound source localization reliability and sound category. The calculation formula is: ; in, Indicates construction acoustic events Spatial retrieval radius; Show sound category The corresponding basic search radius; This represents the positioning uncertainty correction coefficient; Indicates an acoustic event The corresponding sound source localization reliability.

[0109] Based on the spatial unit to which the acoustic event belongs Construction area Spatial search radius The BIM spatial topology map established in step S1 Determine the spatial search range: ; in, For acoustic events The corresponding spatial retrieval range; For acoustic events Its own spatial unit; To and The set of all adjacent spatial units; To and Connected and falling within the spatial retrieval radius A collection of spatial units within.

[0110] If the construction acoustic event is located near a spatial boundary, the spatial units adjacent to that boundary will be added to the spatial search range. If the sound source localization reliability is below a preset threshold, the spatial search range will be expanded to the same construction area. Adjacent spatial units within.

[0111] S3.2: Initial screening of candidate component types based on the construction acoustic association library. This is done according to the sound category probability vector of construction acoustic events. The initial construction acoustic association library established in step S1 Determine the set of candidate component types related to the current sound event.

[0112] The first in the initial construction acoustic association library Related records For component type Based on the sound category probability of acoustic events and the associated library weights The method for calculating the acoustic prior correlation value for this component type is as follows: ; in, For component type Relative acoustic events Acoustic prior correlation values; For acoustic events Identified as a sound category The probability of; This is the association weight for the initial construction acoustic association library, that is, the initial association weight between the sound category and the component type; The first one represents the initial construction acoustic association library. 1 associated record.

[0113] when , If the preset initial screening threshold for component types is used, then the component types will be... Adding a candidate component type to the set is represented as: ; in, For acoustic events The set of candidate component types.

[0114] For cases where the same sound category corresponds to multiple component types, multiple candidate component types are retained. For events where the sound recognition confidence level is below a preset threshold but still belongs to construction sounds, the sound category probability vector is selected. Several sound categories participate in the selection of candidate component types.

[0115] S3.3: Filter existing BIM entity components based on spatial scope and component type. Within the spatial search scope... Within this scope, the system reads the set of BIM entity components that fall within this range, and filters the component types from the set of components located within the spatial retrieval range to ensure that the component types belong to the candidate component type set. The components are used to obtain the initial set of acoustic semantic components.

[0116] For each component in the initial acoustic semantic screening component set Based on the component surface patch model obtained in step S1, the sound source point is calculated. Minimum distance to the component surface: ; in, For acoustic events Components within the spatial retrieval range The shortest distance between the triangular facets of the surface; This indicates finding the minimum value; For acoustic events The corresponding BIM coordinates of the sound source; For components within the spatial retrieval range A collection of triangular facets on the surface of a component; For components within the spatial retrieval range The first set of triangular facets on the surface of the component A triangular facet; This is the shortest distance in Euclidean geometry.

[0117] If the following conditions are met: Then the components Add to the existing set of BIM entity candidate objects.

[0118] For components whose distance from the sound source is less than a preset nearest neighbor threshold, even if their component type does not belong to the candidate component type set. It can also be added to the candidate set as a geometric nearest neighbor and marked as a low-acoustic prior object.

[0119] S3.4: Supplement candidate objects based on BIM topology relationships. Based on the BIM spatial topology map... The existing BIM entity candidate object set is topologically expanded, that is, topologically related components in the BIM spatial topology map that have component attachment, component adjacency, component connection or component affiliation with candidate components in the existing BIM entity candidate set are included in the candidate set, and the candidate set is updated after the topology expansion.

[0120] If candidate components Other components If there are attachment, adjacency, or connection relationships, then... Add it to the candidate set as a candidate object for topological association.

[0121] For components such as pipes, cable trays, supports, and equipment foundations, if they are attached to main structural components such as walls, beams, slabs, and columns: Then As an associated main component, it is added to the candidate object set.

[0122] For main structural components such as walls, floors, beams, and columns, if they are attached to supports, cable trays, pipelines, or equipment foundations, the corresponding attached components will be added to the candidate object set as related construction objects.

[0123] After topology expansion, an updated set of existing BIM entity candidate objects is obtained, namely the updated candidate set after topology expansion, which includes geometrically adjacent components, acoustically semantically related components, and topologically associated components.

[0124] S3.5: Calculate the event-related topology tags for candidate objects. For each candidate object in the updated set of existing BIM entity candidate objects, calculate its relationship with construction acoustic events. Event-related topology tags between them.

[0125] The event-related topology tags include: ① Whether it is located in the same spatial unit as the acoustic event; ② Whether it is located in the same construction area as the acoustic event; ③ Whether it is located in the connected space of the spatial unit to which the acoustic event belongs; ④ Whether it has an attachment relationship with other candidate objects; ⑤ Whether there is a spatial boundary barrier; ⑥ Whether it belongs to the initial acoustic semantic screening component; ⑦ Whether it belongs to the geometric nearest neighbor candidate object.

[0126] For candidate components Set its event-related topology tags, including tags for same spatial units, same construction areas, connected spaces, attachment relationships, spatial barriers, acoustic semantic screening tags, and geometric nearest neighbor tags.

[0127] Spatial barrier markers are used to characterize the BIM coordinates of sound sources. To candidate BIM components Are there any sound-insulating or blocking components along the straight path from the nearest point on the surface? Specifically, first determine the sound source point. To candidate components The closest point on the surface : ; in, Indicates candidate components In the BIM global coordinate system, the surface set can be composed of triangular facets on the component surface; express Any surface point on it; Indicates the distance from the sound source. The nearest component surface point.

[0128] Then, the sound source point closest point to the surface of the candidate component The detection line segments between them are represented in parametric form: ; in, For line segment parameters. When hour, , indicating that the starting point of the line segment is the sound source location point; when hour, This indicates that the endpoint of the line segment is the nearest point on the surface of the candidate component. Therefore, This represents the complete spatial detection line segment from the sound source point to the nearest point on the surface of the candidate component.

[0129] Let the set of barrier components be: ,in, This refers to BIM components that provide spatial separation, such as walls, floors, and enclosed partitions. If separating components exist... Furthermore, the barrier component is not a candidate component. It itself makes: , If a spatial barrier exists between the sound source and the candidate component, then the following settings are made: Otherwise set .

[0130] S3.6: Output the complete set of existing BIM candidate components Components that have undergone spatial range filtering, component type filtering, geometric distance filtering, and BIM topology expansion will be considered as construction acoustic events. The complete set of existing BIM candidate components is represented as: ; in, Indicates construction acoustic events The corresponding complete set of existing BIM candidate components, Indicates construction acoustic events The corresponding number Candidate components.

[0131] for Each candidate component (i.e., construction acoustic events) The corresponding number (1 candidate component), establish a candidate component record, including candidate component number, component type, component material, spatial unit to which the component belongs, construction area to which the component belongs, construction stage of the component, minimum distance from acoustic event to candidate component surface, topological label between candidate component and acoustic event, acoustic prior association value between candidate component and current sound category, and candidate component source status.

[0132] The candidate component source states include: ① acoustic semantic preliminary screening components; ② geometric proximity components; ③ topological association components; ④ low acoustic prior candidate components.

[0133] Output the set of existing BIM candidate components corresponding to all construction acoustic events.

[0134] S4. Based on the set of construction acoustic events and the existing set of BIM candidate components, the uncertainty state of the event is constructed from sound recognition, sound source localization, and candidate component competition. The reliability of various types of evidence is calculated to obtain dynamic evidence weights. Then, the dynamic evidence is weighted and fused to calculate the dynamic explanatory ability of candidate components for construction acoustic events. At the same time, the candidate competitive advantage adaptive matching threshold is combined to make a comprehensive judgment and output reliable matching results, candidate non-unique results, or events with insufficient global explanation.

[0135] S4.1: Construct the uncertain state of sound category.

[0136] Let the first The probability vector of sound category for each construction acoustic event is: To reduce the impact of confusion between similar categories in the construction sound recognition model on subsequent object matching, a sound category confusion matrix is ​​established based on project calibration samples or historical samples. , is represented as: ,in, Indicates the category of real sound. When identified as The probability of.

[0137] For the original sound category probability vector After performing obfuscation correction, the corrected sound category probability vector is obtained: The formula is: ; ; in, To obscure the corrected construction acoustics incident The normalized probability vector; For construction acoustic events The original sound category probability vector; For the originally identified construction acoustic events The The probability of construction-like sounds; Indicates vector normalization; This represents a truncation function; it sets negative numbers to 0 and retains positive numbers. For sound category confusion matrix; It is a minimal constant to prevent the matrix from becoming singular and non-invertible; for An identity matrix of order 1, with all diagonal elements being 1 and the rest being 0. Here, the order of the identity matrix is... The same as the total number of all construction sound categories.

[0138] Based on the corrected sound category probability vector Calculate the dispersion of sound category distribution: ; in, Indicates construction acoustic events The dispersion of sound category distribution; This represents the total number of all construction sound categories. Construction acoustic events after obfuscation correction Belongs to the Normalized probability of construction-related sounds; The closer a value is to 1, the more dispersed the probability distribution of sound categories and the more uncertain the sound categories are; the closer a value is to 0, the more concentrated the model's judgment on a certain sound category is.

[0139] Calculate the discriminant between the highest probability class and the second highest probability class: ; in, Indicates construction acoustic events Category discrimination; , These are the construction acoustic events after the confusion correction. The maximum and second largest values ​​in the normalized probability vector.

[0140] To determine whether the event is confused between similar construction sound categories, the category confusion sensitivity is further calculated. : ; in, Indicates construction acoustic events Category confusion sensitivity; For construction acoustic events belong The corrected probability of the sound type; for Sound-like The risk factor for sound confusion. This applies to easily confused sound categories such as drilling sounds and cutting sounds, hammering sounds and breaking sounds, grinding sounds and cutting sounds. Take the larger value; for clearly different sound categories, Take the smaller value.

[0141] Let the set of audio segments contained within the j-th acoustic event be . ,in, express The number of audio segments included.

[0142] Let the first The corrected sound category probability vector for each audio segment is: The fluctuation of the sound category within the event for: ; in, Indicates the first The optimal sound category determined for each audio segment ; This is an indicator function; it takes the value 1 if the condition is true, and 0 otherwise. This represents the total number of audio segments that are most frequently classified as belonging to the same category among all audio segments.

[0143] The larger the value, the more inconsistent the highest sound category is among different short segments within the same event, and the less stable the sound category judgment is.

[0144] The reliability of a sound category is obtained by combining the dispersion of sound category distribution, category discrimination, category confusion sensitivity, and intra-event category volatility. Then the uncertainty of the sound category is obtained. , is represented as: ; in, , , , This is an adjustment coefficient for the uncertainty of the sound category; It is an exponential function.

[0145] when A larger value indicates that there is uncertainty in the sound category of the construction acoustic event, which reduces the role of acoustic semantic evidence in subsequent candidate component matching and enhances the role of BIM topology evidence, construction phase evidence, and historical stability evidence.

[0146] S4.2: Construct the uncertainty state of sound source localization. For the j-th construction acoustic event, read its internal continuous short-time frame localization results, represented as: ; in, Indicates construction acoustic events The Three-dimensional sound source localization coordinates in a short time frame; For construction acoustic events The total number of short-time frames.

[0147] Based on the signal-to-noise ratio, GCC-PHAT main peak sharpness, and time delay residual of each short-time positioning point, a positioning frame weight is set: ; in, for No. The audio signal-to-noise ratio of a short frame; for No. Cross-correlation peak sharpness of a short time frame; for No. The measured delay of a short frame is consistent with the theoretical delay.

[0148] Calculate the event-level location center based on the location frame weights: , For construction acoustic events The weighted sound source localization center.

[0149] Calculate the positioning covariance matrix based on the discreteness of the positioning points: ; in, For construction acoustic events The positioning covariance matrix; It is a minimal constant to prevent the matrix from becoming singular and non-invertible; for An identity matrix of order 1 is used to ensure that the covariance matrix is ​​positive definite.

[0150] Calculate the volume index of the positioning uncertainty ellipsoid based on the positioning covariance matrix: ; in, For construction acoustic events The positional uncertainty of the ellipsoidal volume logarithmic index; This refers to matrix determinant operations. The larger the value, the greater the spatial uncertainty range of the positioning point.

[0151] Let the set of microphone pairs involved in the localization be . For microphone pair (a, b), the observation arrival time difference is: By the positioning center The theoretical arrival time difference obtained by reverse calculation is Then the whitening delay residual is: ; in, For construction acoustic events The mean of whitening delay residuals; A set of paired microphones for participating in positioning; This represents the total number of pairs; For microphone pair Actual measured sound arrival time difference; For the reason Theoretical arrival time difference calculated in reverse; For microphone pair Delay estimation noise standard deviation.

[0152] Calculate the spatial affiliation jump degree based on the changes in the spatial unit to which the location point belongs in consecutive short-time frames: ; in, For construction acoustic events Spatial affiliation jump degree; This indicates that the positioning point falls into the spatial unit. The number of short-time frames; This represents the cell count that contains the most positioning frames among all spatial cells. The larger the value, the more unstable the attribution of the same acoustic event in the BIM space, and the more likely it is to be affected by reverberation, reflection, or interference from multiple sound sources.

[0153] The reliability of sound source localization is obtained by combining the volume of the localization ellipsoid, the time delay residual, and the spatial affiliation jump degree. Thus, the uncertainty in sound source localization is obtained, expressed as: ; in, For construction acoustic events The uncertainty in sound source localization; , , These are the adjustment coefficients for the uncertainty of sound source localization.

[0154] when A larger value indicates that the sound source localization result of the acoustic event is unstable, which reduces the role of geometric distance evidence in subsequent candidate component matching, and increases the role of spatial topological evidence, construction stage evidence, and historical association evidence.

[0155] S4.3: Construct the uncertainty state of candidate component competition. For the set of candidate components corresponding to the j-th construction acoustic event, first calculate the initial competitive potential energy of each candidate component: ; in, For construction acoustic events Candidate Components (referred to as candidate components) The initial competitive potential energy; BIM coordinates of the sound source To candidate components The minimum Euclidean distance between the triangular facets of the surface; This is the distance attenuation parameter; For construction acoustic events candidate components Acoustic prior association value with the current sound category; For initial acoustic semantic screening of components, use binary tags, where 1 represents a construction acoustic event. candidate components It belongs to the acoustic semantic preliminary screening component; For topology expansion candidate components, use binary labels, where 1 represents a construction acoustic event. candidate components ... , , The adjustment coefficient represents the initial competitive potential energy.

[0156] Normalize the initial competitive potential energy of all candidate components to obtain the initial competitive probability of the candidate components: ; in, For construction acoustic events The total number of candidate components; For construction acoustic events Candidate Components Normalized initial competition probability.

[0157] Calculate the competition entropy of candidate components based on the initial competition probability: ; in, For construction acoustic events The entropy of competition among candidate components; The larger the value, the closer the initial competition among multiple candidate components, and the more difficult it is to distinguish between candidate objects.

[0158] Calculate the effective density of candidate components: ;in, For construction acoustic events The effective density of candidate components, The larger the value, the more candidate components with similar explanatory power are competing for the same functionality.

[0159] Thus, the uncertainty of the candidate competition is expressed as: ; in, , This is an adjustment coefficient for competition uncertainty.

[0160] S4.4: Generate event-level uncertainty state vector and dynamic evidence weights. Based on sound category uncertainty, sound source localization uncertainty, and candidate competition uncertainty, combined with topological conflict degree, construction stage conflict degree, and historical sample insufficiency degree, construct the event-level uncertainty state vector: ; in, For construction acoustic events The uncertain state vector; Indicates construction acoustic events Sound category uncertainty; For construction acoustic events The uncertainty in sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events Uncertainty of conflicts during the construction phase; For construction acoustic events The uncertainty stems from insufficient historical samples.

[0161] The topological conflict degree The expression is: ; in, This indicates the number of candidate components that have spatial barriers, cross-space propagation, non-connected spaces, or conflicting attachment relationships; For construction acoustic events The total number of candidate components.

[0162] The degree of conflict during the construction phase The expression is: ; in, For construction acoustic events The number of candidate components that conflict with the construction behavior corresponding to the current sound during the construction phase.

[0163] The historical sample insufficiency The expression is: ; in, For construction acoustic events The number of historical samples that match the current component, sound, and construction stage; This is the historical sample sufficiency adjustment constant.

[0164] Based on the event-level uncertainty state vector, the reliability of different evidence channels in the current acoustic event is generated: ; ; ; ; ; ; in, For construction acoustic events The reliability of geometric distance evidence; For construction acoustic events The reliability of sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events The reliability of acoustic semantic evidence; For construction acoustic events Reliability of sound category recognition; For construction acoustic events Category confusion sensitivity; For construction acoustic events The volatility of the internal category of the event; For construction acoustic events Uncertainty of the location of a fixed sound source; For construction acoustic events The reliability of BIM topology evidence; For construction acoustic events The reliability of evidence during the construction phase; For construction acoustic events Sound category uncertainty; For construction acoustic events Reliability of historical matching association evidence; For construction acoustic events The uncertainty due to insufficient historical samples; For construction acoustic events The reliability of evidence regarding abnormal constraint penalties; For construction acoustic events Uncertainty of conflicts during the construction phase; , , , , , , , These are the attenuation adjustment coefficients for each item.

[0165] Dynamic evidence weights are generated based on the above evidence reliability: ; in, For the first Similar evidence in construction acoustic events The dynamic weights under the condition are all summed to 1; This is the evidence scaling factor; The evidence bias constant; For construction acoustic events The Reliability of such evidence.

[0166] S4.5: Calculate the geometric interpretation evidence for the candidate components. For candidate components... (i.e., construction acoustic events) candidate components ), calculate the weighted sound source localization center To the candidate component surface patch set Minimum Mahalanobis distance: ; in, From the sound source localization center to the candidate component The minimum Mahalanobis distance of the surface; For construction acoustic events The weighted sound source localization center; Candidate components Any three-dimensional vertex on the surface patch; For construction acoustic events The inverse matrix of the positioning covariance matrix; Candidate components The set of all surface vertices; This indicates that the minimum value is being sought.

[0167] If the surface patch of the candidate component With workable surface marking And the normal vector of the patch is The main direction vector of the sound source is The correction term for the workability consistency of the component surface is: ; in, Candidate components Workable surface matching correction items; For dough pieces The binary markers are 1 for workable surfaces and 0 for non-workable surfaces; For dough pieces The normal vector; The vector representing the main propagation direction of the sound source.

[0168] The geometric interpretation evidence is expressed as follows: ; in, The workable surface reinforcement coefficient.

[0169] S4.6: Calculate acoustic semantics, BIM topology, and construction phase interpretation evidence. This is based on the corrected sound category probability vector and the current construction acoustic association library. Calculate candidate components Evidence for acoustic semantic interpretation : ; in, Construction acoustic events after obfuscation correction Belongs to the Normalized probability of construction-related sounds; This represents the total number of all construction sound categories. For the initial construction acoustic association library; Candidate components Component types; Candidate components Component materials; Candidate components The component construction stage; Indicates the candidate component type, material, construction stage, and number of... The correlation strength between construction sounds.

[0170] Based on the BIM spatial topology map, calculate the spatial unit to which the construction acoustic event belongs. To candidate components Minimum topological path cost between : ; in, For construction acoustic events spatial unit To candidate components The minimum total cost of the topological path; This represents all topological connectivity paths from the spatial unit where the sound source is located to the candidate component; It is a single topological edge; Topological edges The cost of passage.

[0171] In this embodiment, topological edges The passage cost can be calculated by comprehensively considering spatial topological factors, geometric distance attenuation weights, and acoustic matching confidence. First, topological attributes such as spatial units, component adjacency, and connectivity dependencies connected by the topological edge are extracted to obtain the topological weight. Then, a distance attenuation function is constructed by combining the geometric distance between the sound source location point and the BIM component to correct the weight. At the same time, the comprehensive confidence of sound source location and acoustic semantic matching score are introduced as correction coefficients. The three types of indicators—topological constraints, spatial distance, and acoustic evidence—are weighted, summed, and normalized to the [0,1] interval, finally yielding the topological edge. The passage cost is calculated as follows: a higher cost value indicates a higher spatial constraint cost and lower topological connectivity reliability for matching the sound source with the BIM component along that edge; conversely, a lower cost value indicates strong spatial connectivity of the topological path and its suitability as a high-reliability matching support path. Alternatively, other existing methods can be used for calculation, which are not limited here.

[0172] Different topological costs are set for spatial inclusion, attachment, adjacency, connectivity, cross-space, and barrier relationships; if the path passes through barrier components such as walls, floors, and enclosed partitions, the barrier propagation cost is increased.

[0173] The BIM topology interpretation evidence for the candidate components is represented as follows: ; in, For sound source and candidate components The spatial obstruction penalty term is generated by the spatial obstruction marker.

[0174] Calculate the current event time based on the construction schedule or component status. With the allowable construction time window for candidate components Time deviation between: ; in, For construction acoustic events Time of the event With candidate components The components allow for construction windows Time window deviation; This is a time-distance function.

[0175] like If it is within the permitted construction time window, then .

[0176] Explanation of evidence during the construction phase Represented as: ; in, This refers to the decay parameter within the time window. Candidate components Component types; Candidate components The construction phase; For construction acoustic events Corrected sound probability vector; This represents the degree to which the candidate component type and construction stage are adapted to the current modified sound probability vector.

[0177] S4.7: Calculate historical correlation evidence and anomaly constraint evidence. Read historical matching records from the dynamic construction acoustic correlation library related to candidate component type, material, construction stage, sound category, spatial unit, and construction area.

[0178] In the history, candidate component types are defined. With the current sound category The relevant number of correct confirmations is The number of incorrect confirmations is The most recent confirmation time was Then construct the Bayesian historical association posterior mean: ; in, Candidate component type With the The posterior mean of the basic historical association of sound types; , These are prior parameters; Candidate component types in history With the The number of correct confirmations related to sound categories; Candidate component types in history With the Number of error confirmations related to sound.

[0179] Considering the time decay of historical records, we obtain the time-corrected historical correlation strength: ; in, To introduce the effective historical correlation strength after time decay; This is the historical time decay coefficient.

[0180] Candidate Components Historical Connections Explanation Evidence Represented as: ; in, This is a parameter for adjusting the sufficiency of historical samples.

[0181] For candidate components Let the set of abnormal conditions that trigger it be: ; in, Candidate components The set of exception conditions that trigger the event; This represents the total number of exception conditions that trigger the exception. This is a single abnormal condition.

[0182] Abnormal conditions include construction phase prohibitions, conflicts in finished product protection, conflicts in spatial obstruction, strong conflicts between component type and sound category, continuous construction noise in non-construction areas, and the appearance of breaking or cutting noises in accepted components.

[0183] For each abnormal condition Let its risk level be . Trigger strength is Then the probability of abnormal risk of the candidate component is: ; Evidence of abnormal constraints is represented as follows: ; in, Candidate components The overall probability of abnormal risk; Abnormal conditions Trigger strength; Abnormal conditions Risk level; The smaller the value, the less suitable the candidate component is as a matching object for the current construction acoustic event.

[0184] S4.8: Calculate the dynamic interpretability of candidate components for construction acoustic events. Convert geometric interpretation evidence, acoustic semantic interpretation evidence, BIM topology interpretation evidence, construction phase interpretation evidence, historical correlation evidence, and anomaly constraint evidence into logarithmic evidence form: ; in, Candidate components Construction acoustic events The ability to provide dynamic explanations; , , , , , Weighting of various types of dynamic evidence; Candidate components Geometric interpretation evidence; Candidate components Evidence for the acoustic semantic interpretation; Candidate components Evidence for BIM topology interpretation; Candidate components Evidence to explain the construction phase; Candidate components Historical connections explain the evidence; Candidate components Evidence to explain the abnormal constraints; It is a very small constant.

[0185] The dynamic interpretation probability of the candidate component is expressed as: ; in, Candidate components The dynamic interpretation probability; Candidate components Construction acoustic events The ability to provide dynamic explanations; The coefficient for enhancing the consistency of evidence; Candidate components Matching items for workable surfaces; This is the penalty coefficient for abnormal constraints; , , These are the convergence adjustment coefficients for the three types of uncertainties.

[0186] S4.9: Output the matching results of existing BIM candidate components based on the competitive advantage of the candidates. For the... Dynamic interpretation capability of all candidate components corresponding to a construction acoustic event After normalization, the posterior probability of the candidate component competition is obtained: ; in, Candidate components The posterior probability of competition; Candidate components The ability to provide dynamic explanations; The competition temperature parameter is used to adjust the sensitivity of the competition differences between candidate components; the smaller the value, the more obvious the difference between superior and inferior components. For construction acoustic events The total number of candidate components.

[0187] Let the candidate component corresponding to the highest competing posterior probability be: ;in, The optimal candidate component with the highest posterior probability; For construction acoustic events The complete set of candidate components; For maximization operators.

[0188] The highest competing posterior probability is: ; Let the posterior probability of the second-highest competition be... Then the candidate advantage interval is: ; Simultaneously calculate the candidate competition entropy: ; in, The candidate advantage interval is the difference between the best and second-best candidates; It represents the maximum posterior probability; This is the suboptimal posterior probability; It is a very small constant; For construction acoustic events The candidate competition entropy.

[0189] When the following conditions are met simultaneously: ; Then construction acoustic events Reliable matching to existing BIM candidate components And output reliable matching results: ; in, For construction acoustic events Reliable matching results; Optimal candidate component The dynamic interpretation of probability; This indicates that the acoustic event has been reliably attributed to an existing BIM component; This indicates that the source of the matched object is an existing BIM entity component.

[0190] The matching threshold, candidate advantage threshold, competition entropy threshold, and anomaly risk threshold are dynamically adjusted based on the uncertainty state of the event. ; ; ; ; in, For construction acoustic events The reliable matching threshold; For construction acoustic events The candidate discrimination interval threshold; For construction acoustic events The competitive entropy threshold; For construction acoustic events The comprehensive anomaly risk probability of the optimal candidate component; For construction acoustic events The optimal component anomaly risk threshold; , , , These are the corresponding fixed base thresholds; Indicates construction acoustic events Sound category uncertainty; For construction acoustic events The uncertainty in sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events Uncertainty of conflicts during the construction phase; , , , , , These are the adjustment coefficients for the corresponding uncertainties.

[0191] like The matching threshold has been reached, but Less than the candidate dominance threshold, or If the value is greater than the competition entropy threshold, it indicates that multiple existing BIM candidate components can explain the current acoustic event, and the output result is not unique. ; in, For construction acoustic events The candidate results are not unique; The threshold for selecting candidate candidates; The set of all candidate components whose posterior probability is higher than the candidate selection threshold; This indicates that there are multiple solutions for matching existing BIM components; Events marked as such require manual review.

[0192] If the dynamic interpretation probability of all candidate components is lower than the matching threshold, or the highest candidate component has a high anomaly risk, then it is determined that the existing BIM candidate components are insufficient to interpret the acoustic event, and the output is: ; in, For construction acoustic events Insufficient explanation of the event; This indicates that there are no valid matching BIM components; To explain the insufficient marker; The dynamic acoustic object generation process for the event flow to the next step.

[0193] If an acoustic event simultaneously meets the following conditions: ① Below the reliable matching threshold; ②The sound source is located in a non-solid area within a blank BIM area, a temporary work area, or between candidate components; ③ or Not entirely caused by noise; ④ There are events in the same spatial unit or adjacent spatial units that are not adequately explained in history; The acoustic event is then marked as a suspected dynamic acoustic object event, and the following output is given: ; in, For construction acoustic events The suspected dynamic acoustic object; This is a marker for a suspected temporary / mobile dynamic sound source. The dynamic acoustic object generation process for high-priority events to be transferred to the next step.

[0194] By merging the set of events with the suspected dynamic acoustic object, a global set of events with insufficient explanation is obtained.

[0195] S5. Based on the global under-explained events, calculate the joint probability contribution of the events to any point in space to delineate candidate dynamic acoustic regions, calculate the dynamic correlation between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence, generate dynamic acoustic topology nodes under the condition that the node confidence meets a preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multi-class topology associations between topology nodes and original nodes, and update to obtain the extended BIM spatial topology map.

[0196] S5.1: Construct a joint probability volume of sound source and under-explanation. For under-explanation acoustic events... Based on its event-weighted sound source localization center Centered on the sound source, a joint probability contribution of insufficient interpretation is constructed in the BIM space.

[0197] For any point in the BIM space , No. An underexplained event The joint probability contribution to this point is expressed as: ; in, For the first An underexplained event BIM spatial points The joint probability contribution; for The unexplained probability; for The normalization coefficient of sound source intensity; for Duration; For any three-dimensional coordinate point within the BIM space; for The weighted sound source localization center; for The spatial Gaussian distribution variance is used to control the probability diffusion range.

[0198] The unexplained probability refers to the probability that an existing BIM component cannot match the acoustic event in an underexplained event.

[0199] For multiple underexplained acoustic events within the same spatial range, their joint probability field is expressed as: When the following conditions are met within a spatial region: Then, this region is marked as a candidate dynamic acoustic region: ; in, Candidate dynamic acoustic regions; For spatial points The total joint probability field value after superimposing all adjacent events that are underexplained; The threshold for determining candidate dynamic acoustic regions.

[0200] S5.2: Dynamic evidence aggregation for underexplained acoustic events. For two underexplained acoustic events... , Spatial proximity, temporal continuity, sound category similarity, and unexplained consistency are calculated respectively.

[0201] Spatial proximity is represented as: ; in, ; Time continuity is represented as: ; Sound category similarity is represented as: ; Unexplained consistency is represented as: ; Taking all the above factors into account, the dynamic correlation between insufficient acoustic events is obtained: ; in, To explain the insufficiency event , The dynamic correlation degree; Spatial proximity; For time continuity; For sound category similarity; Unexplained consistency; To explain the insufficiency event , The mean variance; , These are events that are not fully explained. , Spatial variance; for The weighted sound source localization center; , They are respectively , The moment the event occurred; This is the time decay parameter; To explain the insufficient acoustic event Corrected sound category probability vector; To explain the insufficient acoustic event The unexplained probability.

[0202] When the following conditions are met: , This is the preset dynamic correlation threshold.

[0203] Therefore, it is considered that the two insufficiently explained acoustic events belong to the same dynamic acoustic event group.

[0204] After aggregation, a set of insufficiently explained acoustic events is formed: ; For the Event groups, create event group records: ; The center location of the event group is indicated as: ; The probability vector of the combined sound category of the event group is represented as follows: ; in, To explain the insufficient set of acoustic events; For the first The incident team explained the insufficient acoustic events; For the first The set of all acoustic events within the event group; For the first Synthetic sound probability vector within the event group; For the first The weighted sound source localization center of the event group; For the first The event group belongs to the BIM space unit; For the first The construction area of ​​the incident team; For the first Event group start time; For the first Total duration of the event group; For the first Unexplained probabilities of event groups; For the first Joint probability contribution of the event group; For the first The incident team provided insufficient explanation for the acoustic incident. The unexplained probability; To explain the insufficient acoustic event Corrected sound category probability vector.

[0205] S5.3: Calculate the confidence score for dynamic node generation. For the first... The group provided insufficient explanation for the acoustic incident. Calculate the confidence level of its dynamic node generation.

[0206] First, calculate the support level of the number of events: ; in, express The number of events supporting the event group; The total number of events within the group; This is a parameter for normalizing the number of events.

[0207] Calculate the duration of support: ; in, for The duration of support for the event group over time; for Total duration of the group; This represents the duration normalization parameter.

[0208] Computational spatial stability: ; in, for Spatial stability of event groups; for Dispersion of sound source locations within an event group; This represents the spatial dispersion normalization parameter.

[0209] The dispersion of sound source locations within an event group can be expressed as: ; Calculate sound category stability: ; in, for Sound category stability of the event group; for The event group's comprehensive sound category probability vector is the first... The probability of a sound type; This represents the total number of all construction sound categories.

[0210] The calculation explains the insufficient support: ; in, for The explanation provided by the event group lacks sufficient support. For the first An underexplained event The unexplained probability.

[0211] The combined confidence level for dynamic node generation is as follows: ; in, for Node confidence of the event group; , , , , Let each indicator have a weighting coefficient, and satisfy the following: .

[0212] If the following conditions are met: , If the node confidence threshold is used, then the event group is considered... The conditions for generating dynamic acoustic topology nodes are met. If not, the event group is marked as an acoustic event group to be observed, and subsequent acoustic events continue to be accumulated.

[0213] S5.4: Determine the node type of the dynamic acoustic topology. For event groups that meet the generation conditions... The dynamic acoustic topology node type is determined based on its sound category composition, spatial stability, duration, joint probability volume morphology, and BIM topology relationship.

[0214] Let the set of dynamic node types be:

[0215]

[0216] Representing temporary acoustic objects Virtual components Temporary work unit and sound source probability volume node .

[0217] Calculate the decision values ​​for the four types of nodes: ; ; ; ; in, for Temporary acoustic object scores for the event group; for The virtual component score of the event group; for The temporary task unit score for the event group; for The source probability volume node score of the event group; , , These are the weighting coefficients for each item in the score of the temporary acoustic object; , , These are the weighting coefficients for each item in the score of the virtual component; , , These are the weighting coefficients for each item in the score of the temporary work unit; , , These are the weight coefficients for each item in the score of the sound source probability volume node; express The proximity strength between the event group and existing BIM components or reserved construction areas in the surrounding area, with a value in the range of [0,1]. express The association strength between the event group and subsequent construction objects, unfinished components, or planned construction objects, with a value in the range of [0,1]. express Average location uncertainty of the event group.

[0218] The dynamic node type is determined as follows: ; in, for The optimal node type for an event group.

[0219] When the event group is concentrated in location, the sound type is singular, and the duration is short, a temporary acoustic object is generated. When the event group is in a stable position and is close to an unformed component, a reserved installation location, or a future construction object, a virtual component is generated. When an event group lasts for a long time and contains multiple types of construction sounds, a temporary work unit is generated. When the location of the event group is unstable but the range of the joint probability volume is obvious, generate a sound source probability volume node.

[0220] S5.5: Generate dynamic acoustic topology nodes. Based on event groups. and its node types Generate dynamic acoustic topology nodes: ; in, for Dynamic acoustic topology nodes of the event group; for The dynamic acoustic topology node numbering of the event group; for The optimal node type for an event group; for The weighted fusion of event groups yields a probability vector of sound categories. for The three-dimensional spatial center coordinates of dynamic nodes within an event group are obtained by weighted average of the sound source localization centers of all acoustic events within the event group. for The BIM spatial unit to which the dynamic nodes within the event group belong; for The construction area to which dynamic nodes within an event group belong; for Joint probability contribution of the event group; for The start time of the event group; for Total duration of the event group; for The original interpretation of dynamic nodes is insufficient for the set of acoustic events; express The dynamic node status of the event group.

[0221] The dynamic node status is determined based on the node generation confidence level and is divided into valid dynamic nodes, candidate dynamic nodes, and failed dynamic nodes, i.e.: When satisfied hour, If the lower confidence threshold is the threshold for a valid dynamic node, then it is a valid dynamic node, denoted as: ; When satisfied hour, If the lower confidence threshold is set to the threshold value for candidate dynamic nodes, then a node is considered a candidate dynamic node or a dynamic node awaiting review, denoted as: ; If an event group does not associate with new, insufficiently explained acoustic events for an extended period, or is subsequently reliably explained by existing BIM components, it is considered a failed dynamic node or an archived dynamic node, represented as... .

[0222] S5.6: Establish the topological relationship between the dynamic acoustic topology nodes and the BIM space. This involves generating the dynamic acoustic topology nodes... Access the BIM spatial topology map established in step S1.

[0223] If dynamic nodes Located in space unit If the elements are internal, then an inclusion relationship is established, which is represented as: ; If dynamic nodes Located in the construction area Within this area, a regional affiliation relationship is established, represented as follows: ; If dynamic nodes With existing BIM components Surface distance satisfies Then, a dynamic adjacency relationship is established, represented as: ,in, This represents the threshold for the adjacency distance between dynamic nodes and existing BIM components. If dynamic nodes With existing dynamic acoustic topology nodes The spatial ranges overlap, or the similarity of the sound categories of the two sounds meets the following criteria. Then, a dynamic node association relationship is established, represented as: ; If two dynamic nodes simultaneously satisfy the conditions of spatial overlap, similar sound category, and temporal continuity, then they are marked as nodes to be merged, represented as: .

[0224] Add dynamic acoustic topology nodes and their topology relationships to the BIM spatial topology map to form an extended BIM spatial topology map: ,in, Represents a set of dynamic acoustic topology nodes. This represents the set of topological relationships between dynamic acoustic topology nodes and spatial units, construction areas, existing BIM components, or other dynamic nodes.

[0225] S5.7: Output the dynamic acoustic topology node set. After completing the generation of dynamic acoustic topology nodes and the establishment of topology relationships, output the dynamic acoustic topology node set: ;in, This represents the number of nodes in the dynamic acoustic topology.

[0226] Each dynamic acoustic topology node is recorded as follows: ; Simultaneously output the expanded BIM space topology map: .

[0227] like Figure 3 As shown in the figure, the left side of the image shows the BIM plan space of the construction site, divided into... , Two independent spatial units, each containing various BIM building components, acoustic acquisition equipment, and sound source targets. (Microphone array) Collect data on construction noise generation. The acoustic events obtained by the algorithm. The corresponding BIM coordinates of the sound source; the dashed circle represents the uncertain range of the sound source, which is affected by the sound source location reliability of the microphone array. The larger the circle, the higher the positioning error. Walls floor slab Liang, Columns are static BIM entity components within the site and are candidate matching targets for sound source attribution. , , Representing the sound source points The Euclidean geometric distances to each BIM component are used as the basic matching criteria. Acoustic events are also considered. Using all BIM candidate components as input, the matching credibility is quantified by integrating six types of independent evidence. Matching results are then divided based on the judgment expressions of four indicators: if a certain BIM component ( Figure 3 middle If all beams meet the criteria, the construction acoustic event is determined to belong to this BIM component, and a clear sound source-component object attribution relationship is output and stored in a reliable matching set; if all BIM candidate components are determined to be under-explained events, dynamic acoustic topology nodes are automatically generated. Record the location, timing, and acoustic characteristics of the unknown sound source for subsequent updates of new components and tracking of abnormal sound sources.

[0228] S6. Based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, establish a matching record between construction acoustic events and dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal levels of construction behavior, and dynamically update the initial construction acoustic association library to achieve dynamic intelligent supervision of the construction site.

[0229] S6.1: Generate construction acoustic event—object matching record. Read the existing BIM candidate component matching result set output from step S4, read the dynamic acoustic topology node set output from step S5, and establish the construction acoustic event object matching record: ; in, Indicates an acoustic event Object matching records, Indicates an acoustic event The acoustic event number; Number the matched objects; To match the source of objects, they are divided into existing BIM components. Dynamic topology nodes ; Indicates an acoustic event Sound categories; For possible construction activities; Indicates an acoustic event The start time of the event, Indicates an acoustic event The duration of the event; For construction acoustic events BIM coordinates; Indicates the maximum posterior probability; Matching status flags.

[0230] For example, for construction acoustic events that reliably match existing BIM components, set: ; in, Optimal candidate component serial number; For the source of the matching object; To match credibility scores; Optimal candidate component The dynamic interpretation of probability; Location reliability for sound sources; The uncertainty in sound source localization; Indicates the matching status flag. This indicates a reliable match.

[0231] For events in step S5 where valid dynamic acoustic topology nodes have been generated, if the acoustic event... Belongs to dynamic nodes For the set of associated events, a dynamic node matching record is established: ; in, Number the dynamic nodes; For acoustic events The unexplained probability; Node confidence; This indicates that the matching status is marked as dynamic node matching.

[0232] For acoustic events in step S4 where the candidate object is not unique, the matching status is marked as manual review, as follows: .

[0233] For dynamic node association events that are in the candidate or low-confidence state in step S5, the matching status is set to "dynamic pending review", as shown below: .

[0234] This results in a set of matching records for all construction acoustic event objects. : ;in, This represents the total number of matching records for construction acoustic event objects in the set.

[0235] S6.2: Write the reliable matching results back to the BIM object. For construction acoustic events that match existing BIM entity components, write the corresponding matching results to the extended properties of that BIM component.

[0236] Construction acoustic incident The matching object is an existing BIM component. Then in the components Acoustic construction behavior attributes are written into it: ; in, This represents the BIM component after write-back. This indicates the extended attributes corresponding to construction acoustic events.

[0237] The extended attributes It should include at least: ① acoustic event number; ② sound category; ③ inferred construction behavior; ④ event occurrence time; ⑤ event duration; ⑥ sound source location; ⑦ matching score; ⑧ identification confidence; ⑨ location confidence; ⑩ verification status.

[0238] It can be represented as: ; in, This indicates whether the status is manually reviewed or automatically confirmed.

[0239] For the same BIM component that corresponds to multiple construction acoustic events within multiple time periods, a sequence of acoustic construction behavior records for the component is formed according to the chronological order of the events: ; in, Representing components Historical records of acoustic construction activities; Indicates the first Record of acoustic construction behavior of structural components This represents the total number of records of acoustic construction activities for structural components.

[0240] S6.3: Write back the dynamic acoustic topology nodes to the BIM space. For dynamic acoustic topology nodes... Based on its node type, center location, sound source-under-interpretation joint probability volume, spatial unit, construction area, and node status, dynamic acoustic object markers are generated in the BIM model: ; in, for Dynamic acoustic topology nodes in event groups Dynamic acoustic object markers generated in the BIM model; Number the dynamic nodes; for The optimal node type for event groups is dynamic acoustic topology nodes. The node type; The coordinates of the node center; for Joint probability contribution of the event group; for The BIM space unit to which the event group belongs; for The construction area to which dynamic nodes within an event group belong; for The weighted fusion of event groups yields a probability vector of sound categories. for Node confidence of the event group; express Dynamic node status of event groups; for The storage status of the BIM model for the event group.

[0241] Dynamic acoustic object labeling can be expressed through BIM custom parameters, IFC extended attributes, external database indexes, or BIM spatial annotation objects.

[0242] If the dynamic node type is a temporary acoustic object, then record its temporary sound source location, main sound category, and under-interpretation probability; If the dynamic node type is a virtual component, then record its possible future construction object type, spatial unit, event evidence set, and node generation confidence. If the dynamic node type is a temporary work unit, then record its work scope, main sound category, duration, construction activity intensity, and set of associated events; If the dynamic node type is a sound source probability volume node, then record its joint probability volume range, probability center, average unexplained probability, and set of associated acoustic events.

[0243] S6.4: Generate thermal tags for basic construction activities. Based on the construction acoustic event records written back to the BIM object and BIM space, generate thermal tags for basic construction activities.

[0244] For BIM objects or spatial units Its time window The activity value of construction activities within the area is represented as: ; Based on construction activity value Generate thermal level markers for corresponding BIM components, spatial units, or dynamic acoustic topology nodes: ; For existing reliable matching events of BIM components: ; For dynamic acoustic topology node association events: ; in, Represents BIM objects or spatial units In the time window The activity level of construction activities within the area; for The collection of all construction acoustic events within the unit; For construction acoustics incident; It represents the maximum posterior probability; for The sound source intensity coefficient; for Duration of the acoustic event; Indicates the construction thermal rating; , , The preset grading threshold is used to classify the thermal levels as 0 (no construction), 1 (low activity), 2 (medium activity), and 3 (high activity).

[0245] The aforementioned thermal markers of basic construction activities, as a form of expression of BIM write-back results, are used to indicate the activity level of construction activities of components, spatial units, or dynamic acoustic nodes within a specified time range.

[0246] S6.5: Obtain matching result feedback information. Obtain feedback information on the matching results of construction acoustic events, including: ① Reliable matching automatic confirmation results; ② Manual verification results; ③ Dynamic acoustic topology node confirmation results; ④ Subsequent construction status verification results; ⑤ Incorrect matching markers; ⑥ Low confidence markers pending verification.

[0247] For the For each matching result, establish a feedback record. : ; in, Indicates the first The acoustic event number of each matching result; For the first The sound category of each matching result; For the first Possible construction behaviors for each matching result; Indicates the first Feedback results of each matching result Indicates the first The credibility of the feedback for each matching result.

[0248] Feedback results This includes: ① Correct match; ② Incorrect match; ③ Behavior type correction; ④ Object type correction; ⑤ Need to continue observation; ⑥ Invalid acoustic event; ⑦ Dynamic node merging confirmation; ⑧ Dynamic node failure confirmation.

[0249] S6.6: Update the dynamic construction acoustics association library. Update the dynamic construction acoustics association library based on construction acoustics event-object matching records and feedback records.

[0250] For a single association record in the construction acoustics association database: ; If the feedback record indicates that the combination of component type, material, construction stage, sound category, and construction behavior matches correctly, the weight of the associated record is increased; if the feedback record indicates that the combination does not match correctly, the weight of the associated record is decreased.

[0251] Let the association weight before the update be... The feedback value is Update step size is The updated new association weight for: ; in, When the feedback result indicates a correct match, Take the higher value; when the feedback result is a match error, Take the lower value; when the feedback result is "pending review", reduce the update step size. Or it may not be updated for the time being; The value (0,1) is used to control the magnitude of weight correction in a single feedback.

[0252] To ensure that the association weights are within a valid range, a normalization constraint is applied to the updated association weights. : ;in, This represents the new association weights after normalization constraints.

[0253] If the feedback results show new combinations not included in the construction acoustics association library, then add a new dynamic association record: ; in, Indicates the type of the newly added object; This indicates the addition of new object materials or object attributes. Indicates the construction stage or the relevant specialty. Indicates the sound category, This indicates potential construction activity. Indicates the initial dynamic association weight. Indicates an abnormal or prohibited condition.

[0254] The updated dynamic construction acoustic association library is represented as follows: .

[0255] By combining the initial construction acoustic association library and the dynamic construction acoustic association library, the construction acoustic association library to be used in the next round of matching is obtained: .

[0256] S6.7: Update the status of dynamic acoustic topology nodes. Update the status of dynamic acoustic topology nodes based on the new construction acoustic event matching results and feedback results.

[0257] For dynamic acoustic topology nodes If subsequent construction acoustic events continue to match the node, then update its event set, duration, node center location, and node confidence.

[0258] The node-related event set has been updated to: ,in, This is the set of events associated with the original node; For newly matched acoustic events.

[0259] The updated node center location is: ; in, This indicates the updated center position of the node.

[0260] If a dynamic node is not associated with any new acoustic events for an extended period of time, or is confirmed as an invalid node after manual review, then the node's status will be set to invalid, archived, or pending deletion.

[0261] If a dynamic node is repeatedly confirmed to stably correspond to a certain type of construction object or construction behavior, then the node type, sound category, and construction behavior are written into the dynamic construction acoustic association library.

[0262] S6.8: Output the final results. Output the set of construction acoustic event matching results, BIM write-back results, dynamic acoustic topology node set, and updated dynamic construction acoustic association library.

[0263] Final output results Represented as: ; in, This represents the set of matching records for construction acoustic event objects. Represents a set of dynamic acoustic topology nodes. This represents the expanded BIM spatial topology map. This indicates the updated construction acoustics association library. This represents a collection of construction acoustic history records for BIM components, spatial units, or dynamic nodes.

[0264] In summary, compared with the prior art, the present invention has the following beneficial effects: This invention addresses several issues, including difficulty in accurately attributing construction audio recognition results to specific BIM objects, unstable component matching due to sound source localization errors, insufficient utilization of BIM spatial topology information, inability to represent unmodeled temporary objects in BIM, and the inability to dynamically update construction acoustic association rules. Through the above technical solution, the accuracy of object matching for construction acoustic events in BIM space, the interpretability of results, and project adaptability can be improved, as detailed below: (1) By standardizing the BIM model data using steps S1.2 and S1.3 and converting the component geometric representation into a surface patch model, the coordinate system, component number, material properties, and geometric representation of BIM components can be unified, solving the problems of inconsistent data formats of different BIM models and difficulty in accurately calling the spatial position of components. By calculating the spatial relationship between the sound source and the component through the component surface patch model, the geometric error caused by using only the component center point or bounding box for judgment can be reduced.

[0265] (2) By establishing BIM spatial topology relationships using step S1.4, the relationships between components and rooms, floors, construction areas, adjacent components, and attached components can be organized into a calculable topology. This solves the problem in the existing technology of only using component coordinates for matching and not making full use of BIM spatial relationships, and enables subsequent sound source object screening to consider constraints such as same space, adjacent space, connectivity and attachment relationships.

[0266] (3) By performing sound recognition, event merging, sound source localization, and BIM coordinate mapping on the construction audio in steps S2.3 to S2.7, continuous construction audio can be transformed into construction acoustic events with sound category, category probability, occurrence time, duration, sound intensity, sound source location, and BIM spatial affiliation. This can reduce the interference of fragmented original audio segments and non-construction sounds on subsequent matching, and enable construction sounds to have a structured data foundation for matching with BIM objects.

[0267] (4) Using steps S3.1 to S3.6 to filter existing BIM candidate components based on BIM spatial topology, the spatial retrieval range and candidate component types can be determined first based on the floor, spatial unit, construction area, and sound category of the construction acoustic event. Then, the existing BIM entity components are filtered and topologically expanded by combining BIM spatial topological relationships, component surface distances, component adjacency relationships, spatial connectivity relationships, attachment relationships, and spatial barrier relationships. This avoids traversing all components in the entire BIM model and reduces the amount of subsequent matching score calculations. At the same time, by combining acoustic semantic initial screening, geometric nearest neighbor screening, and topological association supplementation, the omission of candidate components caused by screening only based on the distance of the sound source point can be reduced, and the completeness and rationality of the existing BIM component candidate set can be improved.

[0268] (5) By using steps S4.1 to S4.9 to correct the dynamic interpretation probability of existing BIM candidate components based on the uncertainty state of the event and the reliability of the evidence, the geometric location of the sound source, the BIM spatial topology relationship, the construction acoustic association library, the construction stage status, the dynamic historical association, and the abnormal constraint conditions can be jointly incorporated into the process of judging the interpretation capability of candidate components. This invention determines the uncertainty state of the current acoustic event based on the uncertainty of the sound category, the uncertainty of the sound source location, the degree of competition of candidate components, the spatial topology conflict, and the construction stage conflict, and then dynamically adjusts the reliability of geometric evidence, acoustic semantic evidence, topological evidence, stage evidence, historical evidence, and abnormal constraint evidence accordingly, thereby avoiding over-reliance on a single piece of evidence when the positioning error is large, the sound category is confused, or the candidate components are dense. By comparing the competitive advantages between the highest candidate component and other candidate components, states such as reliable matching, non-unique candidates, and insufficient interpretation of existing BIM objects can be distinguished, reducing the situation of misattributing construction sounds to components that are close but have unreasonable construction semantics, improving the accuracy and stability of the object attribution judgment between construction acoustic events and existing BIM components, and providing a more reliable judgment basis for the subsequent generation of dynamic acoustic topology nodes.

[0269] (6) By using steps S5.1 to S5.7 to generate dynamic acoustic topology nodes based on the accumulation of insufficiently explained evidence, acoustic events that cannot be reasonably explained by existing BIM candidate components can be further organized into computable and updatable dynamic acoustic objects. By constructing a joint probability volume of sound source and insufficient explanation, it is possible to express the possible distribution of sound sources in the BIM space, and also to reflect the degree to which construction sounds in the area are difficult to explain by existing BIM components. By dynamically aggregating the insufficiently explained acoustic events and combining the number of events, duration, spatial stability, sound category stability and average unexplained probability to calculate the confidence level of dynamic node generation, the situation of accidental generation of dynamic nodes due to occasional noise or single low-confidence events can be reduced. Thus, temporary processing tables, material stacking areas, mobile equipment, temporary supports, unformed components or temporary work areas that are not explicitly expressed in BIM can be represented as temporary acoustic objects, virtual components, temporary work units or sound source probability volume nodes, and connected to the BIM space topology relationship, thereby solving the problem that the BIM model only contains formal design components and is difficult to express temporary construction objects and dynamic construction activity areas on site.

[0270] (7) Using steps S6.2 to S6.5, the matching results and dynamic acoustic nodes are written back to the BIM model. Sound category, construction behavior, event time, matching confidence, and construction activity level can be recorded in BIM components, spatial units, or dynamic nodes. This allows the construction audio recognition results to be transformed from independent sound data into construction behavior records in the BIM space, providing a data foundation for subsequent construction behavior heatmaps, construction status tracking, and abnormal operation judgment.

[0271] (9) By using steps S6.6 to S6.8 to update the dynamic construction acoustic association library based on the matching results, manual review results, and subsequent construction status feedback, the association weights between sound categories, component attributes, and construction behaviors can be continuously corrected. This can reduce the problem of insufficient adaptability of fixed rules to different projects, different construction stages, and different construction methods, and enable the system to gradually form project-specific construction acoustic association relationships after multiple runs.

[0272] Example 2 like Figure 4 As shown, the second embodiment of the present invention also provides a dynamic monitoring system based on BIM spatial topology and construction sound matching, including: The BIM spatial topology generation unit is used to construct a BIM spatial topology model and an initial construction acoustic association library. It performs analysis and standardization processing on the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map. The construction acoustic event set generation unit is used to collect the original audio signals at the construction site, perform preprocessing, segmentation and construction sound recognition, and then merge them to form construction acoustic events. The unit also performs sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a construction acoustic event set. The BIM candidate component screening unit is used to determine the spatial retrieval range based on the construction acoustic event set and the BIM spatial topology map, and at the same time, to perform preliminary screening of candidate component types in conjunction with the initial construction acoustic association library to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, it calculates the minimum distance from the sound source point of the existing BIM entity component to the component surface, generates event-related topology tags, and obtains the existing BIM candidate component set. The component acoustic event matching unit is used to calculate the reliability of various types of evidence to obtain dynamic evidence weights based on the construction acoustic event set and the existing BIM candidate component set, from sound recognition, sound source localization, and candidate component competition to construct the event uncertainty state. Then, it calculates the dynamic explanatory ability of candidate components for construction acoustic events by weighted fusion of dynamic evidence, and combines the candidate competitive advantage adaptive matching threshold for comprehensive judgment, and outputs reliable matching results, candidate non-unique results, or events with insufficient global explanation. The BIM spatial topology map extension unit is used to calculate the joint probability contribution of events to any point in space based on the global under-explained events, so as to delineate candidate dynamic acoustic regions, calculate the dynamic correlation degree between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence degree, generate dynamic acoustic topology nodes under the condition that the node confidence degree meets the preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multiple types of topology associations between topology nodes and original nodes, and update to obtain the extended BIM spatial topology map; The dynamic update unit is used to establish a matching record between construction acoustic events and dynamic acoustic topology nodes based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal level of construction behavior, and dynamically update the initial construction acoustic association library to realize dynamic intelligent supervision of the construction site.

[0273] Example 3 The third embodiment of the present invention also provides a dynamic monitoring device based on BIM spatial topology and construction sound matching, which includes a memory and a processor. The memory stores a computer program, which can be executed by the processor to realize the dynamic monitoring method based on BIM spatial topology and construction sound matching as described above.

[0274] Example 4 The fourth embodiment of the present invention also provides a computer-readable storage medium storing computer-readable instructions. When the computer-readable instructions are executed by the processor of the device where the computer-readable storage medium is located, the dynamic monitoring method based on BIM spatial topology and construction sound matching as described above is implemented.

[0275] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic monitoring method based on BIM spatial topology and construction sound matching, characterized in that, include: S1. Construct a BIM spatial topology model and an initial construction acoustic association library. Analyze and standardize the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map. S2, collect the original audio signals from the construction site, perform preprocessing, segmentation and construction sound recognition, and then merge them to form construction acoustic events. Then, perform sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a set of construction acoustic events. S3. Based on the set of construction acoustic events and the BIM spatial topology map, determine the spatial retrieval range. At the same time, combine the initial construction acoustic association library to perform preliminary screening of candidate component types to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, calculate the minimum distance from the sound source point of the existing BIM entity component to the component surface, generate event-related topology tags, and obtain the existing BIM candidate component set. S4. Based on the set of construction acoustic events and the existing set of BIM candidate components, the uncertainty state of the event is constructed from sound recognition, sound source localization, and candidate component competition. The reliability of various types of evidence is calculated to obtain dynamic evidence weights. Then, the dynamic evidence is weighted and fused to calculate the dynamic explanatory ability of candidate components for construction acoustic events. At the same time, the candidate competitive advantage adaptive matching threshold is combined to make a comprehensive judgment and output reliable matching results, candidate non-unique results, or events with insufficient global explanation. S5. Based on the global under-explained events, calculate the joint probability contribution of the events to any point in space to delineate candidate dynamic acoustic regions, calculate the dynamic correlation between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence, generate dynamic acoustic topology nodes under the condition that the node confidence meets a preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multiple topology associations between topology nodes and original nodes, and update to obtain the extended BIM space topology map; S6. Based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, establish a matching record between construction acoustic events and dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal levels of construction behavior, and dynamically update the initial construction acoustic association library to achieve dynamic intelligent supervision of the construction site.

2. The dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 1, characterized in that... S1 specifically refers to: The BIM spatial topology model is exported as an IFC format file. Data containing at least building component entities, spatial entities, floor entities, spatial structural relationships, material relationships, attribute sets, and geometric representation information in the IFC format file is read to obtain BIM data. Then, the BIM data is standardized to convert the local coordinates of the components into global BIM coordinates, resulting in standardized basic attribute records of the components. The spatial topological relationships include direct inclusion relationships, spatial affiliation relationships, inter-component adjacency relationships, spatial connectivity relationships, and component attachment relationships; Wherein, the direct inclusion relationship is a component Belongs to spatial unit ,Right now ; The spatial attribution relationship is determined by judging the geometric center point of the component. Does it fall within a spatial unit? Within the boundary, if it is a component Belonging to Among them, the geometric center point of the component The expression is: ； in, For components The set of vertices on the surface of the component. Indicates the number of vertices. express The three-dimensional coordinates of the vertices in the diagram; By calculating components With spatial units The overlap ratio of the components The primary home space; the formula for calculating the overlap ratio is: ； in, Representing components With spatial units The overlap ratio; Representing components Located in space unit Internal volume; Representing components The total volume; The adjacency relationship between components is determined by calculating the minimum distance between the surfaces of two components, and the minimum distance is less than a preset adjacency distance threshold. The spatial unit connectivity is such that two spatial units are connected through doorways, corridors, or open boundaries; The component attachment relationship refers to the existence of an IFC support / connection relationship between auxiliary components such as pipes, cable trays, and supports and the main wall / beam / slab, or the minimum distance between the component surfaces is less than a preset attachment threshold, which is expressed as follows: ,in For attachment components, As the main component; Then, a BIM spatial topology map is constructed, consisting of a set of topological nodes and a set of spatial topological edges; wherein, the set of topological nodes includes component nodes, spatial unit nodes, floor nodes, and construction area nodes; and the set of topological edges consists of edges belonging to the spatial topological relationships. The initial construction acoustic association library is established based on construction method knowledge, historical construction audio samples, manual annotation results, and construction task rules. It is used to record the correspondence between component attributes, construction sound categories, and construction behaviors.

3. The dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 1, characterized in that... The construction acoustic events include an acoustic event number, event sound category, event identification confidence level, event start time, event duration, event sound intensity, and a set of audio segments corresponding to the event. The generation process is as follows: The audio segments after segmentation are input into the construction sound recognition model to obtain the sound category probability vector, thereby obtaining the sound category and recognition confidence of the corresponding audio segment; Audio segments that are adjacent audio segments and meet the conditions of having the same sound category and a time interval of less than a set time threshold are merged into the same acoustic event. The event recognition confidence level is the average of the recognition confidence levels of the merged audio segments; The sound intensity of the event is the average sound energy over the duration of the event, expressed as: ； in, Indicates the first Construction acoustics incident The average sound energy; Indicates an acoustic event The number of audio segments included; Indicates an acoustic event The corresponding set of audio clips; For the first The sound energy of an audio segment; When locating the sound source of the construction acoustic event, the sound source location is calculated based on the arrival time difference between different acquisition channels of the microphone array, and the sound source coordinates are solved using global least squares. The expression is as follows: ； in, For acoustic events The corresponding three-dimensional coordinates of the sound source; For the microphone pairs involved in positioning; , The first , The location coordinates of each microphone; The speed of sound in air; For the first , The time difference of arrival of each microphone; Let be the three-dimensional coordinates of the sound source to be solved; Represents Euclidean spatial distance; It is a least squares operator; Then the sound source location is mapped to the BIM coordinate system to obtain the position of the sound source in the BIM coordinate system; During sound source localization, the sound source localization reliability is calculated based on cross-correlation peak sharpness, time delay residual consistency, and continuous frame localization stability. The expression is: ； in, Indicates an acoustic event The corresponding sound source localization reliability; For acoustic events Corresponding peak clarity; For acoustic events The actual measured delay is consistent with the theoretical delay; For acoustic events Corresponding continuous frame positioning stability; , , They are respectively , , Weighting coefficients; The closer the value is to 1, the more reliable the positioning. The closer the value is to 0, the less reliable the positioning is; This results in a set of construction acoustic events that includes the construction acoustic events, the location of the sound source in the BIM coordinate system, the spatial unit to which it belongs, the construction area to which it belongs, and the location confidence of the sound source.

4. The dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 3, characterized in that... The expression for the spatial retrieval range is: ； in, For acoustic events The corresponding spatial retrieval range; For acoustic events Its own spatial unit; To and The set of all adjacent spatial units; To and Connected and falling within the spatial retrieval radius A collection of spatial units within; Spatial search radius The calculation formula is: ； in, Indicates construction acoustic events Spatial retrieval radius; Show sound category The corresponding basic search radius; This represents the positioning uncertainty correction coefficient; Indicates an acoustic event The corresponding sound source localization reliability; When conducting the initial screening of candidate component types, the acoustic prior association value of the component type is calculated based on the sound category probability of the acoustic event and the association weight of the initial construction acoustic association library. Component types with acoustic prior association values ​​exceeding the preset initial screening threshold are included in the candidate component type set. The acoustic prior correlation value is calculated as follows: ； in, For component type Relative acoustic events Acoustic prior correlation values; For acoustic events Identified as a sound category The probability of; The association weights for the initial construction acoustic association library; The first one represents the initial construction acoustic association library. 1 associated record; The formula for calculating the minimum distance from the sound source point to the surface of an existing BIM entity component within the spatial retrieval range is as follows: ； in, For acoustic events Components within the spatial retrieval range The shortest distance between the triangular facets of the surface; This indicates finding the minimum value; For acoustic events The corresponding BIM coordinates of the sound source; For components within the spatial retrieval range A collection of triangular facets on the surface of a component; For components within the spatial retrieval range The first set of triangular facets on the surface of the component A triangular facet; This is the Euclidean shortest distance; like Less than the spatial search radius Then the components Add to the existing BIM entity candidate set; When generating event-related topology tags, based on the existing BIM entity candidate set, topology-related components in the BIM spatial topology map that have component attachment, component adjacency, component connection or component affiliation relationships with candidate components in the existing BIM entity candidate set are included in the candidate set, and the candidate set is updated after topology expansion. Then, after calculating the topology expansion, the candidate objects in the candidate set are updated with construction acoustic events. The event-related topology tags between them are used to obtain a complete set of existing BIM candidate components; wherein, the event-related topology tags include tags of the same spatial unit, tags of the same construction area, tags of connected space, tags of attachment relationship, tags of spatial isolation, tags of acoustic semantic screening, and tags of geometric nearest neighbors; The spatial barrier markers are used to characterize the BIM coordinates of the sound source. To candidate BIM components Are there any sound-insulating or shielding components along the straight path from the nearest point on the surface? 5. A dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 4, characterized in that... The uncertainty states of the construction acoustic events include sound category uncertainty, sound source localization uncertainty, candidate component competition uncertainty, topological conflict degree, construction stage conflict degree, and historical sample insufficiency degree, expressed as: ； in, For construction acoustic events The uncertain state vector; Indicates construction acoustic events Sound category uncertainty; For construction acoustic events The uncertainty in sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events Uncertainty of conflicts during the construction phase; For construction acoustic events The uncertainty due to insufficient historical samples; Among them, the sound category uncertainty The expression is determined by comprehensively considering the dispersion of sound category distribution, category discrimination, category confusion sensitivity, and intra-event category volatility, and is as follows: ； ； ； ； ； in, Indicates construction acoustic events The dispersion of sound category distribution; Indicates construction acoustic events Category discrimination; Indicates construction acoustic events Category confusion sensitivity; Indicates construction acoustic events The volatility of the internal category of the event; , , , This is an adjustment coefficient for the uncertainty of the sound category; It is an exponential function; This represents the total number of all construction sound categories. Construction acoustic events after obfuscation correction Belongs to the Normalized probability of construction-related sounds; It is the natural logarithm; , These are the construction acoustic events after the confusion correction. The maximum and second largest values ​​in the normalized probability vector; , Indexing two different categories of construction noise; For construction acoustic events belong The corrected probability of the sound type; for Sound-like The risk factor for sound confusion; For construction acoustic events The number of audio segments included; For the first The obfuscation-corrected sound category probability vector of each audio segment; Indicates the first The optimal sound category determined for each audio segment ; This is an indicator function; it takes the value 1 if the condition is true, and 0 otherwise. This represents the total number of audio segments that are most frequently classified into the same category among all audio segments. The formula for calculating the normalized probability vector of construction acoustic events after confusion correction is as follows: ； ； in, To obscure the corrected construction acoustics incident The normalized probability vector; For construction acoustic events The original sound category probability vector; For the originally identified construction acoustic events The The probability of construction-like sounds; Indicates vector normalization; This represents a truncation function; it sets negative numbers to 0 and retains positive numbers. For sound category confusion matrix; It is a minimal constant to prevent the matrix from becoming singular and non-invertible; for An identity matrix of order 1; The uncertainty of sound source localization Based on the comprehensive determination of the positioning uncertainty ellipsoid volume index, whitening time delay residual, and spatial affiliation jump degree, the expression is: ； ； ； ； ； ； ； in, For construction acoustic events The positional uncertainty of the ellipsoidal volume logarithmic index; For construction acoustic events The mean of whitening delay residuals; For construction acoustic events Spatial affiliation jump degree; , , These are the adjustment coefficients for the uncertainty of sound source localization; This refers to matrix determinant operations. For construction acoustic events The positioning covariance matrix; For construction acoustic events The total number of short-time frames; For construction acoustic events The Positioning weights for each short frame; For construction acoustic events The Three-dimensional sound source localization coordinates in a short time frame; For construction acoustic events The weighted sound source localization center; It is the transpose matrix; for No. The audio signal-to-noise ratio of a short frame; for No. Cross-correlation peak sharpness of a short time frame; for No. The measured latency of a short frame is consistent with the theoretical latency. A set of paired microphones for participating in positioning; This represents the total number of pairs; For microphone pair Actual measured sound arrival time difference; For the reason Theoretical arrival time difference calculated in reverse; For microphone pair Delay estimation noise standard deviation; This indicates that the positioning point falls into the spatial unit. The number of short-time frames; This represents the cell count that contains the most positioning frames among all spatial cells. The candidate component competition uncertainty The expression is determined by combining the competition entropy of candidate components and the effective density of candidate components, and is as follows: ； ； ； ； ； in, For construction acoustic events The entropy of competition among candidate components; For construction acoustic events Effective density of candidate components; , This is an adjustment coefficient for competition uncertainty; For construction acoustic events The total number of candidate components; For construction acoustic events Candidate Components Normalized initial competition probability; For construction acoustic events Candidate Components The initial competitive potential energy; BIM coordinates of the sound source To candidate components The minimum Euclidean distance between the triangular facets of the surface; This is the distance attenuation parameter; For construction acoustic events candidate components Acoustic prior association value with the current sound category; For initial acoustic semantic screening of components, use binary tags, where 1 represents a construction acoustic event. candidate components It belongs to the acoustic semantic preliminary screening component; For topology expansion candidate components, use binary labels, where 1 represents a construction acoustic event. candidate components ... , , The adjustment coefficient representing the initial competitive potential energy; The topological conflict degree The expression is: ； in, For construction acoustic events The number of candidate components that exist in the space are spatially blocked, propagated across space, in non-connected spaces, or have conflicting attachment relationships; The degree of conflict during the construction phase The expression is: ； in, For construction acoustic events The number of candidate components that conflict with the construction behavior corresponding to the current sound during the construction phase; The historical sample insufficiency The expression is: ； in, For construction acoustic events The number of historical samples that match the current component, sound, and construction stage; This is the historical sample sufficiency adjustment constant.

6. A dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 5, characterized in that... The dynamic interpretation capability is obtained by weighted logarithmic fusion of geometric interpretation evidence, acoustic semantic interpretation evidence, BIM topology interpretation evidence, construction phase interpretation evidence, historical correlation interpretation evidence, and anomaly constraint interpretation evidence, expressed as: ； in, Candidate components Construction acoustic events Dynamic interpretation capability, candidate components For construction acoustic events candidate components ; , , , , , Weighting of various types of dynamic evidence; Candidate components Geometric interpretation evidence; Candidate components Evidence for the acoustic semantic interpretation; Candidate components Evidence for BIM topology interpretation; Candidate components Evidence to explain the construction phase; Candidate components Historical connections explain the evidence; Candidate components Evidence to explain the abnormal constraints; It is a very small constant; The expression for dynamic evidence weight is: ； ； ； ； ； ； ； in, For the first Similar evidence in construction acoustic events The dynamic weights under the condition are all summed to 1; This is the evidence scaling factor; This is the evidence bias constant; For construction acoustic events The Reliability of similar evidence; For construction acoustic events The reliability of geometric distance evidence; For construction acoustic events The reliability of sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events The reliability of acoustic semantic evidence; For construction acoustic events Reliability of sound category recognition; For construction acoustic events Category confusion sensitivity; For construction acoustic events The volatility of the internal category of the event; For construction acoustic events Uncertainty of the location of a fixed sound source; For construction acoustic events The reliability of BIM topology evidence; For construction acoustic events The reliability of evidence during the construction phase; For construction acoustic events Sound category uncertainty; For construction acoustic events Reliability of historical matching association evidence; For construction acoustic events The uncertainty due to insufficient historical samples; For construction acoustic events The reliability of evidence regarding abnormal constraint penalties; For construction acoustic events Uncertainty of conflicts during the construction phase; , , , , , , , These are the attenuation adjustment coefficients for each item; The expression for the geometric interpretation evidence is: ； ； ； in, Weighted sound source localization center To candidate components The minimum Mahalanobis distance of the surface; The coefficient of performance for workable surfaces; Candidate components Workable surface matching correction items; For construction acoustic events The weighted sound source localization center; Candidate components Any three-dimensional vertex on the surface patch; For construction acoustic events The inverse matrix of the positioning covariance matrix; Candidate components The set of all surface vertices; For dough The binary markers are 1 for workable surfaces and 0 for non-workable surfaces; For dough The normal vector; The vector representing the main propagation direction of the sound source; This indicates finding the minimum value; This indicates finding the maximum value; Evidence for Acoustic Semantic Interpretation The expression is: ； in, Construction acoustic events after obfuscation correction Belongs to the Normalized probability of construction-related sounds; This represents the total number of all construction sound categories. For the initial construction acoustic association library; Candidate components Component types; Candidate components Component materials; Candidate components The component construction stage; BIM Topology Interpretation Evidence The expression is: ； ； in, For construction acoustic events spatial unit To candidate components The minimum total cost of the topological path; For sound source and candidate components The spatial obstruction penalty term is generated by the spatial obstruction marker; This represents all topological connectivity paths from the spatial unit where the sound source is located to the candidate component; It is a single topological edge; Topological edges The cost of passage; Explanation of evidence during the construction phase The expression is: ； ； in, For construction acoustic events Time of the event With candidate components The components allow for construction windows Time window deviation; This refers to the decay parameter within the time window. Candidate components Component types; Candidate components The construction phase; For construction acoustic events Corrected sound probability vector; The degree to which the candidate component type and construction stage are adapted to the current modified sound probability vector; It is a time-distance function; Historical Connections Explanation Evidence The expression is: ； ； ； in, This is an adjustment parameter for the sufficiency of historical samples; Candidate component type With the The posterior mean of the basic historical association of sound types; Candidate component types in history With the The number of correct confirmations related to sound categories; Candidate component types in history With the Number of error confirmations related to sound; To introduce the effective historical correlation strength after time decay; This is the historical time decay coefficient; , These are prior parameters; Evidence for Explanation of Abnormal Constraints The expression is: ； ； ； in, Candidate components The overall probability of abnormal risks; Candidate components The set of exception conditions that trigger the event; This represents the total number of exception conditions that trigger the exception. This is a single abnormal condition; Abnormal conditions Trigger strength; Abnormal conditions Risk level; Based on the aforementioned dynamic interpretation capability, the dynamic interpretation probability of the candidate component is obtained, expressed as follows: ； in, Candidate components The dynamic interpretation probability; Candidate components Construction acoustic events The ability to provide dynamic explanations; The coefficient for enhancing the consistency of evidence; Candidate components Matching items for workable surfaces; This is the penalty coefficient for abnormal constraints; , , These are the convergence adjustment coefficients for the three types of uncertainties.

7. A dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 6, characterized in that... The system combines candidate competitive advantage with adaptive matching threshold to make a comprehensive judgment, outputting reliable matching results, non-unique candidate results, or insufficiently explained events, specifically: Based on the dynamic interpretation capability and the dynamic interpretation probability, the competitive posterior probability of all candidate components is obtained through temperature normalization. The candidate advantage interval and candidate competition entropy of the best candidate component and the second best candidate component are calculated. At the same time, the reliable matching threshold, candidate differentiation interval threshold, competition entropy threshold and the optimal component abnormal risk threshold are adaptively generated in combination with the uncertainty state of construction acoustic events. The four indicators of maximum posterior probability, candidate advantage interval, candidate competition entropy and optimal component abnormal risk are compared with the corresponding dynamic thresholds for judgment. If all four judgment indicators meet the judgment conditions, the set of reliable matching results for the candidate components of the current construction acoustic event will be output. Maximum posterior probability only Meets the criteria and has a candidate advantage interval Less than the threshold or candidate competition entropy If the result exceeds the threshold, output a set of non-unique candidate results. If the maximum posterior probability Meeting the judgment criteria or optimal component abnormal risk If the threshold is exceeded, it is determined to be an insufficient set of events; If an underexplained event simultaneously meets the following conditions, it will be additionally marked as a suspected dynamic acoustic object and uniformly included in the set of underexplained events. The conditions are: The maximum posterior probability is lower than the reliable matching threshold, and the sound source is located in a non-physical area between BIM blank areas, temporary work areas, or candidate components, and the sound category uncertainty or sound source location uncertainty is not entirely caused by noise, and there are events with insufficient historical interpretation in the same spatial unit or adjacent spatial units. The expression for the competitive posterior probability is: ； in, Candidate components The posterior probability of competition; Candidate components The ability to provide dynamic explanations; For competing temperature parameters; For construction acoustic events The total number of candidate components; The expression for the optimal candidate component is: ； in, The optimal candidate component with the highest posterior probability; For construction acoustic events The complete set of candidate components; For maximization operators; The expression for the candidate dominance interval is: ； ； in, The candidate advantage interval is the difference between the best and second-best candidates; It represents the maximum posterior probability; This is the suboptimal posterior probability; It is a very small constant; The expression for the candidate competition entropy is: ； in, For construction acoustic events The candidate competition entropy; The judgment expressions for the four indicators are as follows: ； ； ； ； ； in, For construction acoustic events The reliable matching threshold; For construction acoustic events The candidate discrimination interval threshold; For construction acoustic events The competitive entropy threshold; For construction acoustic events The comprehensive anomaly risk probability of the optimal candidate component; For construction acoustic events The optimal component anomaly risk threshold; , , , These are the corresponding fixed base thresholds; Indicates construction acoustic events Sound category uncertainty; For construction acoustic events The uncertainty in sound source localization; For construction acoustic events Uncertainty of competition among candidate components; For construction acoustic events Topological conflict uncertainty; For construction acoustic events Uncertainty of conflicts during the construction phase; , , , , , These are the adjustment coefficients corresponding to the uncertainty; The expression for the reliable matching result set is: ； in, For construction acoustic events Reliable matching results; Optimal candidate component The dynamic interpretation of probability; This indicates that the acoustic event has been reliably attributed to an existing BIM component; This indicates that the source of the matched object is an existing BIM entity component; The expression for the set of candidate non-unique results is: ； in, For construction acoustic events The candidate results are not unique; The threshold for selecting candidate candidates; The set of all candidate components whose posterior probability is higher than the candidate selection threshold; This indicates that there are multiple solutions for matching existing BIM components; Events must be manually reviewed before they can be marked. The expression for the set of events with insufficient explanation is: ； in, For construction acoustic events Insufficient explanation of the event; This indicates that there are no valid matching BIM components; To explain the insufficient marker; The process of generating dynamic acoustic objects for the event flow to the next step; The expression for the suspected dynamic acoustic object is: ； in, For construction acoustic events The suspected dynamic acoustic object; This is a marker for a suspected temporary / mobile dynamic sound source. A dynamic acoustic object generation process for high-priority events to be transferred to the next step; By merging the set of events with the suspected dynamic acoustic object, a global set of events with insufficient explanation is obtained.

8. A dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 7, characterized in that... The expression for the joint probability contribution is: ； in, For the first An underexplained event BIM spatial points The joint probability contribution; for The unexplained probability; for The normalization coefficient of the sound source intensity; for Duration; For any three-dimensional coordinate point within the BIM space; for The weighted sound source localization center; for The spatial Gaussian distribution variance is used to control the probability diffusion range; The expression for the candidate dynamic acoustic region is: ； in, Candidate dynamic acoustic regions; For spatial points The total joint probability field value after superimposing all adjacent events that are underexplained; Threshold for determining candidate dynamic acoustic regions; When calculating the dynamic correlation between underexplained events, for any two underexplained events... , Spatial proximity, temporal continuity, sound category similarity, and unexplained consistency are calculated sequentially and fused to obtain dynamic correlation. The output of the correlation threshold includes underexplained events that meet a preset correlation threshold, resulting in a set of underexplained acoustic event groups, denoted as: ； ； ； in, To explain the insufficient set of acoustic events; For the first The incident team explained the insufficient acoustic events; For the first The set of all acoustic events within the event group; For the first Synthetic sound probability vector within the event group; For the first The weighted sound source localization center of the event group; For the first The event group belongs to the BIM space unit; For the first The construction area of ​​the incident team; For the first Event group start time; For the first Total duration of the event group; For the first The average unexplained probability of the event group; For the first Joint probability contribution of the event group; To explain the insufficient acoustic event The unexplained probability; To explain the insufficient acoustic event Corrected sound category probability vector; The expression for the dynamic correlation degree is: ； ； ； ； ； ； in, To explain the insufficiency event , The dynamic correlation degree; Spatial proximity; For time continuity; For sound category similarity; Unexplained consistency; To explain the insufficiency event , The mean variance; , These are events that are not fully explained. , Spatial variance; for The weighted sound source localization center; , They are respectively , The moment the event occurred; This is the time decay parameter; To explain the insufficient acoustic event Corrected sound category probability vector; To explain the insufficient acoustic event The unexplained probability; When calculating the confidence score for dynamic node generation, for each group of underexplained acoustic events... The event quantity support, temporal duration support, spatial stability, sound category stability, and explanation insufficiency support are calculated and weighted to obtain the following expression: ； ； ； ； ； ； ； in, for Node confidence of the event group; , , , , These are the weighting coefficients for each indicator; express The number of events supporting the event group; The total number of events within the group; This is a parameter for normalizing the number of events; for The duration of support for the event group over time; for Total duration of the group; This represents the duration normalization parameter; for Spatial stability of event groups; for Dispersion of sound source locations within an event group; This represents the spatial dispersion normalization parameter; for Sound category stability of the event group; for The event group's comprehensive sound category probability vector is the first... The probability of a sound type; This represents the total number of all construction sound categories. for The explanation provided by the event group lacks sufficient support. For the first An underexplained event The unexplained probability; When satisfied At that time, dynamic acoustic topology nodes are generated. Set a node confidence threshold; otherwise, The group is labeled as the acoustic event group to be observed; When generating dynamic acoustic topology nodes, the type of dynamic acoustic topology node is determined based on sound category composition, spatial stability, duration, joint probability volume morphology, and BIM topology relationships. Specifically: When the event group is concentrated in location, the sound type is singular, and the duration is short, a temporary acoustic object is generated. When the event group is in a stable position and is close to an unformed component, a reserved installation location, or a future construction object, a virtual component is generated. When an event group lasts for a long time and contains multiple types of construction sounds, a temporary work unit is generated. When the location of the event group is unstable but the range of the joint probability volume is obvious, generate a sound source probability volume node; This results in dynamic node types including temporary acoustic objects, virtual components, temporary work units, and sound source probability volume nodes; The determination value for the dynamic node type is: ； ； ； ； in, for Temporary acoustic object scores for the event group; for The virtual component score of the event group; for The temporary task unit score for the event group; for The source probability volume node score of the event group; , , These are the weighting coefficients for each item in the score of the temporary acoustic object; , , These are the weighting coefficients for each item in the score of the virtual component; , , These are the weighting coefficients for each item in the score of the temporary work unit; , , These are the weight coefficients for each item in the score of the sound source probability volume node; express The strength of the proximity relationship between the event group and existing BIM components or reserved construction areas in the surrounding area; express The strength of the association between the event group and subsequent construction objects, unfinished components, or planned construction objects; express Average location uncertainty of the event group; Then, based on the explanation of the insufficient acoustic event group and its node types Generate dynamic acoustic topology nodes, represented as: ； ； in, for Dynamic acoustic topology nodes of the event group; for The dynamic acoustic topology node numbering of the event group; for The optimal node type for an event group; for The weighted fusion of event groups yields a probability vector of sound categories. for The weighted sound source localization center of the dynamic node within the event group is obtained by weighted averaging of the sound source localization centers of all acoustic events within the event group. for The BIM spatial unit to which the dynamic nodes within the event group belong; for The construction area to which dynamic nodes within an event group belong; for Joint probability contribution of the event group; for The start time of the event group; for Total duration of the event group; for The original interpretation of dynamic nodes is insufficient for the set of acoustic events; express Dynamic node status of event groups; The dynamic node status is determined based on node confidence and is categorized into valid dynamic nodes, candidate dynamic nodes, and failed dynamic nodes. When satisfied hour, If the lower confidence threshold is the threshold for a valid dynamic node, then it is a valid dynamic node, denoted as: ; When satisfied hour, If the lower confidence threshold is set to the threshold value for candidate dynamic nodes, then a node is considered a candidate dynamic node or a dynamic node awaiting review, denoted as: ; If an event group does not associate with new, insufficiently explained acoustic events for an extended period, or is subsequently reliably explained by existing BIM components, it is considered a failed dynamic node or an archived dynamic node, represented as... ; Output dynamic acoustic topology node set , represented as: ； in, The number of nodes in the dynamic acoustic topology; Finally, based on the aforementioned BIM spatial topology map, new dynamic acoustic topology nodes are generated to establish multiple topological associations between the topology nodes and the original nodes, resulting in an extended BIM spatial topology map. Specifically: If dynamic nodes Located in space unit If the elements are internal, then an inclusion relationship is established, which is represented as: ; If dynamic nodes Located in the construction area Within this area, a regional affiliation relationship is established, represented as follows: ; If dynamic nodes With existing BIM components Surface distance satisfies Then, a dynamic adjacency relationship is established, represented as: ,in, This represents the threshold for the adjacency distance between dynamic nodes and existing BIM components; If dynamic nodes With existing dynamic acoustic topology nodes The spatial ranges overlap, or the similarity of the sound categories of the two sounds meets the following criteria. Then, a dynamic node association relationship is established, represented as: ; If two dynamic nodes simultaneously satisfy the conditions of overlapping spatial ranges, similar sound categories, and temporal continuity, then they are marked as nodes to be merged, represented as: .

9. A dynamic monitoring method based on BIM spatial topology and construction sound matching according to claim 8, characterized in that... When establishing a matching record between construction acoustic events and dynamic acoustic topology nodes, Based on the reliable matching results, the construction acoustic events matched to existing BIM components are expressed as follows: ； in, Number the matched objects; Optimal candidate component serial number; To match the source of objects, they are categorized into existing BIM components. Dynamic topology nodes ; To match credibility scores; Optimal candidate component The dynamic interpretation of probability; Location reliability for sound sources; The uncertainty in sound source localization; Indicates the matching status flag. Indicates a reliable match; For events of valid dynamic acoustic topology nodes, if construction acoustic events Belongs to dynamic nodes For the set of associated events, a dynamic node matching record is established, represented as: ； in, Number the dynamic nodes; To explain the insufficiency event The unexplained probability; Node confidence; This indicates that the matching status is marked as dynamic node matching; For acoustic events with non-unique candidate results, the matching status is set to manual review, as shown below: ; For dynamic topology node association events where the dynamic node status is candidate dynamic node or the dynamic node confidence level is below a preset threshold, the matching status is set to dynamic pending review, as follows: ; This results in a matching record between construction acoustic events and dynamic acoustic topology nodes; Then, components marked as reliably matched are written back to the extended attributes and BIM space of the original BIM model objects, generating thermal markers for basic construction behavior. These markers represent the activity level of construction behavior of components, spatial units, or dynamic acoustic topology nodes within a specified time range; where the activity level value is expressed as: ； For existing reliable matching events of BIM components: ; For dynamic acoustic topology node association events: ; in, Represents BIM objects or spatial units In the time window The activity level of construction activities within the area; for The set of all construction acoustic events within the unit; For construction acoustics incident; It represents the maximum posterior probability; for The sound source intensity coefficient; for Duration of the acoustic event; Based on a comparison between the activity level of construction activities and a preset grading threshold, the thermal levels of construction activities are classified, as follows: ； in, Indicates the construction thermal rating; , , The preset grading threshold; When dynamically updating the initial construction acoustic association library, the association weights before the update are used. Feedback value and update step size Update the association weights of the construction acoustics association library to obtain the updated new association weights. The updated association weights are then normalized using the following expression: ； ； in, The new association weights after normalization constraints; , for the construction acoustics association library The feedback value for each associated record is set based on the matching status flag.

10. A dynamic monitoring system based on BIM spatial topology and construction sound matching, used to implement the dynamic monitoring method based on BIM spatial topology and construction sound matching as described in any one of claims 1-9, characterized in that, include: The BIM spatial topology generation unit is used to construct a BIM spatial topology model and an initial construction acoustic association library. It performs analysis and standardization processing on the BIM spatial topology model to discretize the surface of each BIM component into a set of triangular facets, establish spatial topological relationships, and generate a BIM spatial topology map. The construction acoustic event set generation unit is used to collect the original audio signals at the construction site, perform preprocessing, segmentation and construction sound recognition, and then merge them to form construction acoustic events. The unit also performs sound source localization and BIM coordinate system transformation on the construction acoustic events to obtain a construction acoustic event set. The BIM candidate component screening unit is used to determine the spatial retrieval range based on the construction acoustic event set and the BIM spatial topology map, and at the same time, to perform preliminary screening of candidate component types in conjunction with the initial construction acoustic association library to determine the candidate component types related to the current construction acoustic event. Within the spatial retrieval range, it calculates the minimum distance from the sound source point of the existing BIM entity component to the component surface, generates event-related topology tags, and obtains the existing BIM candidate component set. The component acoustic event matching unit is used to calculate the reliability of various types of evidence to obtain dynamic evidence weights based on the construction acoustic event set and the existing BIM candidate component set, from sound recognition, sound source localization, and candidate component competition to construct the event uncertainty state. Then, it calculates the dynamic explanatory ability of candidate components for construction acoustic events by weighted fusion of dynamic evidence, and combines the candidate competitive advantage adaptive matching threshold for comprehensive judgment, and outputs reliable matching results, candidate non-unique results, or events with insufficient global explanation. The BIM spatial topology map extension unit is used to calculate the joint probability contribution of events to any point in space based on the global under-explained events, so as to delineate candidate dynamic acoustic regions, calculate the dynamic correlation degree between under-explained events to form a set of under-explained acoustic event groups, calculate the dynamic node confidence degree, generate dynamic acoustic topology nodes under the condition that the node confidence degree meets the preset threshold, determine the dynamic acoustic topology node type, output the dynamic acoustic topology node set, establish multiple types of topology associations between topology nodes and original nodes, and update to obtain the extended BIM spatial topology map; The dynamic update unit is used to establish a matching record between construction acoustic events and dynamic acoustic topology nodes based on the existing set of BIM candidate components and the set of dynamic acoustic topology nodes, output the matching results and write them back to the original BIM model, classify the thermal level of construction behavior, and dynamically update the initial construction acoustic association library to realize dynamic intelligent supervision of the construction site.

Images