A method and system for selecting noise monitoring points based on a noise map

Abstract

The present invention discloses a method and system for selecting noise monitoring point locations based on a noise map, which relates to the technical field of environmental noise monitoring. It includes constructing a three-dimensional topological network based on a three-dimensional noise intensity matrix, using a quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combining quantum annealing to optimize the weights of nodes in the three-dimensional topological network to generate an acoustic topological feature map. According to the acoustic topological feature map, multi-scale noise propagation features are extracted through a convolutional neural network, and combined with a monitoring device parameter library and coverage constraint conditions, a grid-based spatial demand matrix is constructed. According to the grid-based spatial demand matrix, an optimal noise monitoring network location selection scheme is generated. The present invention realizes the intelligent optimization and dynamic adjustment of the sound propagation path by combining the quantum-inspired topological optimization algorithm with quantum annealing to optimize the weights of nodes in the three-dimensional topological network.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental noise monitoring, and particularly to a method and system for selecting noise monitoring points based on a noise map. Background Art

[0002] The technology of selecting noise monitoring points plays an important role in urban environmental noise control. At present, traditional methods for selecting noise monitoring points are mainly based on empirical rules or simplified acoustic models, such as using equally spaced grid points or static site selection strategies based on two-dimensional noise maps. In recent years, with the development of BIM technology and GIS, some studies have begun to attempt to combine building geometric features and sound propagation characteristics for three-dimensional noise modeling and use heuristic algorithms to optimize the layout of monitoring points. However, existing methods still have obvious deficiencies in aspects such as dynamic optimization of sound propagation paths, extraction of multi-scale noise features, and accurate matching of monitoring equipment and sound field characteristics, resulting in problems such as uneven coverage of the monitoring network and low resource utilization.

[0003] The main defects of the prior art are as follows: on the one hand, traditional site selection methods are difficult to accurately reflect the complex influence of the building environment on sound propagation, especially the modeling accuracy of acoustic phenomena such as reflection and diffraction is insufficient; on the other hand, most optimization algorithms only consider noise features at a single scale and lack collaborative analysis of near-field direct sound, mid-field reflected waves, and far-field attenuation gradients, resulting in the inability of monitoring points to comprehensively capture the spatial heterogeneity of noise propagation. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for selecting noise monitoring points based on a noise map to solve the problems of insufficient modeling accuracy of sound propagation in the building environment and incomplete capture of noise spatial heterogeneity in the prior art.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for selecting noise monitoring points based on a noise map, which includes obtaining sound source monitoring data and a building BIM model of a target area, performing three-dimensional grid division on the spatial range of the target area to generate a three-dimensional noise intensity matrix; based on the three-dimensional noise intensity matrix, constructing a three-dimensional topological network, using a quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combining quantum annealing to optimize the weights of nodes in the three-dimensional topological network to generate an acoustic topological feature map; according to the acoustic topological feature map, extracting multi-scale noise propagation features through a convolutional neural network, and combining a monitoring device parameter library and coverage constraint conditions to construct a grid-based spatial demand matrix, and generating an optimal noise monitoring network location selection scheme according to the grid-based spatial demand matrix; based on the optimal noise monitoring network location selection scheme, performing feasibility evaluation through coverage verification, and dynamically adjusting the optimal noise monitoring network location selection scheme according to the evaluation results.

[0008] As a preferred embodiment of the method for selecting noise monitoring points based on a noise map according to the present invention, wherein: the process of obtaining the building BIM model is to generate a triangular mesh model through Poisson reconstruction;

[0009] Using the NURBS algorithm to fit the outer surface of the building to generate a parametric surface model, and through the IFC geometry parsing engine, identifying the geometric features of the building structure to generate a preliminary building BIM model;

[0010] Inputting the acoustic parameters in the associated material library into the preliminary building BIM model to obtain the building BIM model.

[0011] As a preferred embodiment of the method for selecting noise monitoring points based on a noise map according to the present invention, wherein: the steps of generating the three-dimensional noise intensity matrix are as follows:

[0012] According to the three-dimensional computational domain boundary parameters, establish a Cartesian grid, and adjust the offset of the grid lines in the Cartesian grid from the outer surface of the building BIM model through ray collision detection to generate a building-conforming grid;

[0013] According to the building-conforming grid and the sound source monitoring data, associate the point sound source with the nearest grid center point through the nearest neighbor interpolation algorithm, and establish the global sound pressure level distribution through spatial interpolation to generate a three-dimensional noise intensity matrix.

[0014] As a preferred embodiment of the method for selecting noise monitoring points based on a noise map according to the present invention, wherein: the steps of constructing the three-dimensional topological network are as follows:

[0015] Read the grid cell coordinates and sound pressure level data in the noise intensity matrix through a COO format parser, and generate a three-dimensional grid point set with sound pressure level annotations;

[0016] According to the three-dimensional grid point set, each node in the three-dimensional grid point set is connected by the 26-neighborhood rule to form an initial topological connection relation graph;

[0017] By using the Dijkstra shortest path algorithm, calculate the sound pressure level gradient weights of adjacent nodes in the initial topological connection relation graph, and mark the building obstacle edges as impassable to generate a three-dimensional topological network.

[0018] As a preferred scheme of the noise monitoring point location selection method based on the noise map according to the present invention, wherein: using the quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combining quantum annealing to optimize the weights of nodes in the three-dimensional topological network to generate an acoustic topological feature map, the steps are as follows:

[0019] Calculate the sound pressure level gradient weights of adjacent nodes in the initial topological connection relation graph;

[0020] Map the nodes in the topological network to qubits according to the deterministic mapping rule to form a qubit string;

[0021] Initialize all qubits to a uniform superposition state through Hadamard gate transformation and calculate the optimal sound propagation path;

[0022] Iteratively optimize the weights of nodes in the three-dimensional topological network through the quantum annealing algorithm, and combine with the optimal sound propagation path to generate an acoustic topological feature map.

[0023] As a preferred scheme of the noise monitoring point location selection method based on the noise map according to the present invention, wherein: the extraction of multi-scale noise propagation characteristics by the convolutional neural network means extracting the near-field characteristics of the sound source through small-scale convolutional kernels, extracting the reflected wave characteristics through medium-scale dilated convolutions, and extracting the acoustic diffraction attenuation characteristics of building edges through separable convolutional kernels.

[0024] As a preferred scheme of the noise monitoring point location selection method based on the noise map according to the present invention, wherein: combining the monitoring device parameter library and the coverage constraint conditions to construct a grid-based spatial demand matrix, the steps are as follows:

[0025] Identify the spatial distribution characteristics of multi-scale noise propagation characteristics through the multi-scale sound field decomposition algorithm;

[0026] Divide the monitoring devices into high-frequency, medium-frequency, and low-frequency devices through frequency band division, and define the monitoring device parameter library according to the monitoring devices in different frequency bands;

[0027] Convert the multi-scale noise propagation characteristics into grid-based coverage indicators through the spatial grid discretization algorithm;

[0028] Generate the coverage constraint conditions through Kriging interpolation according to the grid-based coverage indicators;

[0029] Based on the coverage constraint conditions, the candidate set of device deployment is transformed into a grid-based spatial demand matrix through ArcGIS.

[0030] As a preferred solution of the method for selecting noise monitoring point locations based on a noise map according to the present invention, wherein: the generation of the optimal noise monitoring network location selection scheme refers to calculating the spatial layout score of the monitoring points through a non-linear coverage contribution algorithm;

[0031] Based on the spatial layout score of the monitoring points, the genetic algorithm is used to iteratively optimize the spatial layout set of the monitoring points to generate the optimal noise monitoring network location selection scheme.

[0032] As a preferred solution of the method for selecting noise monitoring point locations based on a noise map according to the present invention, wherein: based on the optimal noise monitoring network location selection scheme, feasibility assessment is carried out through coverage verification, and the optimal noise monitoring network location selection scheme is dynamically adjusted according to the assessment results. The steps are as follows:

[0033] Based on the coverage ability data in the device parameter library, the GIS spatial analysis method is used to analyze the coverage of each monitoring device on the grid cells, and the coverage score of the optimal noise monitoring network location selection scheme is calculated through a weighted penalty function;

[0034] Feasibility analysis is carried out according to the coverage score of the optimal noise monitoring network location selection scheme, and the particle swarm optimization algorithm is used to iteratively optimize the spatial layout set of the monitoring points.

[0035] In a second aspect, the present invention provides a system for selecting noise monitoring point locations based on a noise map, including a matrix generation module, a spectrum generation module, a location selection scheme generation module, and an evaluation module;

[0036] The matrix generation module is used to obtain the sound source monitoring data and the building BIM model of the target area, and perform three-dimensional grid division on the spatial range of the target area to generate a three-dimensional noise intensity matrix;

[0037] The spectrum generation module is used to construct a three-dimensional topological network based on the three-dimensional noise intensity matrix, calculate the optimal sound propagation path using a quantum-inspired topological optimization algorithm, and combine quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature spectrum;

[0038] The location selection scheme generation module is used to extract multi-scale noise propagation features through a convolutional neural network according to the acoustic topological feature spectrum, and combine the monitoring device parameter library and the coverage constraint conditions to construct a grid-based spatial demand matrix, and generate the optimal noise monitoring network location selection scheme according to the grid-based spatial demand matrix;

[0039] An evaluation module is used to perform a feasibility evaluation through coverage verification based on the optimal noise monitoring network location selection scheme, and dynamically adjust the optimal noise monitoring network location selection scheme according to the evaluation results.

[0040] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for selecting noise monitoring points based on a noise map as described in the first aspect of the present invention is implemented.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for selecting noise monitoring points based on a noise map as described in the first aspect of the present invention is implemented.

[0042] The beneficial effects of the present invention are as follows: By combining the quantum-inspired topology optimization algorithm with quantum annealing to optimize the weights of the three-dimensional topology network nodes, the intelligent optimization and dynamic adjustment of the sound propagation path are realized, effectively capturing the complex influence of the building environment on the reflection and diffraction of sound waves, so that the generated acoustic topology feature map can accurately characterize the sound field propagation characteristics; at the same time, through the multi-scale feature extraction technology of the convolutional neural network, small-scale convolutional kernels, medium-scale dilated convolutions, and separable convolutional kernels are respectively used to collaboratively analyze the characteristics of near-field direct sound, reflected waves, and diffraction attenuation, realizing the refined analysis of the spatial heterogeneity of noise propagation. Description of the Drawings

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is a flowchart of the method for selecting noise monitoring points based on a noise map in Embodiment 1.

[0045] Figure 2 It is a schematic diagram of the noise monitoring point location selection system based on a noise map in Embodiment 1.

[0046] Figure 3 It is a schematic diagram of the process of constructing a three-dimensional topology network in Embodiment 1.

[0047] Figure 4 It is a flowchart of the execution of the quantum-inspired optimization and annealing algorithm in Embodiment 1. Detailed Embodiments

[0048] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification.

[0049] In the following description, many specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0050] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0051] Example 1, referring to Figures 1 to 4 , this example provides a method for selecting noise monitoring points based on a noise map, including the following steps:

[0052] Step 1: Obtain the sound source monitoring data and the building BIM model of the target area, and perform a three-dimensional grid division on the spatial range of the target area to generate a three-dimensional noise intensity matrix;

[0053] The sound source monitoring data includes sound pressure level time history records, spectral characteristics, sound source spatial positions, and environmental parameters.

[0054] It should be noted that the sound pressure level time history records are collected by a first-class sound level meter that complies with the IEC 61672-1 standard, and the time series waveform data is recorded in the A / C / Z weighting mode; the spectral characteristics are obtained by a real-time spectrum analyzer, and the frequency domain energy distribution is calculated based on the 1 / 3 octave or narrowband Fourier transform; the sound source spatial position is measured by a multi-channel microphone array, and the three-dimensional coordinates and propagation direction angles are determined by combining the beamforming algorithm and the time difference positioning technology; the environmental parameters are synchronously collected by integrating temperature and humidity sensors and an anemometer, and the background noise data is measured and recorded by a sound level meter in a state where the sound source is silent. All data are marked with metadata identifiers such as sensor models, calibration dates, and sampling configurations.

[0055] Obtain the high-density point cloud data of the target building through a ground three-dimensional laser scanner. Based on the high-density point cloud data, use a statistical outlier removal algorithm to remove noise points, and generate a triangular mesh model through Poisson reconstruction;

[0056] Furthermore, high-density point cloud data of the target building is obtained through the scanning resolution of the terrestrial three-dimensional laser scanner and the layout scheme of scanning stations. After the high-density point cloud data is imported into the point cloud processing software, multi-station point cloud registration is first performed, and the iterative closest point algorithm is used to achieve millimeter-level accuracy alignment. For the registered high-density point cloud data, a statistical outlier removal algorithm based on k-nearest neighbors is adopted, and the neighborhood radius and standard deviation multiple threshold are set to identify and remove discrete noise points and flying points. The denoised high-density point cloud data is processed by the Poisson surface reconstruction algorithm, and the octree depth and sampling weight parameters are set to generate a triangular mesh model with a continuous topological structure and retaining the geometric features of the building. The vertex normal vector of the triangular mesh model is obtained by local plane fitting calculation.

[0057] Based on the triangular mesh model, the NURBS algorithm is used to fit the building outer surface to generate a parametric surface model, and through the IFC geometry parsing engine, the geometric features of the building structure are identified to generate a preliminary building BIM model (Building Information Model);

[0058] Furthermore, based on the triangular mesh model, first, the NURBS algorithm is used for surface fitting. By setting the number of control points (not less than 16 per square meter) and the node vector parameters in the triangular mesh model, the discrete triangular patches are converted into continuous NURBS surface patches. During the fitting process, the least squares method is used to optimize the geometric deviation between the surface and the original mesh (controlled within ±5 mm). After the generated parametric surface model is imported into the BIM software environment, the IFC geometry parsing engine automatically identifies the geometric features in the NURBS surface: for the plane area (curvature < 0.001 / m), it is directly converted into an IfcWallStandardCase entity; for the doubly-curved surface (such as a dome), it is converted into an IfcCurtainWall and the NURBS parametric definition is retained; for the intersection area of structural members, an IfcConnectionGeometry connection relationship is generated through spatial boolean operations. All identified geometric features are assigned entity type attributes (such as IfcWall / IfcSlab) according to the IFC4.3 standard, and at the same time, the material partition information of the original triangular mesh model is inherited. The finally generated preliminary building BIM model contains a complete building component classification tree (including geometric details at the LOD350 level) and an attribute set conforming to the ISO 16739 specification.

[0059] Through the transfer matrix method, the acoustic parameters in the associated material library are obtained and input into the preliminary building BIM model to obtain the building BIM model.

[0060] Furthermore, the transfer matrix method was used to calculate the propagation characteristics of sound waves in multilayered media. Based on the sound pressure level decay curve measured in a reverberation chamber and the normal-incidence sound absorption coefficient measured with an impedance tube, the material's acoustic parameters, such as dynamic resistivity, porosity, and tortuosity, were derived. After matching and verifying these acoustic parameters with standard material records in the material library, they were input into the preliminary building BIM model as a property set. Specifically, the acoustic parameter attributes were extended within the IfcMaterial resource definition, the acoustic transmission characteristics of the multilayer structure were configured within the IfcMaterialLayerSetUsage, and the acoustic parameters were bound to the building components in the BIM model via the IfcRelAssociatesMaterial relationship entity. The resulting building BIM model includes complete acoustic material property definitions. Each component's acoustic parameters are annotated with a measurement error range (e.g., ±5%), conforming to the property extension specifications of the ISO 16739-1 standard.

[0061] It should be noted that the acoustic parameters in the associated material library refer to the inherent properties such as sound absorption coefficient and sound insulation value assigned to building components (such as walls and glass) in the building BIM model.

[0062] Based on the sound source monitoring data of the building BIM model, the spatial scope is defined and the boundary parameters of the three-dimensional calculation domain are generated;

[0063] Furthermore, based on the geometric data of the exterior wall components of the building BIM model and the coordinate data of the monitoring points in the sound source monitoring data, the following steps are performed to generate the three-dimensional calculation domain boundary parameters: first, the vertex coordinates of all exterior wall components in the building BIM model are extracted to calculate the building outer contour polygon; then, combined with the spatial position coordinates of each monitoring point in the sound source monitoring data, the horizontal boundary range is expanded outward by 50 meters based on the building outer contour polygon; in the vertical direction, the vertical boundary range is taken from the bottom elevation of the building BIM model to the highest elevation plus 20 meters; finally, the three-dimensional calculation domain boundary parameters including the minimum and maximum values of the X-axis, the minimum and maximum values of the Y-axis, and the minimum and maximum values of the Z-axis are output.

[0064] According to the boundary parameters of the 3D computational domain, a Cartesian grid is established. Through ray collision detection, the offset between the grid lines in the Cartesian grid and the outer surface of the building BIM model is adjusted to generate a building-fitting grid.

[0065] Furthermore, based on the maximum and minimum values of the X / Y / Z axes in the boundary parameters of the three-dimensional computational domain, an initial Cartesian grid is established, and the grid line spacing is evenly distributed according to the preset resolution. The outer surface geometric data of the building BIM model is converted into a set of triangular patches, and ray collision detection is performed on each Cartesian grid line: detection rays parallel to the coordinate axes are emitted from the grid line endpoints, the intersection points with the triangular patches of the building BIM model are calculated, and the grid line endpoints are adjusted to a position 0.3 meters away from the nearest intersection point. Repeat the above process until all grid lines are detected, and finally generate a building-conforming grid that maintains a 0.3-meter offset distance from the outer surface of the building BIM model. The grid lines are automatically disconnected in the internal area of the building BIM model to form cavities. The bounding box acceleration algorithm is used to optimize the collision detection efficiency during the processing to ensure that the building-conforming grid strictly follows the range constraints of the three-dimensional computational domain boundary parameters.

[0066] Based on the building-conforming grid and the sound source monitoring data, the point sound sources are associated with the nearest grid center points through the nearest neighbor interpolation algorithm, and the global sound pressure level distribution is established through spatial interpolation methods to generate a three-dimensional noise intensity matrix.

[0067] Furthermore, a spatial index structure of the center points of each unit of the building-conforming grid is established, and the nearest neighbor interpolation algorithm is used to accelerate the nearest neighbor search; the coordinates of the point sound sources in the sound source monitoring data are matched with the center points of the grid units, and the nearest grid unit corresponding to each point sound source is determined through Euclidean distance calculation, and the sound pressure level values are recorded. After the sound source matching is completed, the grid units not directly associated with the sound sources are processed based on the Kriging spatial interpolation method: a 50-meter range of variation and a nugget value parameter of 0.2 are set; the weight coefficients of each monitoring point are obtained by solving the Kriging equations, and the estimated sound pressure level values of the unsampled grid units are calculated. Finally, a three-dimensional noise intensity matrix containing the sound pressure level data of all grid units is output. Each element in the matrix corresponds to the A-weighted sound pressure level value at a specific position of the building-conforming grid, and the data storage uses the COO sparse format to record non-zero elements. The processing process strictly follows the original measurement values of the sound source monitoring data and the geometric structure of the building-conforming grid to ensure that the spatial distribution characteristics of the three-dimensional noise intensity matrix are consistent with the actual situation.

[0068] Step 2: Based on the three-dimensional noise intensity matrix, construct a three-dimensional topological network, use the quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combine quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature map;

[0069] Read the grid unit coordinates and sound pressure level data in the noise intensity matrix through the COO format parser, and generate a three-dimensional grid point set with sound pressure level annotations;

[0070] Further, the COO format parser reads the non-zero elements of the noise intensity matrix line by line, extracts the row index of each non-zero element as the horizontal position coordinate of the grid cell, the column index as the vertical position coordinate of the grid cell, and the numerical value as the sound pressure level data; combines the horizontal position coordinate, the vertical position coordinate with a preset height reference value to form a spatial position point, and associates the sound pressure level data as an attribute annotation to the corresponding spatial position point, and finally outputs a three-dimensional grid point set composed of spatial position points and sound pressure level annotations.

[0071] According to the three-dimensional grid point set, the 26-neighborhood rule is used to connect each node in the three-dimensional grid point set to form an initial topological connection relationship graph;

[0072] Further, based on the spatial coordinate data of the three-dimensional grid point set, search for the nearest neighbor nodes in 26 directions such as front-back, left-right, up-down, etc. within the spatial cube range for each node in the three-dimensional grid point set. When the distance between two nodes is less than or equal to the grid cell reference distance, establish a two-way connection relationship, record the start and end coordinates of the connection line segment, and finally generate an initial topological connection relationship graph composed of all valid connection relationships. The connection process strictly follows the 26-neighborhood rule, and only uses the existing node coordinates and sound pressure level data in the three-dimensional grid point set for calculation, without adding additional attributes or constraint conditions.

[0073] Through the Dijkstra shortest path algorithm, calculate the sound pressure level gradient weight between adjacent nodes in the initial topological connection relationship graph, and mark the building obstacle edges as impassable to generate a three-dimensional topological network;

[0074] Further, in the initial topological connection relationship graph, take the nodes of the three-dimensional grid point set as vertices, calculate the absolute value of the sound pressure level difference between adjacent nodes as the edge weight; identify the edges corresponding to the locations of building obstacles; the Dijkstra shortest path algorithm (Dijkstra's algorithm) takes each vertex as the starting point and iteratively calculates the shortest paths to all other vertices, and the path length is obtained by accumulating the edge weights; finally, output a three-dimensional topological network including vertex coordinates, sound pressure level data, edge connection relationships, and shortest path information.

[0075] It should be noted that the building obstacle edge refers to the edge that connects two nodes and passes through the solid components of the building BIM model in the three-dimensional topological network. The identification process of the building obstacle edge is as follows: First, extract the triangular patch geometric data of all solid components such as IfcWall and IfcSlab in the building BIM model, and construct an axis-aligned bounding box acceleration structure; then perform the intersection detection of the line segment and the triangular patch for each edge in the topological network. When the number of intersection points between the straight line segment of the edge and the triangular patch of the building BIM model is odd, it is determined as a building obstacle edge. The separation axis theorem is used in the identification process for accurate collision detection to ensure that the result is completely consistent with the geometric boundary of the building BIM model.

[0076] Calculate the sound pressure level gradient weight of adjacent nodes in the initial topological connection relationship diagram, and the expression is:

[0077] ;

[0078] Among them, is the adjacent sound pressure weight of node and node , is the A-weighted sound pressure level value of node , is the A-weighted sound pressure level value of node , is the three-dimensional coordinate vector of node , is the three-dimensional coordinate vector of node , is a minimum constant to prevent zero distance, represents the index of a single node in the three-dimensional topological network, j is the index of the adjacent node to .

[0079] It should be noted that the process of calculating the sound pressure level gradient weight of adjacent nodes is as follows: for each pair of adjacent nodes in the initial topological connection relationship diagram, first obtain the three-dimensional coordinate vector of node and the three-dimensional coordinate vector of node , and calculate the Euclidean distance between them; at the same time, read the A-weighted sound pressure level value of node and the A-weighted sound pressure level value of node , and obtain the absolute difference ; take the ratio of the sound pressure level difference to the distance as the basic weight, and then add the minimum constant ϵ to prevent zero distance, and finally obtain the adjacent sound pressure weight between node and node .

[0080] After numbering the nodes of the three-dimensional topological network in the order of spatial coordinates, establish a deterministic mapping rule for the one-to-one correspondence between the node index and the quantum bit register, and map each node in the three-dimensional topological network to a quantum bit according to the deterministic mapping rule, and form a quantum bit string;

[0081] Furthermore, the three-dimensional topological network nodes are sorted mainly according to the horizontal direction component of the spatial position, secondarily according to the vertical direction component, and finally according to the height direction component to establish a strictly ordered node number sequence; the node number range is determined according to the capacity of the qubit register, and a linear mapping relationship between the node index and the qubit register position is established, where the node number directly corresponds to the physical position index of the qubit register; in the order of the sorted node numbers, each three-dimensional topological network node is sequentially mapped to a qubit in the qubit register, maintaining the correspondence between the node attributes and the qubit states; finally, a qubit string is formed.

[0082] Based on the qubit string, all qubits are initialized to a uniform superposition state through Hadamard gate transformation, and the sound propagation path score is calculated. The expression is:

[0083] ;

[0084] Among them, is the sound propagation path score, is the distance penalty coefficient (default 0.1), which is used to balance the weights of the sound pressure level and the path length, and is used to identify and traverse the specific position points on the path , is the candidate path from the starting point to the ending point, which is composed of a series of continuously connected nodes in the three-dimensional topological network;

[0085] It should be noted that based on the qubit string, first, a Hadamard gate transformation (Hadamard gate transformation) is applied to all qubits to make the qubits in a uniform superposition state of the 0 state and the 1 state; then, all possible path combinations represented by the qubit string are traversed. For each candidate path P, the A-weighted sound pressure level values of all nodes on the path are calculated sum, and at the same time, the Euclidean distance between adjacent nodes in the path is accumulated ; the sum of the sound pressure levels and the result of multiplying the total path length by the distance penalty coefficient are added to obtain the value of the path evaluation function ; the amplitude of the quantum state corresponding to the optimal path is enhanced through the quantum amplitude amplification algorithm, and finally the sound propagation path that minimizes is measured as the optimal sound propagation path.

[0086] Based on the A-weighted sound pressure level values of the nodes of the three-dimensional topological network and the adjacent sound pressure weights , a weighted network structure data is generated through the conversion of the adjacency list data structure;

[0087] It should be noted that each node in the three-dimensional topological network is stored as a vertex element of the adjacency list, and the vertex attribute records the A-weighted sound pressure level value of the node. Traverse all valid connection edges in the initial topological connection relation graph, and create edge records for each pair of adjacent nodes and in the adjacency list. The edge attribute stores the adjacent sound pressure weight. For the connection relation corresponding to the building obstacle edge, mark the weight of the connection relation corresponding to the building obstacle edge as infinity in the adjacency list; the finally generated weighted network structure data includes a vertex set, an edge set, and the corresponding sound pressure level value and weight parameter, completely retaining the geometric connection relation and acoustic characteristics of the three-dimensional topological network.

[0088] Based on the weighted network structure data, initialize the quantum annealing parameters.

[0089] Furthermore, according to the number of vertices and connection relations in the weighted network structure data, determine the number of qubits required for the quantum annealing algorithm, and based on the adjacent sound pressure weight and the building obstacle edge marking in the weighted network structure data, construct the Hamiltonian of the Ising model (Ising model) that describes the acoustic propagation path optimization problem, where the coupling term coefficient is determined by and the local field term coefficient is determined by the vertex sound pressure level value ; for example, set the initial temperature of the quantum annealing to 10 times the maximum value in the weighted network structure data, and the termination temperature to 0.1 times the minimum value. The annealing time is set to 20 microseconds according to the characteristics of the quantum processor. This step realizes the accurate mapping of the acoustic propagation path optimization problem to the quantum annealing calculation model, providing standardized input parameters for the subsequent quantum annealing calculation.

[0090] According to the initialized quantum annealing parameters, iteratively optimize the weights of the nodes in the three-dimensional topological network through the quantum annealing algorithm, and combine the optimal acoustic propagation path to generate an acoustic topological feature map.

[0091] Furthermore, based on the weighted network structure data, first determine the calculation scale required for the quantum annealing, and match the number of qubits with the network vertex scale; then construct a physical model that describes the acoustic propagation characteristics, where the network connection relation and acoustic parameters are directly converted into model parameters; set the temperature parameter of the annealing process, the initial value is determined by the maximum connection strength of the network, and the termination value is determined by the minimum connection strength; iteratively optimize the network node parameters through the quantum annealing process, and adjust the connection weight while maintaining the original network topology; finally, fuse the optimization result with the acoustic propagation path data to form a topological feature map that reflects the sound field distribution and propagation characteristics.

[0092] Step 3: According to the acoustic topology feature map, extract multi-scale noise propagation features through a convolutional neural network, and combine the monitoring device parameter library and the coverage constraint conditions to construct a grid-based spatial demand matrix. According to the grid-based spatial demand matrix, generate an optimal noise monitoring network site selection scheme;

[0093] The multi-scale noise propagation features include the near-field features of the sound source, the reflection wave features, and the acoustic diffraction attenuation features. The extraction process is as follows:

[0094] Based on the acoustic topology feature map, extract the near-field features of the sound source through a small-scale convolutional kernel, extract the reflection wave features through a medium-scale dilated convolution, and extract the acoustic diffraction attenuation features at the building edge through a separable convolutional kernel;

[0095] Furthermore, in the process of processing the acoustic topology feature map, first use a 3×3 small-scale convolutional kernel to densely scan the map, slide and calculate in the near-field area of the sound source with a step size of 0.5 meters to extract the local detailed features of the rapid change of the sound pressure level; then use a 5×5 medium-scale dilated convolutional kernel with a dilation rate of 2 to sparsely sample in the reflection area at an interval of 1 meter to capture the periodic intensity distribution pattern formed by the reflection wave; at the same time, apply a depthwise separable convolutional kernel to perform directional convolution operations along the building edge contour, where a 3×1 horizontal convolutional kernel detects the lateral diffraction attenuation and a 1×3 vertical convolutional kernel detects the longitudinal diffraction attenuation. The three convolutional operations use the same input data but are performed independently. The small-scale convolutional kernel focuses on the 0-5 meter near-field area, the medium-scale dilated convolution covers the 5-20 meter reflection area, and the separable convolutional kernel specifically processes the data within 0.3 meters of the building edge. All convolutional results maintain the same spatial reference system and coordinate accuracy as the original acoustic topology feature map, and finally output a three-channel feature map containing near-field details, reflection features, and diffraction attenuation, completely retaining the multi-scale physical characteristics of the sound field propagation.

[0096] Through the multi-scale sound field decomposition algorithm, identify the spatial distribution characteristics of the multi-scale noise propagation features;

[0097] It should be noted that during the execution of the multi-scale sound field decomposition algorithm, the acoustic topology feature map is sent to three parallel processing channels for feature extraction. The first channel uses a Gaussian difference filter bank to process the near-field area and extracts the point source radiation features through Gaussian kernel operations of different scales; the second channel uses a multi-directional Gabor filter bank to scan the mid-field area and captures the reflection wave propagation pattern using direction-sensitive filter kernels; the third channel processes the far-field area based on the signal time-frequency analysis method and separates the environmental noise components through non-stationary signal decomposition techniques. After the processing results of each channel are spatially registered and feature fused, a feature distribution map containing three propagation mechanisms of direct sound, reflected sound, and environmental noise is generated, completely characterizing the spatial distribution law of the propagation characteristics of different-scale sound fields.

[0098] Based on the spatial distribution characteristics of multi-scale noise propagation features, the monitoring devices are divided into high-frequency, medium-frequency, and low-frequency devices through frequency band division, and a monitoring device parameter library containing monitoring devices in different frequency bands is defined;

[0099] Furthermore, based on the spatial distribution characteristics of multi-scale noise propagation features, first, according to the frequency component distribution range in the sound field decomposition result, the monitoring devices are divided into three categories: high-frequency monitoring devices, medium-frequency monitoring devices, and low-frequency monitoring devices; the configuration parameters of high-frequency monitoring devices focus on the ability to capture the near-field sound pressure that changes rapidly, the parameters of medium-frequency monitoring devices emphasize the steady-state measurement accuracy of the reflected sound field, and the parameters of low-frequency monitoring devices optimize the long-term environmental noise monitoring performance; the technical specifications, frequency response range, sensitivity index and other parameters of the three types of monitoring devices are stored in a structured manner to form a monitoring device parameter library containing monitoring devices in different frequency bands.

[0100] Based on the spatial distribution characteristics of multi-scale noise propagation features, through feature intensity clustering analysis, the near-field sound source concentration area, reflection wave interference area, and far-field attenuation gradient area are marked;

[0101] Furthermore, in the analysis of the spatial distribution characteristics of multi-scale noise propagation features, first, the three groups of data of the near-field characteristics of the sound source, reflection wave characteristics, and acoustic diffraction attenuation characteristics are standardized, and the spectral clustering algorithm is used to divide the spatial area into three categories: the first type of area is dominated by the near-field characteristics of the sound source with a high sound pressure level gradient, marked as the near-field sound source concentration area; the second type of area is mainly characterized by the reflection wave characteristics with periodic intensity changes, marked as the reflection wave interference area; the third type of area is marked by the continuous attenuation of acoustic diffraction attenuation characteristics, marked as the far-field attenuation gradient area. This step realizes the accurate division of the noise propagation area based on multi-physical characteristics.

[0102] According to the near-field sound source concentration area, reflection wave interference area, and far-field attenuation gradient area, through the spatial grid discretization algorithm, the multi-scale noise propagation features are transformed into grid coverage indicators;

[0103] Furthermore, based on the spatial distribution characteristics of the near-field sound source concentration area, reflection wave interference area, and far-field attenuation gradient area, the monitoring area is discretized into regular grid cells by using an adaptive grid division method. Each grid cell is assigned corresponding feature weight values according to the characteristic intensity ratio of the near-field sound source concentration area, reflection wave interference area, and far-field attenuation gradient area, and the noise coverage indicator value of the regular grid cell is obtained through weighted fusion calculation.

[0104] According to the grid coverage indicator, the coverage constraint condition is generated by Kriging interpolation method.

[0105] Further, based on the grid coverage index matrix, the ordinary Kriging interpolation method is used to calculate the semi-variogram of the coverage index values between grid cells, and the spherical model is fitted to determine the range, sill value, and nugget effect parameters; then the Kriging equations are established, and the optimal linear unbiased estimate of the point to be estimated is solved using the coverage index values of the known grid cells; finally, the coverage constraint conditions are generated.

[0106] It should be noted that the coverage constraint conditions include the quantitative requirements for the coverage degree of the noise propagation characteristics at each position in the monitoring area, which are used to guide the optimal deployment of monitoring equipment. For example, at the boundary of an industrial plant area, it is required that the noise coverage index is not less than 0.8 to ensure the complete monitoring of the main noise sources.

[0107] Based on the monitoring equipment parameter library, high-frequency, medium-frequency, and low-frequency equipment are respectively matched to the near-field sound source concentration area, the reflected wave interference area, and the far-field attenuation gradient area to generate a candidate set of equipment deployments;

[0108] Further, according to the technical specifications of the high-frequency monitoring equipment, medium-frequency monitoring equipment, and low-frequency monitoring equipment in the monitoring equipment parameter library, the high-frequency monitoring equipment is deployed to the near-field sound source concentration area to capture the rapidly changing near-field noise characteristics, the medium-frequency monitoring equipment is arranged in the reflected wave interference area to monitor the steady-state noise formed by the reflected waves, and the low-frequency monitoring equipment is arranged in the far-field attenuation gradient area to collect the environmental noise propagated over long distances; by traversing all possible installation positions in the three types of characteristic areas, different configuration schemes of the high-frequency monitoring equipment, medium-frequency monitoring equipment, and low-frequency monitoring equipment are combined to form a candidate set of equipment deployments containing complete information such as equipment type, installation position, and technical parameters.

[0109] Based on the coverage constraint conditions, the candidate set of equipment deployments is transformed into a grid-based spatial demand matrix through ArcGIS;

[0110] Further, in the ArcGIS (Geographic Information System) platform, the spatial position coordinates of the candidate set of equipment deployments are superimposed and analyzed with the coverage constraint conditions, and the position of each candidate equipment is mapped to the corresponding grid cell; based on the technical parameters of the high-frequency monitoring equipment, medium-frequency monitoring equipment, and low-frequency monitoring equipment in the candidate set of equipment deployments, the minimum number of equipment configurations required for each grid cell to meet the coverage constraint conditions is calculated; through spatial join and attribute calculation, a grid-based spatial demand matrix containing grid cell ID, coordinate range, equipment type requirements, and coverage compliance status is generated. The matrix data is spatially referenced consistently with the original coverage constraint conditions, accurately reflecting the specific requirements of each area for monitoring equipment.

[0111] Based on the grid-based spatial demand matrix, the spatial layout score of the monitoring points is calculated through the non-linear coverage contribution algorithm , and the expression is:

[0112] ;

[0113] wherein, is the gridded spatial demand matrix, represents the index variable of the grid cell in the gridded spatial demand matrix in the gridded spatial demand matrix, represents the th grid cell's demand intensity, represents the th grid cell's central spatial coordinate, represents the th monitoring device's detection level, is the cost penalty coefficient (default 0.3), the set of spatial layouts of the monitoring points, is the th monitoring device's location coordinate;

[0114] Furthermore, when calculating the score of the spatial layout of the monitoring points , first traverse each grid cell in the gridded spatial demand matrix to obtain the grid cell demand intensity and the central coordinate ; for each monitoring device in the set of spatial layouts of the monitoring points , calculate the square of the Euclidean distance between the location coordinate and the grid cell center , combine the monitoring device detection level and a very small constant to obtain the coverage contribution degree of the device to the grid cell ; sum up the contribution degrees of all monitoring devices to the same grid cell, normalize it through the hyperbolic tangent function, and then multiply it by the grid cell demand intensity ; after accumulating the calculation results of all grid cells, subtract the product of the total cost of the set of spatial layouts of the monitoring points and the cost penalty coefficient to finally obtain the layout score . The calculation process strictly uses the original data of the gridded spatial demand matrix and the set of spatial layouts of the monitoring points to ensure that the score result accurately reflects the coverage effect and economy of the layout scheme.

[0115] Based on the score of the spatial layout of the monitoring points , use the genetic algorithm to iteratively optimize the set of spatial layouts of the monitoring points to generate the optimal noise monitoring network location selection scheme.

[0116] Furthermore, in the genetic algorithm optimization process, a population containing multiple sets of spatial layouts of monitoring points is first initialized, and each individual represents a possible device distribution plan. During the iteration process, the spatial layout score of each set of spatial layouts of monitoring points is calculated. Based on the score results, the device position information in different sets of spatial layouts of monitoring points is exchanged through single-point crossover operation, and the Gaussian mutation operation is used to randomly perturb the coordinates of some devices. After the selection, crossover, and mutation operations for a preset number of generations, the set of spatial layouts of monitoring points with the highest spatial layout score is output as the optimal noise monitoring network site selection plan. The optimization process completely depends on the mathematical definition of the spatial layout score of the monitoring points and standard genetic operators, ensuring that the final site selection plan achieves an optimal balance between coverage performance and deployment cost.

[0117] Step 4: Based on the optimal noise monitoring network site selection plan, conduct a feasibility assessment through coverage verification, and dynamically adjust the optimal noise monitoring network site selection plan according to the assessment results.

[0118] Based on the coverage ability data in the device parameter library, using the GIS spatial analysis method, analyze the coverage of each monitoring device for grid cells, and calculate the coverage score of the optimal noise monitoring network site selection plan through a weighted penalty function. The expression is:

[0119] ;

[0120] where is the coverage score of the optimal noise monitoring network site selection plan, represents the weight coefficient of the th grid cell, represents the coverage standard of the th grid cell, is the effective coverage radius of the th monitoring device;

[0121] Furthermore, when calculating the coverage score of the optimal noise monitoring network site selection plan, first traverse each grid cell in the gridded spatial demand matrix , and obtain the grid cell weight coefficient and the coverage standard ; for each monitoring device in the set of spatial layouts of monitoring points , judge whether the Euclidean distance between the position coordinate and the grid cell center is within the effective coverage radius Within the range, record the coverage situation through the indicator function; accumulate the coverage contributions of all monitoring devices to the grid cells and calculate the coverage gap value of the grid cells in the grid-based spatial demand matrix , take zero when the result is negative; square the gap value of each grid cell and multiply it by the weight coefficient , and accumulate the calculation results of all grid cells to obtain the final coverage score ; strictly use the grid-based spatial demand matrix and the original parameters of the spatial layout set of monitoring points to ensure that the scoring result accurately reflects the actual coverage gap situation of the layout plan.

[0122] Based on historical monitoring data, define the feasibility threshold H1;

[0123] Furthermore, when defining the feasibility threshold H1, first collect all valid noise monitoring records in the historical monitoring data, and extract the sound pressure level measurement value sequences of each monitoring point; conduct percentile analysis on the historical monitoring data, for example, select the 95th percentile value as the reference value; combine the environmental noise standard limit and the technical parameters of the monitoring equipment, and add a safety margin on the basis of the reference value. In the example, 3dB is used as the typical safety margin value, and finally determine the specific value of the feasibility threshold H1.

[0124] When ≥H1, it is considered that the current optimal noise monitoring network location selection plan is feasible;

[0125] When <H1, it is considered that the current optimal noise monitoring network location selection plan is infeasible;

[0126] When the current optimal noise monitoring network location selection plan is feasible, through the particle swarm optimization algorithm, iteratively optimize the spatial layout set of monitoring points until ≥H1 or the maximum number of iterations is reached.

[0127] Furthermore, during the execution of the particle swarm optimization algorithm, first initialize a particle swarm containing multiple spatial layout sets of monitoring points. In the example, the number of particles is set to 50, and each particle represents a possible spatial layout set of monitoring points and randomly assign the initial position and velocity. Calculate the coverage score of each particle during each iteration , for example, when ≥H1, immediately terminate the optimization; otherwise, update the particle velocity and position according to the particle's individual best position and the group's best position. In the example, the inertia weight linearly decreases from 0.9 to 0.4. The iteration process continues until the coverage score Reach the feasibility threshold H1 or reach the preset maximum number of iterations (set to 200 times for example), and finally output the set of optimal monitoring point spatial layouts that satisfy ≥H1.

[0128] This embodiment also provides a noise monitoring point location selection system based on a noise map, including: a matrix generation module, a spectrum generation module, a location selection scheme generation module, and an evaluation module;

[0129] The matrix generation module is used to obtain the sound source monitoring data and the building BIM model of the target area, perform three-dimensional grid division on the spatial range of the target area, and generate a three-dimensional noise intensity matrix;

[0130] The spectrum generation module is used to construct a three-dimensional topological network based on the three-dimensional noise intensity matrix, calculate the optimal sound propagation path using a quantum-inspired topological optimization algorithm, and combine quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature spectrum;

[0131] The location selection scheme generation module is used to extract multi-scale noise propagation features from the acoustic topological feature spectrum through a convolutional neural network, combine the monitoring device parameter library and the coverage constraint conditions to construct a grid-based spatial demand matrix, and generate an optimal noise monitoring network location selection scheme according to the grid-based spatial demand matrix;

[0132] The evaluation module is used to perform a feasibility evaluation based on the optimal noise monitoring network location selection scheme through coverage verification, and dynamically adjust the optimal noise monitoring network location selection scheme according to the evaluation result.

[0133] This embodiment also provides a computer device applicable to the case of the noise monitoring point location selection method based on a noise map, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the noise monitoring point location selection method based on a noise map proposed in the above embodiment.

[0134] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0135] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for selecting noise monitoring point locations based on a noise map as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read-Only Memory (EPROM for short), Programmable Read-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0136] In summary, in the present invention: the quantum-inspired topology optimization algorithm combines quantum annealing to optimize the weights of three-dimensional topology network nodes, realizing the intelligent optimization and dynamic adjustment of sound propagation paths, effectively capturing the complex effects of the building environment on sound wave reflection and diffraction, and enabling the generated acoustic topology feature map to accurately characterize the sound field propagation characteristics; at the same time, through the convolutional neural network multi-scale feature extraction technology, small-scale convolutional kernels, medium-scale dilated convolutions, and separable convolutional kernels are respectively used to jointly analyze the characteristics of near-field direct sound, reflected waves, and diffraction attenuation, realizing the refined analysis of the spatial heterogeneity of noise propagation.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for selecting noise monitoring point locations based on a noise map, characterized in that: including Obtain the sound source monitoring data and building BIM model of the target area, perform three-dimensional grid division on the spatial range of the target area to generate a three-dimensional noise intensity matrix, where the sound source monitoring data includes sound pressure level time history records, spectral characteristics, sound source spatial positions, and environmental parameters; Based on the three-dimensional noise intensity matrix, construct a three-dimensional topological network, use a quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combine quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature map; According to the acoustic topological feature map, extract multi-scale noise propagation features through a convolutional neural network, and combine the monitoring device parameter library and coverage constraint conditions to construct a grid-based spatial demand matrix. According to the grid-based spatial demand matrix, generate an optimal noise monitoring network site selection plan; Based on the optimal noise monitoring network site selection plan, conduct a feasibility assessment through coverage verification, and dynamically adjust the optimal noise monitoring network site selection plan according to the assessment results; The steps of using the quantum-inspired topological optimization algorithm to calculate the optimal sound propagation path, and combining quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature map are as follows: Calculate the sound pressure level gradient weights of adjacent nodes in the initial topological connection relation graph; Map the nodes in the topological network to quantum bits according to the deterministic mapping rule to form a quantum bit string; Initialize all quantum bits to a uniform superposition state through Hadamard gate transformation and calculate the optimal sound propagation path; Iteratively optimize the weights of the nodes in the three-dimensional topological network through the quantum annealing algorithm, and combine the optimal sound propagation path to generate an acoustic topological feature map; The extraction of multi-scale noise propagation features through the convolutional neural network refers to extracting the near-field features of the sound source through small-scale convolutional kernels, extracting the reflected wave features through medium-scale dilated convolutions, and extracting the acoustic wave diffraction attenuation features at the building edges through separable convolutional kernels.

2. The method for selecting noise monitoring points based on a noise map according to claim 1, characterized in that: The process of obtaining the building BIM model is to generate a triangular mesh model through Poisson reconstruction; Use the NURBS algorithm to fit the building outer surface to generate a parametric surface model, and through the IFC geometry parsing engine, identify the geometric features of the building structure to generate a preliminary building BIM model; Input the acoustic parameters in the associated material library into the preliminary building BIM model to obtain the building BIM model.

3. The method for selecting noise monitoring points based on a noise map according to claim 2, wherein: The steps of generating the three-dimensional noise intensity matrix are as follows: According to the three-dimensional computational domain boundary parameters, establish a Cartesian grid, and adjust the offset of the grid lines in the Cartesian grid from the outer surface of the building BIM model through ray collision detection to generate a building-conforming grid; According to the building-conforming grid and the sound source monitoring data, associate the point sound source to the nearest grid center point through the nearest neighbor interpolation algorithm, and establish the global sound pressure level distribution through spatial interpolation methods to generate a three-dimensional noise intensity matrix.

4. The method for selecting noise monitoring points based on a noise map according to claim 1, wherein: The steps of constructing the three-dimensional topological network are as follows: Read the grid cell coordinates and sound pressure level data in the noise intensity matrix through a COO format parser and generate a three-dimensional grid point set with sound pressure level annotations; According to the three-dimensional grid point set, connect each node in the three-dimensional grid point set using the 26-neighborhood rule to form an initial topological connection relation graph; By using the Dijkstra shortest path algorithm, calculate the sound pressure level gradient weights of adjacent nodes in the initial topological connection relation graph, mark the building obstacle edges as impassable, and generate a three-dimensional topological network.

5. The method for selecting noise monitoring points based on a noise map according to claim 1, wherein: The steps of constructing the grid space demand matrix by combining the monitoring device parameter library and the coverage constraint conditions are as follows: Identify the spatial distribution characteristics of multi-scale noise propagation characteristics through the multi-scale sound field decomposition algorithm; Divide the monitoring devices into high-frequency, medium-frequency, and low-frequency devices through frequency band division, and define the monitoring device parameter library according to the monitoring devices in different frequency bands; Convert the multi-scale noise propagation characteristics into grid coverage indicators through the spatial grid discretization algorithm; Generate the coverage constraint conditions by using the Kriging interpolation method according to the grid coverage indicators; Based on the coverage constraint conditions, convert the device deployment candidate set into a grid space demand matrix through ArcGIS.

6. The method for selecting noise monitoring points based on a noise map according to claim 5, wherein: The generation of the optimal noise monitoring network site selection scheme refers to calculating the spatial layout score of the monitoring points through the non-linear coverage contribution algorithm; Based on the spatial layout score of the monitoring points, iteratively optimize the spatial layout set of the monitoring points through the genetic algorithm to generate the optimal noise monitoring network site selection scheme.

7. The method for selecting noise monitoring points based on a noise map according to claim 1, characterized in that: The steps of performing feasibility evaluation through coverage verification based on the optimal noise monitoring network site selection scheme and dynamically adjusting the optimal noise monitoring network site selection scheme according to the evaluation results are as follows: Based on the coverage ability data in the device parameter library, use the GIS spatial analysis method to analyze the coverage of each monitoring device for the grid cells, and calculate the coverage score of the optimal noise monitoring network site selection scheme through a weighted penalty function; Conduct feasibility analysis according to the coverage score of the optimal noise monitoring network site selection scheme, and iteratively optimize the spatial layout set of the monitoring points through the particle swarm optimization algorithm.

8. A noise monitoring point location selection system based on a noise map, based on the noise monitoring point location selection method according to any one of claims 1 to 7, characterized in that: Including a matrix generation module, a spectrum generation module, a site selection scheme generation module, and an evaluation module; The matrix generation module is used to obtain the sound source monitoring data and the building BIM model of the target area, perform three-dimensional grid division on the spatial range of the target area, and generate a three-dimensional noise intensity matrix; The spectrum generation module is used to construct a three-dimensional topological network based on the three-dimensional noise intensity matrix, calculate the optimal sound propagation path by using the quantum-inspired topological optimization algorithm, and combine quantum annealing to optimize the weights of the nodes in the three-dimensional topological network to generate an acoustic topological feature spectrum; The site selection scheme generation module is used to extract multi-scale noise propagation characteristics through a convolutional neural network according to the acoustic topological feature spectrum, combine the monitoring device parameter library and the coverage constraint conditions to construct a grid space demand matrix, and generate the optimal noise monitoring network site selection scheme according to the grid space demand matrix; The evaluation module is used to perform feasibility evaluation through coverage verification based on the optimal noise monitoring network site selection scheme, and dynamically adjust the optimal noise monitoring network site selection scheme according to the evaluation results.

Images