A digital twin modeling method and system for vibration and noise of large complexes

Abstract

This invention discloses a digital twin modeling method and system for vibration and noise in large-scale complexes, relating to the field of computer data processing technology. For the target complex, a sensor array is deployed and a twin is reconstructed to build a three-dimensional twin. The vibration and noise sources of the target complex are identified. Vibration modes are discretized and aggregated, and vibration equations for each aggregate class are mined. An explicit scene transfer framework is introduced to determine the explicit modeling requirements of the vibration and noise scene. Assisting the vibration equations, vibration and noise distribution based on acoustic medium boundaries is performed on the three-dimensional twin to generate a vibration and noise model. This invention addresses the technical problem in existing technologies where the level of detail and flexibility in vibration and noise modeling for complex architectural scenes is insufficient, leading to inadequate accuracy and adaptability in dynamic modeling. It achieves detailed vibration and noise modeling and dynamic visualization for complex architectural scenes, possessing significant advantages of high accuracy and high adaptability.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, specifically to a method and system for digital twin modeling of vibration and noise in large complexes. Background Art

[0002] Large-scale integrated complexes combine transportation, commerce, and residence, resulting in complex structures and densely packed functional areas, leading to increasingly prominent vibration and noise problems. This is especially true for integrated transportation hubs that combine stations and urban areas; due to the high density of people and numerous equipment, complex acoustic-vibration coupling phenomena can easily cause structural resonance, environmental disturbances, and other issues, affecting user safety and comfort.

[0003] Existing technologies often employ localized monitoring or simplified models for analysis, making it difficult to fully reflect the actual acoustic and vibration propagation characteristics. This is especially true in environments with multi-source heterogeneous interference, where it is challenging to effectively identify vibration and noise sources and quantify their impact. Furthermore, the lack of systematic modeling methods for the acoustic and vibration characteristics of complex building structures limits their application in acoustic diagnostics, structural optimization, and intelligent operation and maintenance.

[0004] Therefore, there is an urgent need for a modeling method that can integrate sensor data, structural information and physical laws to achieve refined and flexible modeling and evaluation of vibration and noise in large complexes. Summary of the Invention

[0005] This application provides a digital twin modeling method and system for vibration and noise of large complexes, which addresses the technical problem that the existing technology lacks the refinement and flexibility of vibration and noise modeling for complex building scenarios, resulting in insufficient accuracy and adaptability of dynamic modeling.

[0006] In view of the above problems, this application provides a digital twin modeling method and system for vibration and noise of large complexes.

[0007] In a first aspect, this application provides a digital twin modeling method for vibration and noise of a large complex. The method includes: deploying a sensor array and reconstructing a twin for the target complex to build a three-dimensional twin, wherein the sensor array and the modeling platform are connected by communication; determining the vibration and noise sources of the target complex, discretizing and aggregating the vibration and noise sources by vibration mode, and mining the vibration equations of each aggregation class, wherein the vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregation class; introducing an explicit scene transfer framework to determine the explicit modeling requirements of the vibration and noise scene, assisting the vibration equations, performing vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin to generate a vibration and noise model, and establishing a mapping between the vibration and noise model and the three-dimensional twin, wherein the vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

[0008] Secondly, this application provides a digital twin modeling system for vibration and noise of a large complex. The system includes: a first construction unit, used for deploying a sensor array and performing twin reconstruction on the target complex to construct a three-dimensional twin, wherein the sensor array and the modeling platform are connected by communication; an aggregation mining unit, used for determining the vibration and noise sources of the target complex, discretizing and aggregating the vibration and noise sources by vibration mode, and mining the vibration equations of each aggregation class, wherein the vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregation class; and a second construction unit, used for introducing an explicit scene transfer framework, determining the explicit modeling requirements of the vibration and noise scene, assisting the vibration equations, performing vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin, generating a vibration and noise model, and establishing a mapping between the vibration and noise model and the three-dimensional twin, wherein the vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application provides a digital twin modeling method for vibration and noise in large complexes. For the target complex, a sensor array is deployed and a twin is reconstructed to build a 3D twin. The vibration and noise sources of the target complex are identified. These sources are discretized and aggregated based on vibration patterns. Vibration equations for each aggregate class are extracted. An explicit scene transfer framework is introduced to determine the explicit modeling requirements of the vibration and noise scene. Assisting the vibration equations, vibration and noise distribution based on acoustic medium boundaries is applied to the 3D twin to generate a vibration and noise model. A mapping is established between the vibration and noise model and the 3D twin. The vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements. This method addresses the technical problem in existing technologies where the level of detail and flexibility in vibration and noise modeling for complex architectural scenes is insufficient, leading to inadequate accuracy and adaptability in dynamic modeling. It achieves detailed vibration and noise modeling and dynamic visualization for complex architectural scenes, exhibiting significant advantages of high accuracy and high adaptability. Attached Figure Description

[0011] Figure 1 This application provides a schematic flowchart of a digital twin modeling method for vibration and noise of a large complex.

[0012] Figure 2 This application provides a schematic diagram of the process for mining the vibration equations of each aggregate class in a digital twin modeling method for vibration and noise of a large complex.

[0013] Figure 3This application provides a schematic diagram of the structure of a digital twin modeling system for vibration and noise of a large complex.

[0014] Explanation of reference numerals in the attached figures: First building unit 11, aggregation mining unit 12, second building unit 13. Detailed Implementation

[0015] This application provides a digital twin modeling method and system for vibration and noise of large complexes to solve the technical problem that the existing technology lacks the level of refinement and flexibility in vibration and noise modeling for complex building scenarios, resulting in insufficient accuracy and adaptability of dynamic modeling.

[0016] Example 1: Figure 1 As shown, this application provides a digital twin modeling method for vibration and noise of large complexes, the method comprising:

[0017] S1: For the target complex, deploy sensor arrays and perform twin reconstruction to build a three-dimensional twin, wherein the sensor arrays and the modeling platform establish a communication connection.

[0018] In this embodiment, the target complex refers to a large building, transportation hub, activity center, etc., with multiple functional areas, complex structural form, and significant vibration and noise interference. To achieve vibration and noise modeling of this complex, a sensor array deployment is first required. That is, based on the building's structural layout, functional zoning, and acoustically sensitive areas, various types of sensor units are installed at typical locations. The sensor units may include accelerometers, sound pressure meters, micro-vibration probes, etc., and their function is to acquire multidimensional data related to vibration and noise.

[0019] Preferably, the sensor array is deployed in a distributed network manner, interconnecting with the backend modeling platform via wireless or wired communication protocols to ensure the continuity and real-time nature of data acquisition. Subsequently, combining the infrastructure data of the target complex, such as building structure data, design drawings, and sensor calibration information, a physical twin reconstruction of the target twin is performed. This involves using multi-source data fusion and employing BIM modeling tools, laser scanning, or image reconstruction to replicate and generate a three-dimensional visualization model reflecting the structural characteristics of the target complex, serving as the three-dimensional twin. The three-dimensional twin is a virtual-to-real mirror image of the target complex, its geometric shape and height are identical to the actual building, and it possesses structural identification capabilities, serving as a physical carrier for subsequent vibration and noise propagation modeling and situational evolution simulation.

[0020] Throughout the process, the sensor array establishes a communication connection with the modeling platform. For example, by deploying communication modules or IoT middleware, it can be ensured that the sensor-collected data can be transmitted to the modeling platform in real time to support model reconstruction and dynamic updates, thereby achieving closed-loop operation of the twin system.

[0021] S2: Identify the vibration and noise sources of the target complex, and discretize and aggregate the vibration and noise sources according to the vibration mode to mine the vibration equations of each aggregate class. The vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregate class.

[0022] In this embodiment of the application, based on the construction of a three-dimensional twin, the vibration and noise sources of the target complex are further determined in order to achieve modeling and simulation of the vibration and noise propagation law.

[0023] Specifically, the vibration and noise sources refer to physical equipment or functional areas that generate mechanical vibrations or sound pressure disturbances during the operation of the complex. Examples include: transportation hubs such as railways, activity areas, and large power generation equipment.

[0024] Subsequently, the vibration noise sources are discretely aggregated based on vibration modes. A vibration mode refers to the typical frequency structure, vibration directionality, and amplitude characteristics exhibited by the vibration noise source during operation. Discrete aggregation involves clustering all identified noise sources based on the similarity of their vibration modes, grouping noise sources with similar modal characteristics into the same cluster, thereby reducing modeling complexity and improving model reusability. For example, pipeline equipment with multiple frequency peaks concentrated in the 50Hz-70Hz range and exhibiting similar periods and spatial distributions can be grouped into the same cluster. Within the same cluster, the vibration noise sources remain spatially discrete, i.e., based on their assembly location within the target complex.

[0025] After completing the aggregation classification, the vibration equations of each aggregation are further explored. These vibration equations are mathematical expressions that describe the propagation and attenuation of acoustic signals in a spatial medium under specific vibration modes, reflecting the dynamic response characteristics under different medium conditions.

[0026] Specifically, a data statistical analysis and mining approach is employed. This involves combining the operational data of typical noise sources within the aggregate class to analyze their acoustic and vibration propagation patterns in various transmission media (such as concrete and air), extracting their diffusion paths, attenuation rates, and boundary reflection characteristics, and establishing corresponding model representations. The vibration equations, based on the propagation parameters of the acoustic medium, construct a relationship between vibration and noise in three-dimensional space and the gradual attenuation with distance and medium variations, serving as the foundation for subsequent vibration and noise scenario simulations and model evolution.

[0027] Furthermore, the vibration noise sources are discretely aggregated based on vibration modes. Step S2 of this application includes:

[0028] Set an encoding method; traverse the vibration noise sources, identify the source vibration patterns and perform same-pattern clustering to determine N clusters; according to the encoding method, encode and identify the N clusters, wherein vibration noise sources within the same cluster have the same encoding.

[0029] In this embodiment, to achieve structured management and unified model processing of vibration and noise sources, an encoding method is first set to identify and classify the subsequently identified vibration and noise source aggregations. The encoding method refers to a method based on predefined classification rules. For example, an encoding structure is set in conjunction with dimensions such as vibration mode characteristics, spatial location, and noise intensity level. This encoding structure may include fields such as a prefix indicating the aggregation type, an infix indicating vibration mode characteristics, and a suffix indicating geographical location information. A unified format such as VM-XX-YY is preferred, where VM represents the vibration mode, XX represents the mode number, and YY represents the region or equipment number, to ensure the uniqueness and resolvability of the encoding.

[0030] After setting the encoding method, the vibration noise sources are traversed, that is, all vibration noise sources identified in the previous stage are analyzed and processed one by one. Specifically, by calling up the historical sensing data of each vibration noise source, frequency domain analysis, modal feature extraction, and operational behavior discrimination are performed to identify the vibration mode of each source. For example, the vibration mode can be characterized by characteristic indicators such as dominant frequency, amplitude distribution, harmonic structure, and excitation directionality.

[0031] Subsequently, based on the identification results, homogeneous clustering is performed, that is, cluster analysis is conducted using modal similarity to group noise sources with similar vibration mode characteristics into the same category, determining a total of N clusters. Each cluster represents a set of vibration noise sources with similar vibration behavior and similar physical response mechanisms, which facilitates the commonalization of the model and equation fitting in subsequent steps.

[0032] Furthermore, according to the aforementioned coding method, the above N aggregate classes are coded and identified, that is, each aggregate class is assigned a unique code according to a predetermined coding rule to achieve unified indexing in the modeling system, database, and analysis process. For example, if the water pumps in multiple underground pump rooms are classified into the high-frequency vibration category, they can be uniformly identified as VM-01-PB, where PB represents the Pump Basement area, used for rapid retrieval and batch modeling. Through the above steps, the classification and coding management of vibration noise sources can be efficiently realized, supporting subsequent vibration equation mining and propagation simulation based on aggregate classes.

[0033] Furthermore, such as Figure 2 As shown, the vibration equations for each aggregate class are extracted. Step S2 of this application includes:

[0034] For the first cluster, a first vibration mode is determined; using the first vibration mode as an index, vibration records of the same type are called, and clustering is performed based on the acoustic medium to determine M groups of vibration records, where each group of vibration records corresponds to an acoustic medium; the M groups of vibration records are traversed to explore the spatial diffusion attenuation trend under dynamic vibration noise, and a linear transformation is performed and added to the first vibration equation.

[0035] In this embodiment, vibration equation mining is performed on a set of vibration noise sources that have been clustered and encoded. Specifically, a first vibration mode is determined for a first cluster. The first cluster refers to a representative set of vibration noise sources obtained after the aforementioned same-mode clustering process; and the first vibration mode is the common or dominant vibration behavior characteristic of the noise sources in the cluster, which can be composed of parameters such as dominant frequency components, harmonic structure, excitation direction, and time variation trend, and serves as the basic characteristic parameter set of the noise propagation behavior of this type.

[0036] Using the first vibration mode as an index, similar vibration records are retrieved. That is, the characteristic parameters corresponding to this mode are used as search conditions to filter vibration records that match its characteristics from the vibration database or historical operational data. The vibration records refer to vibration response data collected by the sensor array during building operation and stored according to time and location, including time-domain signals, frequency-domain spectra, and spatial positioning information. The above search operation can employ mechanisms such as fuzzy matching or threshold determination to ensure that the retrieved data has sufficient similarity and representativeness.

[0037] Furthermore, the vibration records of the same type in the first vibration mode are clustered based on the acoustic medium. That is, the vibration records obtained above are classified and aggregated according to the type of acoustic medium in their structural environment. The acoustic medium refers to the physical space medium through which vibration and noise propagate, such as reinforced concrete, glass curtain walls, and air. Its medium properties determine the propagation rate, attenuation degree, and reflection characteristics of the sound and vibration signals. By identifying the spatial location information and structural tags carried in the vibration records, a medium property mapping relationship is constructed, and finally, the vibration records of the same type are divided into M groups, each group corresponding to a type of acoustic medium, thereby providing a physical hierarchical basis for modeling the propagation law.

[0038] Furthermore, after grouping the media, the M groups of vibration records are traversed to uncover the spatial diffusion and attenuation trend of dynamic vibration noise. This involves analyzing multiple vibration records within each group to extract features such as signal amplitude changes, frequency attenuation, and propagation delay along different propagation paths, thus forming a diffusion and attenuation trend specific to the medium. This model characterizes the energy loss and spatial influence range of vibration noise as it propagates from its source point through a specific medium. Preferably, a continuous and predictable trend expression is formed by fitting attenuation curves and establishing distance-intensity functions.

[0039] Finally, the trend relationships mined from various acoustic media are integrated and added to the first vibration equation. Through the above steps, the constructed first vibration equation can comprehensively reflect the propagation characteristics of specific vibration modes in different media, laying the foundation for subsequent multi-source scenario modeling.

[0040] Similarly, the above steps are performed for each aggregate class until the Nth vibration equation corresponding to the Nth aggregate class is determined. The first vibration equation is then integrated and normalized up to the Nth vibration equation, which is taken as the vibration equation.

[0041] S3: Introduce an explicit scene transfer framework to determine the explicit modeling requirements of the vibration and noise scene, assist the vibration equation, perform vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin, generate a vibration and noise model, and establish a mapping between the vibration and noise model and the three-dimensional twin, wherein the vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

[0042] In this embodiment, to enhance the adaptability and specificity of vibration and noise modeling, an explicit scene transfer framework is further introduced to achieve dynamic adaptation and model optimization of noise propagation patterns under complex usage environments. The explicit scene transfer framework refers to a driving mechanism introduced into the modeling process to describe specific application scenarios, define modeling boundary conditions, and guide model selection and parameter configuration. This framework can transform actual operating scenarios into modelable problem structures through manual setting, rule matching, or knowledge graph reasoning, thereby enhancing the model's generalization ability and engineering adaptability in diverse scenarios.

[0043] Driven by the aforementioned framework, the explicit modeling requirements for the vibration and noise scenario are first determined. These explicit requirements refer to the clear definition of the specific modeling objectives and constraints for the acoustic and vibration problems in the current simulation scenario. For example, whether it is a single excitation source scenario, a multi-source superposition scenario, acoustic sensitivity analysis of a specific functional area, or propagation analysis under a specific medium path, etc. These requirements can be extracted from operational sensor data, user interaction settings, or historical case rules, and serve as the input prerequisites for subsequent model construction, guiding the invocation of vibration equations and the setting of spatial boundaries.

[0044] Furthermore, based on the clearly defined modeling requirements, the vibration equation is further assisted to perform vibration and noise distribution based on acoustic medium boundaries on the three-dimensional twin. Specifically, the three-dimensional twin is used as the modeling carrier. Combining the vibration and noise sources involved in the scene and their propagation paths, propagation boundary conditions are constructed according to preset acoustic medium parameters (including medium type, conduction coefficient, attenuation coefficient, reflection boundary, etc.). Based on this, the twin space is partitioned and calculated to simulate the spatial distribution characteristics of vibration and noise diffusion in the complex. Through the above calculation process, a vibration and noise model is generated. This model is used to describe the spatial distribution and attenuation characteristics of vibration and noise in the multi-medium structure under the target scene.

[0045] The vibration noise model is a reduced-order gray box model based on the medium frame and vibration noise elements. That is, under the three-dimensional twin frame, it only involves the spatial state of vibration noise elements. For example, vibration noise elements may include source intensity, spectral characteristics, periodicity, medium propagation parameters, etc.

[0046] Finally, a mapping is established between the vibration and noise model and the 3D twin. This involves associating and binding each physical element in the model with the spatial structure, component identifiers, and temporal state of the twin. This allows the vibration and noise model to be visualized and dynamically simulated on the 3D twin, enabling intuitive analysis and predictive evaluation of the noise situation within the target complex. This mapping also provides an indexing basis for subsequent model calls, scene migrations, and intelligent scheduling, forming one of the core components of the overall digital twin modeling system.

[0047] Furthermore, the explicit modeling requirement is based on a single-source scenario or a scenario with multiple distributed sources superimposed on each other, based on vibration and noise sources.

[0048] Furthermore, generating a vibration noise model, step S3 of this application includes:

[0049] Based on the vibration and noise scenario, the target vibration and noise source and its coverage area are determined, wherein the coverage area is located within the target complex; the target vibration and noise source is traversed, and the vibration equation is matched and associated; based on the coverage area, the three-dimensional twin is bounded to determine a scene twin based on the vibration and noise scenario; a vibration and noise model is constructed using the scene twin as a reference.

[0050] In this embodiment of the application, the explicit modeling requirement is based on a single-source scenario or a multi-source superposition scenario of vibration and noise sources. Specifically, it refers to determining the configuration type of vibration and noise sources to be focused on in the current modeling process based on the operating status and usage scenario of the target complex during the vibration and noise modeling process.

[0051] Among them, the single-source scenario refers to the existence of only a single dominant vibration and noise source in a certain period of time or a specific area. Its propagation path and influence range are relatively concentrated. The modeling focus is on accurately depicting the sound and vibration diffusion pattern of this single source. The multi-source superposition scenario indicates that there are multiple vibration and noise sources with different spatial locations and significant differences in intensity within the same modeling period. Their propagation process may involve mutual interference, overlap, and superposition effects. The modeling of this type of scenario needs to consider sound field coupling and multi-path propagation characteristics, which significantly increases the complexity.

[0052] Further, generating a vibration and noise model includes the following key steps: First, based on the vibration and noise scenario, determine the target vibration and noise source and its coverage area. The target vibration and noise source refers to the acoustic and vibration interference source identified as the main modeling object in the current scenario, which can be filtered through system scheduling information, sensor response intensity, or usage scenario strategies. The coverage area refers to the boundary of the area that may have an impact in the building space, starting from the target noise source, according to its acoustic and vibration propagation characteristics. It is usually estimated based on a preset sound pressure threshold or energy attenuation ratio, and its location range in three-dimensional space is determined. This range should be limited to within the building structure of the target complex.

[0053] Subsequently, the target vibration and noise sources are traversed, and the vibration equations are matched and associated. Specifically, each vibration and noise source involved in the current scene is processed one by one, its corresponding aggregation class code is identified, and the corresponding vibration equations established earlier are called accordingly, realizing rapid adaptation of physical laws and parameter reuse. This step ensures that the modeling of each noise source is based on its real physical characteristics, improving the accuracy and interpretability of the simulation results.

[0054] Simultaneously, based on the aforementioned coverage area, the 3D twin is bounded to determine a scene twin based on the vibration and noise scenario. Specifically, within the 3D twin, spatial substructures related to the current vibration and noise propagation are extracted to form a scene twin within a defined range for modeling and simulation. This scene twin possesses spatial structural boundaries, medium properties, and component identifiers, closely mirroring the actual scene and providing a computational platform for model execution.

[0055] Finally, based on the scene twin, a vibration and noise model is constructed. That is, with the defined spatial region, the matched vibration equation and the identified vibration and noise source as the core input, a simulation model that can describe the dynamic response characteristics of vibration and noise in the target scene is generated through physical propagation calculation and model parameter loading.

[0056] Furthermore, based on the aforementioned scene twin, a vibration and noise model is constructed. Step S3 of this application includes:

[0057] Receive distributed sensing data from the sensor array; for the scene twin, determine the sensing data of each target vibration and noise source based on the distributed sensing data, and perform diffusion attenuation deduction by combining the associated vibration equation to determine the vibration and noise model.

[0058] In this embodiment, to achieve dynamic modeling of the vibration and noise propagation pattern in a scene twin, distributed sensing data from a sensor array is first received. The sensors collect and upload the sampled data to the modeling platform in real time, forming continuous distributed sensing data. This data possesses time synchronization, spatial coverage, and type diversity, and can be used to reflect the real-time state of vibration and noise in different structural regions.

[0059] Furthermore, after acquiring the distributed sensing data, for the scene twin, the sensing data of each target vibration and noise source is determined based on the distributed sensing data. Specifically, the scene twin is a spatial substructure extracted from a 3D twin for the current acoustic and vibration modeling task. Based on this, by mapping the sensor deployment locations to the positions of the twin components, sensor data falling at different coverage locations within the scene twin are selected, completing a one-to-one correspondence between each target vibration and noise source and its sensing data. This process helps to clarify the actual sensing response characteristics dominated by each noise source, providing data support for subsequent physical modeling.

[0060] Furthermore, diffusion attenuation is extrapolated by combining the associated vibration equation. This vibration equation is typically expressed as a composite of spatial diffusion and temporal attenuation terms. By fusing this vibration equation with the identified source sensor data, and by setting boundary conditions (such as initial sound pressure level), initial parameters (such as dominant frequency and wave velocity), and medium property parameters (such as density and elastic modulus), the propagation process of the acoustic vibration signal in the twin structure can be extrapolated and calculated, predicting its amplitude attenuation and frequency variation trends at various spatial locations.

[0061] Finally, based on the above deduction results, a vibration noise model was determined. This vibration noise model is a modeling entity that quantifies and visualizes the propagation status of all major vibration and noise sources within the defined spatial area in the current scenario. It not only reflects the dynamic acoustic and vibration response state but also provides a basis for subsequent acoustic evaluation, risk diagnosis, and mitigation strategies.

[0062] Furthermore, diffusion attenuation is extrapolated. Step S3 of this application includes:

[0063] For the aforementioned scene twin, an acoustic medium architecture is determined. Based on the acoustic medium architecture, starting with sensing data based on the target vibration noise source, and using the diffusion equation iteration under acoustic medium changes as a condition, a diffusion attenuation trend is deduced to determine the vibration noise model. The vibration noise model is a reduced-order gray box model under the twin architecture that uses quantized particles to measure vibration noise for spatial coverage.

[0064] In this embodiment of the application, during the process of constructing a model for a specific vibration and noise scenario, the acoustic medium architecture must first be determined for the scenario twin. The scenario twin refers to the structural subset extracted from the complete 3D twin for this modeling task, possessing clear geometric boundaries, component information, and functional zoning. Based on this structure, the acoustic medium covering the space is identified, that is, the type of medium that the vibration and noise traverses during propagation, such as different physical materials.

[0065] Each type of medium, due to its different density, elastic modulus, damping characteristics, etc., has a significant impact on the propagation path and attenuation rate of acoustic and vibration signals. Therefore, it is necessary to construct a complete acoustic medium architecture, that is, based on the spatial distribution of components, labeling the medium type and boundary conditions corresponding to each spatial unit, thereby forming a spatial continuum of medium properties, providing a physical constraint basis for the subsequent diffusion of vibration and noise.

[0066] Furthermore, after determining the medium architecture, the propagation starts with the sensor data based on the target vibration and noise source. That is, the previously identified target vibration and noise source is selected, and its associated initial sensor response data is extracted as the propagation starting point. For example, this sensor data may include parameters such as source sound pressure level, dominant frequency, and excitation period, which constitute the input boundary conditions for simulation calculations and are used to drive the propagation behavior of acoustic and vibration signals in the medium structure.

[0067] Subsequently, the diffusion attenuation trend is deduced based on the iterative diffusion equation under the condition of acoustic medium change. Specifically, as the signal propagation path advances, the type of acoustic medium will continuously change, that is, switch from one material to another. At this time, the propagation parameters need to be dynamically adjusted according to the conduction characteristics of each medium to realize the dynamic deduction of the iterative diffusion of acoustic energy in the structure, and ensure the continuity and physical rationality of the propagation trajectory.

[0068] Finally, based on the simulation results, the vibration and noise model was determined. This model is a reduced-order gray-box model within a twin architecture, using a quantized distribution of vibration and noise for spatial coverage. Simultaneously, by introducing order reduction, the propagation dimensions and number of variables are simplified, improving the model's computational efficiency. The quantized particles in the model are discretized representations of acoustic and vibrational energy propagating in space. Each particle carries sound pressure intensity, frequency information, and propagation path attributes, enabling high-precision spatial coverage modeling within a three-dimensional twin structure. This model accurately characterizes the propagation pattern, attenuation patterns, and impact range of vibration and noise within the target complex, providing crucial support for acoustic assessment and mitigation strategies.

[0069] Furthermore, establishing the mapping between the vibration noise model and the three-dimensional twin, step S3 of this application includes:

[0070] A storage database is constructed; by introducing spatiotemporal codes, explicit modeling requirements based on vibration and noise scenarios are mapped and identified to vibration and noise models, and stored in the storage database; based on the retrieval information, retrieval and display are performed in the storage database and overlay display based on the three-dimensional twin.

[0071] Furthermore, the spatiotemporal code includes a timestamp code and a spatial location code located within the target complex.

[0072] In this embodiment of the application, in order to achieve unified management and efficient retrieval of vibration and noise models, a storage database is first constructed. The storage database refers to a database used to centrally store various vibration and noise models and their associated information.

[0073] Furthermore, by introducing spatiotemporal codes, explicit modeling requirements based on vibration and noise scenarios are mapped and identified to vibration and noise models, and stored in the storage database. Specifically, explicit modeling requirements refer to model visualization requirements corresponding to vibration and noise scenarios, such as the noise propagation process in a certain area, or comparing sound field changes under different equipment operation; while vibration and noise models are the mathematical expression of the sound and vibration propagation laws in this scenario.

[0074] Over time, vibration and noise models may be constructed based on different time points and different scenario requirements. To facilitate rapid subsequent retrieval, spatiotemporal codes are introduced as a mapping index to ensure that the model and its applicable scenario correspond one-to-one in the database and can be retrieved quickly.

[0075] Furthermore, the spatiotemporal code includes a timestamp code and a spatial location code within the target complex. The timestamp code identifies the temporal characteristics of the model's generation or application, supporting dynamic tracking and historical review of noise states across different time periods. The spatial location code is a spatial positioning identifier established based on the geometric structure and functional zoning of the target complex, representing the distribution of the scenario-based vibration and noise models. This can be expressed, for example, through floor numbers, component numbers, or area labels, to clearly define the architectural space range corresponding to the model. The combination of these two codes constitutes a unique identifier, enabling precise location and retrieval of any model.

[0076] When there is a need for scenario-based vibration and noise modeling, a vibration and noise model is constructed based on the above steps, identified based on spatiotemporal codes, and stored in the storage database.

[0077] Based on the above structure, upon receiving an external retrieval request, the system can retrieve information from the stored database and display it overlaid on the 3D twin. The retrieval information may include parameters such as the target time period, query area, and model type. The system quickly locates the target model based on the spatiotemporal code and automatically maps and loads the model into the 3D twin, achieving the overlay presentation and dynamic visualization of the vibration and noise model representing acoustic vibration in the virtual building environment, improving model application efficiency and user interaction experience. Through this mechanism, a complete data loop is established from model generation and storage management to scene retrieval and 3D twin-based overlay, providing strong information support for digital twin modeling of vibration and noise in large complexes.

[0078] This application provides a digital twin modeling method for vibration and noise of large complexes, which has the following technical advantages:

[0079] 1. Deploy a distributed sensor array within the target complex and reconstruct a highly accurate 3D digital twin by combining building structure information with sensor spatial location information. Extract modal features of vibration and noise sources, employ modal similarity clustering algorithms to achieve multi-source classification and aggregation, and establish a coding system for identification and management. This effectively simplifies the management dimensions of vibration and noise sources, improves the reusability of model parameters, and enhances the flexibility of system scheduling. Based on aggregation classes and acoustic medium types, mine and construct diffusion attenuation vibration equations under different propagation paths, supporting dynamic parameter configuration. This accurately reflects the propagation behavior of vibration and noise in heterogeneous structural media, improving the physical reliability of simulation. Addressing the pain points of multi-source coupling of vibration under direct sensing modeling, it can be flexibly configured and constructed according to vibration and noise modeling requirements.

[0080] 2. Establish explicit scene migration logic for identifying modeling requirements, guiding the invocation of vibration equations and local mapping modeling of the 3D twin. This enables rapid adaptation of vibration and noise models to different scenarios, improving the versatility and response efficiency of model construction. By integrating real-time sensor data with pre-built vibration equations, dynamic simulation of acoustic and vibration propagation patterns is performed within the scene twin. This allows for dynamic simulation of acoustic and vibration responses under current building operating conditions, enhancing the real-time performance and accuracy of monitoring and diagnosis.

[0081] 3. Construct a diffusion iteration model based on acoustic medium changes, introduce quantized particles to represent propagation states, and build a reduced-order gray-box model to reduce model dimensionality and computational load while retaining key physical features, achieving efficient modeling and rapid response. Construct a spatiotemporal code system with timestamps and spatial locations as dimensions to uniformly manage the mapping relationship between modeling requirements and model entities, and store and retrieve them in the database. Support accurate model invocation and overlay display across multiple scenarios, time periods, and regions, improving system maintainability and intelligence.

[0082] In summary, it achieves refined modeling and dynamic visualization of vibration and noise in complex building scenarios, with significant advantages of high accuracy and high adaptability.

[0083] Example 2: Based on the same inventive concept as the vibration and noise digital twin modeling method for a large complex in the foregoing examples, such as... Figure 3 As shown, this application provides a digital twin modeling system for vibration and noise of a large complex, the system comprising:

[0084] The first building unit 11 is used to deploy sensor arrays and perform twin reconstruction on the target complex to build a three-dimensional twin, wherein the sensor array and the modeling platform have a communication connection.

[0085] The aggregation mining unit 12 is used to determine the vibration and noise sources of the target complex, and to discretize and aggregate the vibration and noise sources according to the vibration mode, and to mine the vibration equations of each aggregation class, wherein the vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregation class.

[0086] The second building unit 13 is used to introduce an explicit scene transfer framework, determine the explicit modeling requirements of the vibration and noise scene, assist the vibration equation, perform vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin, generate a vibration and noise model, and establish a mapping between the vibration and noise model and the three-dimensional twin, wherein the vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

[0087] Furthermore, the aggregation mining unit 12 is also used to perform the following steps: setting an encoding method; traversing the vibration noise sources, identifying the source vibration patterns and performing same-pattern clustering processing to determine N aggregation classes; and encoding and identifying the N aggregation classes according to the encoding method, wherein vibration noise sources within the same aggregation class have the same encoding.

[0088] Furthermore, the aggregation mining unit 12 is also used to perform the following steps: for the first aggregation class, determine the first vibration mode; using the first vibration mode as an index, call the vibration records of the same type, perform clustering based on the acoustic medium, determine M groups of vibration records, wherein each group of vibration records corresponds to an acoustic medium; traverse the M groups of vibration records, mine the spatial diffusion attenuation trend under dynamic vibration noise, perform linear transformation and add it to the first vibration equation.

[0089] Furthermore, the explicit modeling requirement is based on a single-source scenario or a scenario with multiple distributed sources superimposed on each other, based on vibration and noise sources.

[0090] Furthermore, the second construction unit 13 is also used to perform the following steps: determining the target vibration noise source and coverage area according to the vibration noise scene, wherein the coverage area is located in the target complex; traversing the target vibration noise source, matching and associating the vibration equation, defining the three-dimensional twin according to the coverage area, and determining the scene twin based on the vibration noise scene; and constructing a vibration noise model based on the scene twin.

[0091] Furthermore, the second building unit 13 is also used to perform the following steps: receiving distributed sensing data from the sensing array; for the scene twin, determining the sensing data of each target vibration noise source based on the distributed sensing data, and performing diffusion attenuation deduction in combination with the associated vibration equation to determine the vibration noise model.

[0092] Furthermore, the second construction unit 13 is also used to perform the following steps: for the scene twin, determine the acoustic medium architecture; based on the acoustic medium architecture, starting with the sensing data based on the target vibration noise source, and taking the diffusion equation iteration based on the acoustic medium change as a condition, perform diffusion attenuation situation deduction to determine the vibration noise model, wherein the vibration noise model is a reduced-order gray box model under the twin architecture that uses quantized particles that measure vibration noise for spatial coverage.

[0093] Furthermore, the second construction unit 13 is also used to perform the following steps: constructing a storage database; mapping and identifying the explicit requirements for modeling based on vibration and noise scenarios and vibration and noise models by introducing spatiotemporal codes, and storing them in the storage database; and retrieving and displaying the data in the storage database based on the retrieval information and the overlay display based on the three-dimensional twin.

[0094] Furthermore, the spatiotemporal code includes a timestamp code and a spatial location code located within the target complex.

[0095] Through the foregoing detailed description of a vibration and noise digital twin modeling method for a large complex, those skilled in the art can clearly understand the vibration and noise digital twin modeling method and system for a large complex in this embodiment. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to the method section.

[0096] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A digital twin modeling method for vibration and noise of a large complex, characterized in that, The method includes: For the target complex, sensor arrays are deployed and twin reconstruction is carried out to construct a three-dimensional twin, wherein the sensor arrays and the modeling platform are connected by communication. The vibration and noise sources of the target complex are identified, and the vibration and noise sources are discretized and aggregated according to the vibration mode. The vibration equations of each aggregate are extracted, wherein the vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregate. An explicit scene transfer framework is introduced to determine the explicit modeling requirements of the vibration and noise scene. This framework assists the vibration equation in performing vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin, generating a vibration and noise model. A mapping between the vibration and noise model and the three-dimensional twin is then established. The vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

2. The method for digital twin modeling of vibration and noise in a large complex as described in claim 1, characterized in that, Discrete aggregation of the vibration noise sources based on vibration modes includes: Set the encoding method; Traverse the vibration noise sources, identify the source vibration modes and perform same-mode clustering to determine N clusters; According to the encoding method, the N aggregate classes are encoded and identified, wherein vibration noise sources within the same aggregate class have the same encoding.

3. The method for digital twin modeling of vibration and noise in a large complex as described in claim 2, characterized in that, The vibration equations for each type of aggregate were extracted, including: For the first aggregation class, determine the first vibration mode; Using the first vibration mode as an index, similar vibration records are called up, and clustering is performed based on the acoustic medium to determine M groups of vibration records, where each group of vibration records corresponds to an acoustic medium. By traversing the M sets of vibration records, the spatial diffusion attenuation trend under dynamic vibration noise is explored, linear transformation is performed, and the results are added to the first vibration equation.

4. The method for digital twin modeling of vibration and noise in a large complex as described in claim 1, characterized in that, The explicit modeling requirement is based on single-source scenarios or multiple-source superposition scenarios of vibration and noise sources.

5. The method for digital twin modeling of vibration and noise in a large complex as described in claim 4, characterized in that, Generate a vibration noise model, including: Based on the vibration and noise scenario, the target vibration and noise source and its coverage area are determined, wherein the coverage area is located within the target complex; The target vibration and noise sources are traversed, the vibration equations are matched and associated, and the three-dimensional twin is bounded according to the coverage area to determine the scene twin based on the vibration and noise scene. A vibration and noise model is constructed based on the scene twin.

6. The method for digital twin modeling of vibration and noise in a large complex as described in claim 5, characterized in that, Based on the scene twin, a vibration noise model is constructed, including: Receive distributed sensing data from the sensor array; For the scene twin, the sensing data of each target vibration and noise source is determined based on the distributed sensing data, and the diffusion attenuation is deduced by combining the associated vibration equation to determine the vibration and noise model.

7. The method for digital twin modeling of vibration and noise in a large complex as described in claim 6, characterized in that, Perform diffusion attenuation simulation, including: For the aforementioned scenario twin, determine the acoustic medium architecture; Based on the acoustic medium architecture, starting with sensing data based on the target vibration noise source, and taking the diffusion attenuation situation as a condition based on the diffusion equation iteration under acoustic medium change, the vibration noise model is determined. The vibration noise model is a reduced-order gray box model under the twin architecture that uses quantized particles that measure vibration noise for spatial coverage.

8. The method for digital twin modeling of vibration and noise in a large complex as described in claim 1, characterized in that, Establishing the mapping between the vibration noise model and the three-dimensional twin includes: Build and store the database; By introducing spatiotemporal codes, explicit modeling requirements based on vibration and noise scenarios are mapped and identified to vibration and noise models, and stored in the storage database. Based on the search information, the data is retrieved from the stored database and displayed overlaid on the three-dimensional twin.

9. The method for digital twin modeling of vibration and noise in a large complex as described in claim 8, characterized in that, The spatiotemporal code includes a timestamp code and a spatial location code located within the target complex.

10. A digital twin modeling system for vibration and noise of a large complex, characterized in that, The system implementing the vibration and noise digital twin modeling method for a large complex as described in any one of claims 1-9 includes: The first building unit is used to deploy sensor arrays and perform twin reconstruction on the target complex to build a three-dimensional twin, wherein the sensor array and the modeling platform are connected by communication. The aggregation mining unit is used to determine the vibration and noise sources of the target complex, and to discretize and aggregate the vibration and noise sources according to the vibration mode, and to mine the vibration equations of each aggregation class. The vibration equations are the vibration and noise diffusion and attenuation relationship based on the acoustic medium under the vibration characteristics of each aggregation class. The second building unit is used to introduce an explicit scene transfer framework, determine the explicit modeling requirements of the vibration and noise scene, assist the vibration equation, perform vibration and noise distribution based on the acoustic medium boundary on the three-dimensional twin, generate a vibration and noise model, and establish a mapping between the vibration and noise model and the three-dimensional twin. The vibration and noise model is a reduced-order gray box model based on the medium framework and vibration and noise elements.

Images