A wireless channel twin method and device for complex scenarios

Abstract

The application discloses a wireless channel twin method and device for a complex scene, and belongs to the technical field of wireless networks. The method comprises the following steps: receiving a simulation instruction input by a user, wherein the simulation instruction comprises scene information and base station information of a target scene; acquiring a scene keyword in the scene information, and determining a scene model matched with the scene keyword in a scene database as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of actually measured spatial data and text description information; setting N base station models in the target scene model according to the base station information, wherein N is a positive integer; and performing wireless signal simulation testing according to the target scene model and the N base station models, so as to obtain a first simulation result of wireless signals in the target scene model under the N base station models. According to the embodiment of the application, the efficiency of simulation testing can be improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless networks, and particularly relates to a wireless channel twin method and device for complex scenes. BACKGROUND

[0002] Wireless signal simulation testing refers to simulating the propagation process of wireless signals in a virtual scene to evaluate communication performance indicators such as signal coverage and path loss under different base station configurations.

[0003] In the prior art, simulation testing often requires a user to manually import map data or building models of a scene to be simulated according to the scene, and manually set parameters such as scale and position to construct a scene model, and also manually add base stations in the scene model. Each of the above steps highly depends on human experience, resulting in a very tedious simulation testing process and low overall efficiency of simulation testing. SUMMARY

[0004] The embodiments of the application provide a wireless channel twin method and device for complex scenes, which can solve the problem of low efficiency of existing simulation testing.

[0005] In a first aspect, the embodiments of the application provide a wireless channel twin method for complex scenes, and the method comprises:

[0006] receiving a simulation instruction input by a user, the simulation instruction comprising scene information and base station information of a target scene;

[0007] obtaining a scene keyword in the scene information, and determining a scene model in a scene database that matches the scene keyword as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information;

[0008] setting N base station models in the target scene model according to the base station information; N is a positive integer;

[0009] performing wireless signal simulation testing according to the target scene model and the N base station models to obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0010] In some embodiments, the target scene model comprises at least one entity model, and the performing wireless signal simulation testing according to the target scene model and the N base station models to obtain a first simulation result of wireless signals in the target scene model under the N base station models comprises:

[0011] acquiring material parameters of each of the entity models;

[0012] determining a signal propagation model corresponding to the target scene model according to a type of the target scene model;

[0013] generating signal source parameters according to the N base station models;

[0014] performing simulation testing on a process of propagation of the signal source parameters in the target scene model according to the material parameters of the at least one entity model and the signal propagation model, to obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0015] In some embodiments, before the scene model matching the scene keyword in the scene database is determined as the target scene model corresponding to the target scene, the method further comprises:

[0016] acquiring multi-modal spatial data associated with at least one scene;

[0017] performing structural processing on the multi-modal spatial data to obtain three-dimensional grid data;

[0018] constructing at least one scene model according to the three-dimensional grid data, and constructing the scene database according to the at least one scene model, wherein the at least one scene model comprises the target scene model.

[0019] In some embodiments, the structural processing on the multi-modal spatial data to obtain three-dimensional grid data comprises:

[0020] determining a scene range represented by the multi-modal spatial data;

[0021] determining a grid resolution according to the scene range;

[0022] constructing a three-dimensional grid according to the grid resolution;

[0023] mapping the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data.

[0024] In some embodiments, the measured spatial data comprises laser point cloud data;

[0025] The mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data comprises:

[0026] mapping a plurality of point cloud points in the laser point cloud data into grid cells of the three-dimensional grid;

[0027] According to position information and attribute information of each point cloud point in the network element, statistical features of at least two point cloud points in the network element are determined, and the three-dimensional grid data is obtained.

[0028] In some embodiments, the measured spatial data includes image data, and the mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes:

[0029] At least one entity object in the image data is obtained by performing semantic segmentation on the image data through an image segmentation algorithm.

[0030] Position information and attribute information of the at least one entity object are determined.

[0031] According to the position information, the at least one entity object is mapped into a network element of a three-dimensional grid, and the attribute information of the at least one entity object is added to the network element corresponding to each entity object, and the three-dimensional grid data is obtained.

[0032] In some embodiments, the mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes:

[0033] The semantic understanding of the text description information of the scene is performed to determine the entity information contained in the text description information.

[0034] The three-dimensional grid element is converted from the entity information.

[0035] The three-dimensional grid element is mapped into a network element of the three-dimensional grid to obtain the three-dimensional grid data.

[0036] In some embodiments, the multi-modal spatial data includes text description information of a scene, and after the construction of at least one scene model according to the three-dimensional grid data, the method further includes:

[0037] In the case that the first structure information of an entity object is missing in any one of the at least one scene model, structure information in a preset database and the object type of the entity object has a mapping relationship, the structure information is determined as the first structure information.

[0038] In some embodiments, after the construction of at least one scene model according to the three-dimensional grid data, the method further includes:

[0039] The type and appearance attribute of each entity model in the at least one scene model are determined.

[0040] determining material parameters of each entity model according to the type and appearance attribute of the entity model through a material prediction model;

[0041] receiving a first input of a user on a first entity model; the first entity model is an entity model in any one of the scene models;

[0042] in response to the first input, determining the material parameters input by the first input as the material parameters of the first entity model.

[0043] In some embodiments, after the material parameters input by the first input are determined as the material parameters of the first entity model, the method further comprises:

[0044] setting P base station models in a first scene model according to the positions and the number of base stations in a first scene, wherein the first scene model is any one of the scene models, the first scene model is a scene model of the first scene, and there are P base stations in the first scene, and P is a positive integer;

[0045] performing simulation testing according to the first scene model and the P base station models to obtain a second simulation result of wireless signals in the P base stations in the first scene model;

[0046] obtaining an actual propagation result of the wireless signals in the first scene;

[0047] adjusting the material parameters of each entity model in the first scene model according to the actual propagation result and the second simulation result.

[0048] In some embodiments, the adjusting the material parameters of each entity model in the first scene model according to the actual propagation result and the second simulation result comprises:

[0049] obtaining a coverage accuracy weight coefficient, a calculation efficiency weight coefficient, and a parameter stability coefficient;

[0050] constructing an objective function according to the coverage accuracy weight coefficient, the calculation efficiency weight coefficient, and the parameter stability coefficient;

[0051] inputting the actual propagation result and the second simulation result into the objective function to obtain an objective function value;

[0052] adjusting the material parameters according to the objective function value.

[0053] In a second aspect, the embodiments of the present application provide a wireless channel twin device for a complex scene, and the device comprises:

[0054] The first receiving module is configured to receive a simulation instruction input by a user, wherein the simulation instruction comprises scene information of a target scene and base station information;

[0055] The first obtaining module is configured to obtain a scene keyword in the scene information, and determine a scene model in a scene database that matches the scene keyword as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information.

[0056] The first setting module is configured to set N base station models in the target scene model according to the base station information, wherein N is a positive integer.

[0057] The first simulation module is configured to perform a wireless signal simulation test according to the target scene model and the N base station models, to obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0058] In a third aspect, an embodiment of the present application provides a simulation test device, which comprises a processor and a memory storing computer program instructions.

[0059] The processor implements the above wireless channel twinning method for a complex scene when executing the computer program instructions.

[0060] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the above wireless channel twinning method for a complex scene.

[0061] In a fifth aspect, an embodiment of the present application provides a computer program product, which comprises computer program instructions, and the computer program instructions are executed by a processor to implement the above wireless channel twinning method for a complex scene.

[0062] In the present application, a simulation instruction input by a user can be parsed, and a target scene model can be automatically selected according to scene information and base station information in the simulation instruction, and N base station models can be set in the target scene model, and then a simulation test can be performed in the target scene model to generate a simulation result of wireless signals in the target scene. Since the deployment of the base station model and the setting of the scene model are both automatically completed by the system, the steps of manually setting a scene and manually placing a base station are saved, thereby significantly reducing the simulation preparation time and improving the execution efficiency of the overall simulation test. BRIEF DESCRIPTION OF DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly introduced as follows. Obviously, the drawings described below only constitute some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0064] Figure 1 is a flow diagram of a wireless channel twinning method for a complex scenario provided by an embodiment of the present application;

[0065] Figure 2 is a structural diagram of a wireless channel twinning device for a complex scenario provided by an embodiment of the present application;

[0066] Figure 3 is a hardware structural diagram of a simulation test device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0067] The features and exemplary embodiments of various aspects of the present application will be described in detail below, in order to make the purposes, technical solutions and advantages of the present application more clear and apparent, the present application will be further described in detail below in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, but not to limit the present application. The present application can be implemented without some of these specific details by those skilled in the art. The following description of the embodiments is only intended to provide a better understanding of the present application by showing examples of the present application.

[0068] It should be noted that, in this paper, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the elements defined by the statement "include" do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0069] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The embodiments will be described in detail below in combination with the drawings.

[0070] Specifically, to solve the problems in the prior art, the embodiment of the present application provides a wireless channel twin method and device for a complex scenario. First, the wireless channel twin method for a complex scenario provided by the embodiment of the present application is introduced.

[0071] Figure 1 A flowchart of the wireless channel twin method for a complex scenario provided by an embodiment of the present application is shown. The method comprises the following steps:

[0072] S110, receiving a simulation instruction input by a user, the simulation instruction comprising scene information of a target scenario and base station information.

[0073] In the embodiment, the simulation instruction can be a set of operation requests input by the user in a natural language or a structured manner, used to instruct the system to perform wireless channel simulation calculation under the target scenario. The simulation instruction comprises scene information of the target scenario and base station information, wherein the scene information is used to describe the environmental content in the target scenario in the simulation process, for example, the scene information can comprise the location of the target scenario and the scene type of the target scenario, and the base station information is used to describe the configuration content of the base station to be deployed in the simulation process, for example, the base station information can comprise the number of base stations, the layout of the base stations and the frequency band of the base stations.

[0074] S120, obtaining a scene keyword in the scene information, and determining a scene model matched with the scene keyword in a scene database as a target scene model corresponding to the target scenario, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information.

[0075] In the embodiment, after the system receives the simulation instruction input by the user, the system can obtain a scene keyword in the scene information, and then determine a scene model matched with the scene keyword in a scene database as a target scene model corresponding to the target scenario, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information.

[0076] The scene database refers to a data set composed of multiple structured scene models, each of which represents a specific type of physical space environment. The scene database is constructed based on multi-modal spatial data associated with at least one scene. The multi-modal spatial data includes real spatial data and text description information. The real spatial data refers to real measurement data that can directly represent the spatial form collected by sensors or surveying equipment. The real spatial data can include laser point cloud data, satellite image data, and unmanned aerial photography data. The text description information is provided by the user through natural language to semantically describe the scene space. The text description information can include: "CBD area with high-rise buildings, average building height about 50 meters", or "indoor scene, ceiling height 3 meters, space with multi-faceted glass wall".

[0077] After the system receives the simulation instruction of the user, the simulation instruction is subjected to semantic analysis to extract the scene keywords in the simulation instruction. The scene keywords refer to core words that can represent the semantic features of the target scene in the user input, such as "subway" and "tunnel". Then, the scene keywords can be matched with the semantic labels or description information of each scene model in the scene database. Through keyword comparison, semantic similarity calculation or rule mapping, the matching degree of each scene model with the input keywords is evaluated. Finally, the scene model with the highest matching degree is selected as the target scene model corresponding to the target scene for subsequent simulation.

[0078] S130, setting N base station models in the target scene model according to the base station information; N is a positive integer.

[0079] In this embodiment, the simulation instruction input by the user also includes base station information, which refers to structured simulation input data describing one or more of the number of base stations, spatial position, transmission parameters, and deployment methods. The system can set N base station models in the target scene model according to the base station information in the simulation instruction.

[0080] For example, the simulation instruction input by the user contains "deploy 3 base stations equidistantly along a straight line in the subway tunnel". The system will identify "subway tunnel" as the target scene model, and extract the base station number N = 3 and the layout method "equidistantly along a straight line" from the base station information. The base station information can also include the frequency band, transmission power, antenna height, direction angle, and other parameters of each base station. The system automatically calculates three appropriate installation positions on the tunnel path in the target scene model according to these information, and creates three virtual base station models at the corresponding positions with the set transmission characteristics, providing signal sources for subsequent signal propagation simulation.

[0081] S140, performing wireless signal simulation test according to the target scene model and the N base station models to obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0082] In this embodiment, the first simulation result can be the key propagation indicators of the wireless signal coverage, path loss distribution, and multipath characteristics of the N base stations in the target scene during simulation.

[0083] Specifically, after obtaining the target scene model and the N base station models, the system can first take each base station model as a signal transmitting source to transmit a virtual electromagnetic wave in the three-dimensional target scene according to the transmitting parameters thereof. The simulation engine combines the spatial distribution and material parameters of each entity in the target scene model, calls a preset propagation model to simulate the propagation process of the signal in the scene, and calculates the reflection, refraction, diffraction, and attenuation behaviors thereof at different positions. Finally, the system aggregates the signal propagation effects of the base stations in the entire scene to obtain the first simulation result.

[0084] In this embodiment, the simulation instruction input by the user can be parsed, and the target scene model can be automatically selected according to the scene information and the base station information in the simulation instruction. The N base station models can be set in the target scene model, and then the simulation test can be performed in the target scene model to generate the simulation result of the wireless signal in the target scene. Since the deployment of the base station models and the setting of the scene model are automatically completed by the system, the steps of manually setting the scene and placing the base stations are omitted, thereby significantly reducing the simulation preparation time and improving the execution efficiency of the overall simulation test.

[0085] As an optional embodiment, the target scene model includes at least one entity model, and the wireless signal simulation test according to the target scene model and the N base station models to obtain the first simulation result of the wireless signals in the target scene model under the N base station models includes:

[0086] Obtaining material parameters of each entity model;

[0087] Determining a signal propagation model corresponding to the target scene model according to the type of the target scene model;

[0088] Generating signal source parameters according to the N base station models;

[0089] Performing simulation test on the process of the signal source parameters propagating in the target scene model according to the material parameters of the at least one entity model and the signal propagation model to obtain the first simulation result of the wireless signals in the target scene model under the N base station models.

[0090] In this embodiment, the material parameters of the entity model refer to a parameter set characterizing the electromagnetic properties of entity objects such as buildings, vegetation, etc., which can be automatically determined by a material prediction model according to the entity type and appearance attribute of each entity model, or corrected by manual input of the user. The material parameters can include dielectric constant, magnetic permeability, electrical conductivity, and other key electromagnetic properties, which are used to accurately calculate the attenuation effect of signals passing through different materials. The type determination of the target scene model refers to dynamically selecting a propagation mechanism model that adapts to the target scene according to the type of the target scene, for example, automatically selecting a multipath propagation model containing reflection and diffraction for a densely populated urban area, and selecting a line-of-sight propagation model for an open area. The signal source parameter refers to the transmission end characteristic parameter generated according to the base station position, transmission power, and antenna pattern, which is used to construct the spatial distribution characteristics of the simulation input signal.

[0091] Specifically, in the simulation test process, the entity models in the target scene can be first determined automatically, and then the material parameters of each entity model are determined, and then the type to which the target scene belongs is identified according to a scene classification algorithm, wherein the type of the target scene can include high-rise building groups, industrial plants, or natural terrain, etc., and the corresponding signal propagation model of the type is automatically called.

[0092] Then, the system can perform propagation path calculation on the signal source parameters generated by each base station based on the material parameters of each entity model and in combination with the signal propagation model corresponding to the target scene, simulate the process of reflection, transmission, and attenuation of signals with different materials in a three-dimensional scene, and thus obtain the wireless signal propagation effect of the N base stations in the scene, i.e., the first simulation result.

[0093] For example, when the simulation instruction input by the user is "deploy 4 base stations in a city CBD area for 5G channel simulation", the system first matches the target scene model of the city CBD from the scene database according to the scene keyword "city CBD". Then, the system extracts the material parameters of each building, road, greenery, etc. in the target scene model, such as glass for high-rise outer walls and asphalt for the ground. Since the signal propagation in a high-density urban environment is complex, the system automatically selects a ray tracing model as the propagation model to accurately simulate multiple reflection and diffraction paths.

[0094] Then, according to the base station information input by the user, the system automatically deploys 4 base stations on both sides of the main street and generates corresponding signal source parameters for each base station, including frequency, transmit power, antenna height, directional diagram, etc. The simulation engine takes these base stations as signal sources, combines material parameters and ray tracing models to calculate the propagation path and energy attenuation of signals between buildings, simulate the reflection, transmission and shielding of each ray path, and finally obtain the first simulation result of the 4 base stations in the urban CBD scene. The first simulation result can include key indicators such as signal coverage heat map, path loss distribution, and multipath structure. This result can be used to evaluate the effect of base station deployment and network coverage quality.

[0095] Through the above technical solution, the application effectively solves the problem of insufficient simulation accuracy caused by the dependence of material parameter setting on manual experience, and realizes intelligent matching of propagation models in different scenarios. By automatically obtaining material electromagnetic parameters, the influence of manual input errors on simulation results is eliminated; based on the dynamic selection of propagation models according to scene types, the simulation process is more in line with the actual propagation environment; the collaborative computing mechanism of material parameters and propagation models significantly improves the accuracy of wireless signal coverage prediction in complex scenarios.

[0096] As an optional embodiment, before the scene model matched with the scene keyword in the scene database is determined as the target scene model corresponding to the target scene, the method further comprises:

[0097] Obtain multi-modal spatial data associated with at least one scene;

[0098] Structurally process the multi-modal spatial data to obtain three-dimensional grid data;

[0099] Construct at least one scene model according to the three-dimensional grid data, and construct the scene database according to the at least one scene model, wherein the at least one scene model includes the target scene model.

[0100] In this embodiment, multi-modal spatial data refers to scene-related data containing at least one of measured spatial data and text description information. Specifically, multi-modal spatial data can include laser point cloud, satellite image, unmanned aerial photography, or natural language description, etc. different modal data.

[0101] And the structured processing refers to converting multi-modal spatial data of different formats into a unified three-dimensional grid data structure. Specifically, grid resolution adaptive adjustment algorithm and spatial data mapping algorithm can be used to achieve this. The purpose is to eliminate the heterogeneity of multi-source data in spatial expression. The three-dimensional grid data refers to a spatial data organization form composed of regular grid cells. Specifically, cubic grid cells or hexagonal grid cells can be used for division.

[0102] After obtaining the three-dimensional mesh data, at least one scene model can be constructed according to the three-dimensional mesh data, and the scene model construction refers to converting the three-dimensional mesh data into a computable scene model containing entity objects and attributes thereof.

[0103] Through the technical solution, the application solves the problem of low scene modeling efficiency caused by format differences of multi-modal data, realizes automatic fusion processing of different format spatial data, reduces the need for manual intervention, and improves the construction speed and standardization degree of complex scene models.

[0104] As an optional embodiment, the structured processing of the multi-modal spatial data to obtain three-dimensional mesh data includes:

[0105] Determine the scene range represented by the multi-modal spatial data;

[0106] Determine the mesh resolution according to the scene range;

[0107] Construct a three-dimensional mesh according to the mesh resolution;

[0108] Map the multi-modal spatial data into the three-dimensional mesh to obtain the three-dimensional mesh data.

[0109] In this embodiment, after obtaining the multi-modal spatial data, the system can first analyze the spatial boundary information contained in the multi-modal spatial data to determine the scene range it represents, such as a certain urban area with a length of 500 meters, a width of 300 meters, and a height of 50 meters. Then, according to the scene range and simulation accuracy requirements, the system can automatically set an appropriate mesh resolution, such as dividing the space into cubic mesh units with a side length of 1 meter. Then, the system constructs a corresponding three-dimensional mesh structure based on the resolution to form a regular three-dimensional space framework.

[0110] Taking multi-modal spatial data including laser point cloud, satellite image, and natural language description as an example. The system can project each point in the laser point cloud into the corresponding mesh unit according to its spatial coordinates, and calculate the geometric features of the points in the mesh, such as the centroid position and point density; at the same time, the building, road, vegetation and other object categories segmented from the satellite image are mapped to the corresponding mesh according to their geographical location; if the input contains natural language description, such as "20-story office building on the east side", the system will add the structural information parsed to the mesh at the corresponding position. In this way, data of different modalities are unified and structured into a regular three-dimensional mesh, and standardized three-dimensional mesh data that can be used for subsequent modeling and simulation is finally obtained. Taking a city block as an example, point cloud data provides building shapes, image segmentation identifies road and green distribution, and natural language supplements building height and purpose, and the fusion of the three forms a three-dimensional mesh scene model with complete structure and clear semantics.

[0111] The above scheme improves the automation degree and semantic integrity of scene modeling by uniformly structuring multi-modal spatial data into a standardized three-dimensional grid, and provides a high-precision and high-consistency input basis for subsequent wireless channel simulation.

[0112] As an optional embodiment, the measured spatial data includes laser point cloud data.

[0113] The mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes:

[0114] Mapping a plurality of point cloud points in the laser point cloud data into grid cells of the three-dimensional grid;

[0115] For a network cell including at least two mapped point cloud points, determining statistical features of the at least two point cloud points in the network cell according to the position information and attribute information of each point cloud point in the network cell to obtain the three-dimensional grid data.

[0116] In this embodiment, the three-dimensional grid cell can be a cubic grid with a side length of 1 meter. After obtaining the laser point cloud data, the laser point cloud data includes a plurality of point cloud points. The system can first map each point cloud point in the laser point cloud data into the corresponding three-dimensional grid cell according to the spatial coordinates of each point cloud point. For each grid cell containing a plurality of point cloud points, the system calculates statistical features according to the position information and attribute information of the plurality of point cloud points included in these grid cells. The statistical features can include the centroid position, point density, and average height, which are used to represent the structure of the object corresponding to the cell. Taking a city block as an example, the grid in the high-rise area has a large number of points and a high height concentration. The system determines it as a dense high-rise building area accordingly, thereby forming structured three-dimensional grid data.

[0117] In the above manner, the regularity and computational efficiency of spatial representation can be significantly improved while preserving the original point cloud precision information, making subsequent scene recognition and propagation modeling more efficient and accurate, and having good compatibility and scalability.

[0118] As an optional embodiment, the measured spatial data includes image data, and the mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes:

[0119] Performing semantic segmentation on the image data by an image segmentation algorithm to obtain at least one entity object in the image data;

[0120] Determining the position information and attribute information of the at least one entity object;

[0121] mapping the at least one entity object into a grid cell of a three-dimensional grid according to the position information, and adding attribute information of the at least one entity object into the grid cell corresponding to each entity object, to obtain the three-dimensional grid data.

[0122] In this embodiment, the image data can include satellite images or image data taken by a drone. After obtaining the image data, the system can first perform an image segmentation algorithm on the input image data, identify a plurality of entity objects in the image data, such as buildings, roads, vegetation, etc., and assign a semantic label to each object. Then, the system extracts the geographic coordinates or pixel position information of these entity objects in the image and projects and maps them into the corresponding grid cells in the three-dimensional grid, while binding the attribute information of the objects to the corresponding grid cells. The attribute information can include the category, height estimate, and coverage rate of the entity object, etc.

[0123] Taking the image of a city block as an example, the system can identify a high-rise building from the image, then map the high-rise building into the grid cells of the three-dimensional grid, and mark its position and height attribute in the three-dimensional grid, realizing the conversion from two-dimensional image to structured three-dimensional space.

[0124] Through the semantic segmentation in the above manner, automatic structure recognition and accurate positioning of the image data are realized, so that unstructured image information can be efficiently embedded into regular three-dimensional grids, significantly improving the semantic integrity and spatial expression accuracy of the scene model, and providing a reliable input basis for subsequent simulation.

[0125] As an optional embodiment, the mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes:

[0126] performing semantic understanding on the text description information of the scene to determine entity information contained in the text description information;

[0127] converting the entity information into a three-dimensional grid element;

[0128] mapping the three-dimensional grid element into a grid cell of the three-dimensional grid to obtain the three-dimensional grid data.

[0129] In this embodiment, after obtaining the text description information of the scene, the system can first perform semantic understanding on the text description information of the scene to identify the entity information contained therein. For example, the entity information can be “a 20-story office building on the east side” or “a green coverage rate of 30% in the southern area”.

[0130] Then, the system can structure these entity information into three-dimensional grid elements, for example, generate a building body unit with corresponding height and attributes according to "20-story office building", and generate a grid label of the corresponding vegetation area according to the green coverage rate. Subsequently, these three-dimensional grid elements are positioned and mapped into specific grid units of the three-dimensional grid, completing the conversion of semantic information to spatial structure. Taking the office building as an example, the system can create a grid object with "building, height = 60 meters, material = glass curtain wall" attributes in the specified area of the grid.

[0131] In the above manner, abstract natural language description can be converted into quantifiable and calculable spatial data, realizing the fusion of text and spatial model, and greatly improving the automation level and semantic controllability of scene modeling.

[0132] As an optional embodiment, the multi-modal spatial data includes text description information of the scene, and after constructing at least one scene model according to the three-dimensional grid data, the method further includes:

[0133] For any one of the at least one scene model, in the case that the first structure information of the entity object existing in the scene model is missing, the structure information in the preset database and the object type of the entity object existing in the mapping relationship is determined as the first structure information.

[0134] In this embodiment, the preset database refers to a structured knowledge base that stores the mapping relationship between different object types and their standard structure information, which can be implemented by a graph database or a relational database, for example, an object type and a structure feature are associated by a Neo4j graph database, which is used to quickly retrieve standard structure parameters matching the target object type.

[0135] Among them, the structure information of the object type existing in the mapping relationship refers to the standardized structure attribute associated with the entity object type defined in the database, which can be implemented by entity relationship mapping in a knowledge graph, for example, the "office building" object type is mapped to the standard floor height and load-bearing wall distribution parameters. The "subway tunnel" object type is mapped to the metal lining structure. This feature uses the association between types and structures to ensure that the completed information and the actual physical characteristics of the scene model maintain logical consistency.

[0136] After constructing at least one scene model according to the three-dimensional grid data, it can be judged whether each scene model has a missing structure information. If the scene model has a missing structure information, a scene completion algorithm can be used to fill in the missing structure information through knowledge graph reasoning.

[0137] For example, taking any one of the at least one scene model as an example, when the system detects that a certain entity object in the scene model lacks necessary first structure information, the system can search for a standard structure configuration corresponding to the entity object in a preset structure information database or knowledge graph according to the object type of the entity object.

[0138] For example, when the user only describes a "subway tunnel" without providing specific structural details, the subway tunnel is an entity object, and the structural details of the subway tunnel are the first structure information. The system can automatically infer the first structure information according to the definition of the object type "tunnel" in the knowledge graph, such as having a metal lining, and fill the structure element into the corresponding entity object in the scene model, thereby completing the completion of the first structure information.

[0139] In the above manner, the completion mechanism of the first structure information can be automatically completed, the input burden of the user on the structural details is significantly reduced, the completeness and automation degree of the scene modeling are improved, and accurate calculation based on complete and effective scene information in the simulation process is ensured.

[0140] As an optional embodiment, after the at least one scene model is constructed according to the three-dimensional mesh data, the method further includes:

[0141] determining the type and appearance attribute of each entity model in the at least one scene model;

[0142] determining the material parameters of each entity model according to the type and appearance attribute of the entity model through a material prediction model;

[0143] receiving a first input of a first entity model by the user; the first entity model is an entity model in any one of the scene models;

[0144] in response to the first input, determining the material parameters input by the first input as the material parameters of the first entity model.

[0145] In this embodiment, the type of the entity model refers to the physical object category represented by the model, for example, the type of the entity model can include buildings, vegetation, or roads, and the type of the entity model can be identified through semantic labeling in the three-dimensional mesh, which can provide basic classification information for material parameter prediction. The appearance attribute of the entity model can include surface texture, geometric shape, and color features, which can be extracted from the texture data of the three-dimensional mesh by using an image recognition algorithm.

[0146] The material prediction model refers to a parameter mapping model based on a machine learning algorithm. Specifically, a deep neural network can be used to train a historical material dataset, thereby generating a material prediction model trained to convergence. Through the input of the combination of type and appearance attributes, the material prediction model can intelligently deduce the material parameters of each entity model.

[0147] Specifically, after constructing the scene model according to the three-dimensional grid data, the system can further identify the type (such as buildings, roads, trees, etc.) and appearance attributes (such as color, texture, shape features) of each entity model, and input these information into the material prediction model. The material prediction model automatically infers the possible material parameters of each entity in combination with the existing training data. For example, “high-rise building + glass facade” is predicted as a glass curtain wall with a dielectric constant of 4.2 and a conductivity of 0.01 S / m.

[0148] In addition, the system also supports user manual correction of any entity model. Specifically, the first input can be a parameter correction instruction input by the user through the interactive interface. If the user believes that the material parameters of the first entity model predicted by the material prediction model are incorrect, the user can input the first input to the first entity model, and then the system can update the material parameters input by the first input as the material parameters of the first entity model in response to the first input.

[0149] Specifically, after completing the three-dimensional scene modeling, the system first identifies the types and extracts the appearance features of each entity in the scene. For example, the first entity model can be a glass curtain wall entity model. For the entity model identified as a glass curtain wall, the system inputs the surface smoothness, light transmittance and other appearance attributes into the material prediction model, and the material prediction model automatically generates the reflection coefficient and transmission loss parameters of the glass curtain wall. If the glass curtain wall is a new photovoltaic glass material, the user believes that the material parameters of the glass curtain wall predicted by the material prediction model are incorrect, and the user can directly modify the material parameters of the glass curtain wall through the first input, thereby covering the prediction results of the model. This processing method not only ensures the automatic configuration efficiency of standard materials, but also provides a manual intervention channel for special materials.

[0150] Compared with the prior art, the traditional scheme completely relies on manual experience to set material parameters, and needs to consult materials and manually input electromagnetic parameters one by one. The present scheme realizes automatic parameter deduction through type classification and appearance feature analysis, greatly reducing the manual operation steps. For entities not covered by the material library, the present scheme realizes dynamic updating of parameters through real-time manual input, ensuring the continuity of the simulation process.

[0151] As an optional embodiment, after the material parameter input by the first input is determined as the material parameter of the first entity model, the method further comprises:

[0152] According to the position and number of base stations in the first scene, P base station models are set in a first scene model, wherein the first scene model is any one of the scene models, the first scene model is the scene model of the first scene, there are P base stations in the first scene, and P is a positive integer;

[0153] According to the first scene model and the P base station models, a simulation test is performed to obtain a second simulation result of wireless signals in the first scene model under the P base stations;

[0154] An actual propagation result of the wireless signals in the first scene is obtained;

[0155] According to the actual propagation result and the second simulation result, the material parameters of the entity models in the first scene model are adjusted.

[0156] In this embodiment, the first scene model refers to a three-dimensional digital model constructed according to the spatial data of the real first scene. The base station model refers to a virtual signal source constructed according to the spatial coordinates and signal transmission parameters of the real base station. Specifically, the position and number of base stations in the first scene can be obtained. If it is determined that there are P base stations in the first scene and the positions of the base stations are known, the latitude and longitude information of the P real base stations is converted into spatial coordinates in the three-dimensional first scene model by using a coordinate mapping algorithm, to ensure the consistency of the simulation environment and the base station model in the physical scene.

[0157] After the first base station model is determined and the P base stations are deployed in the first base station model, the actual propagation result of the wireless signals in the first scene can be obtained. The actual propagation result refers to the key indicator data of the wireless signals after propagation in the real environment, such as received signal strength (RSSI), path loss value, or multipath delay, which is obtained by field measurement or historical record and used as a standard for comparison of the simulation result.

[0158] Specifically, the system can automatically set P corresponding base station models as signal transmission sources in the first scene model according to the number and positions of the base stations known in the first scene, and perform simulation tests based on this to calculate the propagation of wireless signals in the first scene and generate a second simulation result. Subsequently, the system obtains the actual propagation result of the wireless signals at the corresponding positions in the first scene, i.e., the actual propagation result, for evaluating the accuracy of the simulation output. If there is a difference between the second simulation result and the actual propagation result, the system will adjust the material parameters of each entity in the scene model based on the difference, such as the reflection characteristics of the building surface or the penetration loss of the vegetation area, to optimize the model accuracy.

[0159] Through the above technical solution, the technical problems of low material parameter adjustment efficiency and poor accuracy in the traditional method are effectively solved. By automatically collecting the propagation data of the real scene and comparing it with the simulation result in real time, the parameter optimization period is significantly shortened, and the time consumption of manual repeated trial and error is avoided. By establishing a reverse parameter correction mechanism based on gradient descent, the matching accuracy of material properties and real physical characteristics is improved, and the simulation deviation caused by incorrect experience judgment is reduced.

[0160] As an optional embodiment, the adjusting the material parameters of each entity model in the first scene model according to the actual propagation result and the second simulation result comprises:

[0161] obtaining a coverage accuracy weight coefficient, a calculation efficiency weight coefficient, and a parameter stability coefficient;

[0162] constructing a target function according to the coverage accuracy weight coefficient, the calculation efficiency weight coefficient, and the parameter stability coefficient;

[0163] inputting the actual propagation result and the second simulation result into the target function to obtain a target function value;

[0164] adjusting the material parameters according to the target function value.

[0165] In this embodiment, the system can first obtain three weight coefficients for controlling the optimization direction of the material parameters, i.e., a coverage accuracy weight coefficient p, a calculation efficiency weight coefficient q, and a parameter stability coefficient m. Among them, the coverage accuracy coefficient is used to measure the error size between the simulation result and the actual propagation result, and the higher the value, the more attention is paid to the simulation accuracy in the material parameter optimization process; the calculation efficiency coefficient is used to measure the resource consumption in the simulation process, such as GPU memory occupation and calculation time; and the parameter stability coefficient is used to measure the deviation between the newly adjusted material parameters and the historical verified parameters, encouraging parameter reuse and solution stability.

[0166] According to the three coefficients, a target function is constructed, which can be:

[0167] F = p x coverage error + q x computational resource consumption + m x parameter variation range

[0168] wherein p + q + m = 1, for balancing the weights among the three.

[0169] The system then inputs the actual propagation result and the second simulation result into the target function, respectively calculates the simulation error, the resource consumption and the parameter stability degree, and thus obtains a comprehensive evaluation value, i.e., the target function value. The reinforcement learning model takes the target function value as a feedback signal, repeatedly adjusts the material parameters of each entity in the scene by trial and error, gradually optimizes the comprehensive performance among the simulation accuracy, the calculation efficiency and the parameter stability, and finally obtains the optimal parameter configuration to drive the simulation result to be closer to the real signal propagation characteristics.

[0170] In the above scheme, the target function considering the simulation accuracy, the calculation efficiency and the parameter stability is constructed to guide the system to intelligently adjust the material parameters, realize the simulation optimization with high precision, low cost and reusability, and significantly improve the credibility and engineering applicability of the simulation result.

[0171] Based on the wireless channel twinning method for complex scenarios provided in the above embodiments, correspondingly, the application also provides a specific implementation mode of a wireless channel twinning device for complex scenarios. Please refer to the following embodiments.

[0172] Firstly, referring to Figure 2 The wireless channel twinning device 200 provided by the embodiments of the application comprises the following modules:

[0173] The receiving module 201 is configured to receive a simulation instruction input by a user, wherein the simulation instruction comprises scene information of a target scene and base station information.

[0174] The first obtaining module 202 is configured to obtain a scene keyword in the scene information, and determine a scene model in a scene database matched with the scene keyword as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information.

[0175] The first setting module 203 is configured to set N base station models in the target scene model according to the base station information; N is a positive integer.

[0176] The first simulation module 204 is configured to perform wireless signal simulation testing according to the target scene model and the N base station models, and obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0177] The device can parse the simulation instruction input by the user, automatically select a target scene model according to scene information and base station information in the simulation instruction, set N base station models in the target scene model, and then perform simulation testing in the target scene model to generate a simulation result of wireless signals in the target scene. Since the deployment of the base station model and the setting process of the scene model are automatically completed by the system, the steps of manually setting the scene and placing the base station are saved, thereby significantly reducing the simulation preparation time and improving the execution efficiency of the overall simulation testing.

[0178] As an implementation manner of the present application, the first simulation module 204 can further include:

[0179] A first acquisition unit configured to acquire material parameters of each of the entity models;

[0180] A first determination unit configured to determine a signal propagation model corresponding to the target scene model according to a type of the target scene model;

[0181] A first generation unit configured to generate signal source parameters according to the N base station models;

[0182] A first simulation unit configured to perform simulation testing on a process of propagation of the signal source parameters in the target scene model according to the material parameters of the at least one entity model and the signal propagation model, to obtain a first simulation result of wireless signals in the target scene model under the N base station models.

[0183] As an implementation manner of the present application, the wireless channel twin device for complex scenes 200 can further include:

[0184] A second acquisition module configured to acquire multi-modal spatial data associated with at least one scene;

[0185] A structuring module configured to perform structured processing on the multi-modal spatial data to obtain three-dimensional grid data;

[0186] A construction module configured to construct at least one scene model according to the three-dimensional grid data, and construct the scene database according to the at least one scene model, wherein the at least one scene model includes the target scene model.

[0187] As an implementation manner of the present application, the structuring module can be further configured to:

[0188] A second determination unit configured to determine a scene range represented by the multi-modal spatial data;

[0189] A third determination unit configured to determine a grid resolution according to the scene range.

[0190] a constructing unit, configured to construct a three-dimensional grid according to the grid resolution;

[0191] a mapping unit, configured to map the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data.

[0192] As an implementation form of the present application, the mapping unit is specifically configured to:

[0193] map a plurality of point cloud points in the laser point cloud data into grid cells of the three-dimensional grid;

[0194] determine, for a network cell including at least two mapped point cloud points, statistical features of the at least two point cloud points in the grid cell according to position information and attribute information of the point cloud points in the network cell, to obtain the three-dimensional grid data.

[0195] As an implementation form of the present application, the mapping unit is specifically configured to:

[0196] perform semantic segmentation on the image data by an image segmentation algorithm to obtain at least one entity object in the image data;

[0197] determine position information and attribute information of the at least one entity object;

[0198] map the at least one entity object into grid cells of a three-dimensional grid according to the position information, and add the attribute information of the at least one entity object into the grid cells corresponding to the entity objects, to obtain the three-dimensional grid data.

[0199] As an implementation form of the present application, the mapping unit is specifically configured to:

[0200] perform semantic understanding on the text description information of the scene to determine entity information contained in the text description information;

[0201] convert the entity information into a three-dimensional grid element;

[0202] map the three-dimensional grid element into grid cells of the three-dimensional grid to obtain the three-dimensional grid data.

[0203] As an implementation form of the present application, the wireless channel twin device 200 for a complex scene can further include:

[0204] a supplement module, configured to, for any one of the at least one scene model, in a case where first structure information of an entity object in the scene model is missing, determine structure information in a preset database and having a mapping relationship with the object type of the entity object as the first structure information.

[0205] As an implementation manner of the present application, the wireless channel twin device 200 for complex scenes can further include:

[0206] a first determining module configured to determine the type and appearance attribute of each entity model in the at least one scene model;

[0207] a second determining module configured to determine the material parameter of each entity model according to the type and appearance attribute of the entity model through a material prediction model;

[0208] a second receiving module configured to receive a first input of a user to a first entity model; the first entity model is an entity model in any one of the scene models;

[0209] a correcting module configured to determine the material parameter input by the first input as the material parameter of the first entity model in response to the first input.

[0210] As an implementation manner of the present application, the wireless channel twin device 200 for complex scenes can further include:

[0211] a second setting module configured to set P base station models in a first scene model according to the position and number of base stations in a first scene, wherein the first scene model is any one of the scene models, the first scene model is the scene model of the first scene, there are P base stations in the first scene, and P is a positive integer;

[0212] a second simulation module configured to perform simulation test according to the first scene model and the P base station models to obtain a second simulation result of wireless signals under the P base stations in the first scene model;

[0213] a third obtaining module configured to obtain an actual propagation result of the wireless signals in the first scene;

[0214] an adjusting module configured to adjust the material parameter of each entity model in the first scene model according to the actual propagation result and the second simulation result.

[0215] As an implementation manner of the present application, the adjusting module is specifically configured to:

[0216] obtain a coverage accuracy weight coefficient, a calculation efficiency weight coefficient, and a parameter stability coefficient;

[0217] construct a target function according to the coverage accuracy weight coefficient, the calculation efficiency weight coefficient, and the parameter stability coefficient;

[0218] input the actual propagation result and the second simulation result into the target function to obtain a target function value.

[0219] adjusting the material parameters according to the target function value.

[0220] The wireless channel twin device for complex scenarios provided by the embodiments of the present application can implement each step in the method embodiments described above, and thus will not be described again here to avoid repetition.

[0221] Figure 3 A hardware structure schematic diagram of the simulation test device provided by the embodiments of the present application is shown.

[0222] The simulation test device can include a processor 1001 and a memory 1002 storing computer program instructions.

[0223] Specifically, the processor 1001 can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement one or more embodiments of the present application.

[0224] The memory 1002 can include a mass storage for data or instructions. By way of example and not limitation, the memory 1002 can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory 1002 can include removable or non-removable (or fixed) media. Where appropriate, the memory 1002 can be internal or external to the integrated gateway disaster recovery device. In certain embodiments, the memory 1002 is non-volatile solid-state memory.

[0225] The memory can include read-only memory (ROM), random-access memory (RAM), magnetic disk storage mediums, optical storage mediums, flash memory devices, electrical, optical, or other physical / tangible memory storage devices. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software that, when executed (by one or more processors), is operable to perform operations described with reference to the methods according to an aspect of the present disclosure.

[0226] The processor 1001 reads and executes the computer program instructions stored in the memory 1002 to implement any one of the wireless channel twin methods for complex scenarios in the embodiments described above.

[0227] In one example, the simulation test device can further include a communication interface 1003 and a bus 1010. As shown in Figure 3 the processor 1001, the memory 1002, and the communication interface 1003 are connected through the bus 1010 and complete communication with each other.

[0228] The communication interface 1003 is mainly used to realize the communication between the modules, devices, units and / or equipment in the embodiments of the application.

[0229] The bus 1010 includes hardware, software or both to couple the components of the simulation test device to each other. By way of example, and without limitation, the bus can include an accelerated graphics port (AGP) or other graphics bus, an enhanced industry standard architecture (EISA) bus, a front-side bus (FSB), a hypertransport (HT) interconnect, an industry standard architecture (ISA) bus, an infiniband interconnect, a low pin count (LPC) bus, a memory bus, a microchannel architecture (MCA) bus, a peripheral component interconnect (PCI) bus, a PCI-express (PCI-X) bus, a serial advanced technology attachment (SATA) bus, a video electronics standards association local (VLB) bus, or another suitable bus or combination of two or more of these. Where appropriate, the bus 1010 can include one or more buses. Although the embodiments of the application are described and illustrated with a specific bus, the application contemplates any suitable bus or interconnect.

[0230] The simulation test device can be based on the above embodiments, thereby realizing the wireless channel twinning method and device for complex scenarios as described above.

[0231] In addition, in combination with the wireless channel twinning method for complex scenarios in the above embodiments, the embodiments of the application can provide a computer storage medium to realize. The computer storage medium has computer program instructions stored thereon; the computer program instructions are executed by the processor to realize any one of the wireless channel twinning methods for complex scenarios in the above embodiments, and can achieve the same technical effect, to avoid repetition, which will not be repeated here. Among them, the above computer readable storage medium can include a non-transitory computer readable storage medium, such as a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, etc., which is not limited here.

[0232] In addition, the embodiments of the application also provide a computer program product, including computer program instructions, which can realize the steps and corresponding contents of the above method embodiments when executed by the processor.

[0233] It is to be understood that the application is not limited to particular configurations and processes described herein and shown in the drawings, as such can include any keeps that are within the scope of the application. Detailed descriptions of known methods are omitted so as not to obscure the application in unnecessary detail. In the above embodiments, a number of specific steps are described and illustrated as examples. However, the methods of the application are not limited to the specific steps described and illustrated, as such can include any number of additional or different steps, vary the order of the steps, or eliminate any number of the steps, as would be understood by one of skill in the art upon reading the present disclosure.

[0234] The functional blocks shown in the structural block diagrams above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, and the like. When implemented in software, the elements of the application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. A "machine-readable medium" includes any medium that can store or transfer information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, and the like. The code segments can be downloaded via computer networks such as the Internet, intranets, and the like.

[0235] It is also to be understood that the example embodiments described in this application are based on a series of steps or apparatuses to describe some methods or systems. However, the application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be performed simultaneously.

[0236] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other processing devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other processing devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. Such processes can be realized by a

[0237] The above is merely specific implementation of the present application, and those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein again. It should be understood that the protection scope of the present application is not limited to this, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical range disclosed in the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A method for complex scenario oriented wireless channel twin, characterized in that, The method comprises: receiving a simulation instruction input by a user, the simulation instruction comprising scene information of a target scene and base station information; obtaining a scene keyword in the scene information, and determining a scene model in a scene database that matches the scene keyword as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of measured spatial data and text description information; setting N base station models in the target scene model according to the base station information; N is a positive integer; performing wireless signal simulation testing according to the target scene model and the N base station models to obtain a first simulation result of wireless signals in the target scene model under the N base station models; wherein the target scene model comprises at least one entity model, and the performing wireless signal simulation testing according to the target scene model and the N base station models to obtain the first simulation result of wireless signals in the target scene model under the N base station models comprises: obtaining material parameters of each entity model; determining a signal propagation model corresponding to the target scene model according to a type of the target scene model; generating signal source parameters according to the N base station models; performing simulation testing on a process of propagation of the signal source parameters in the target scene model according to the material parameters of the at least one entity model and the signal propagation model, to obtain the first simulation result of wireless signals in the target scene model under the N base station models.

2. The method of claim 1, wherein, Before the determining a scene model in a scene database that matches the scene keyword as a target scene model corresponding to the target scene, the method further comprises: obtaining multi-modal spatial data associated with at least one scene; performing structured processing on the multi-modal spatial data to obtain three-dimensional grid data; constructing at least one scene model according to the three-dimensional grid data, and constructing the scene database according to the at least one scene model, wherein the at least one scene model comprises the target scene model.

3. The method of claim 2, wherein, The performing structured processing on the multi-modal spatial data to obtain three-dimensional grid data comprises: determining a scene range represented by the multi-modal spatial data; determining a grid resolution according to the scene range; constructing a three-dimensional grid according to the grid resolution; mapping the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data.

4. The method of claim 3, wherein, The measured spatial data comprises laser point cloud data. The mapping the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data comprises: mapping a plurality of point cloud points in the laser point cloud data into grid cells of the three-dimensional grid; for a network cell comprising at least two mapped point cloud points, determining statistical features of the at least two point cloud points in the grid cell according to position information and attribute information of each point cloud point in the network cell to obtain the three-dimensional grid data.

5. The method of claim 3, wherein, The measured spatial data includes image data, and the mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes: performing semantic segmentation on the image data by an image segmentation algorithm to obtain at least one entity object in the image data; determining position information and attribute information of the at least one entity object; mapping the at least one entity object into a grid cell of a three-dimensional grid according to the position information, and adding attribute information of the at least one entity object into the grid cell corresponding to each entity object to obtain the three-dimensional grid data.

6. The method of claim 3, wherein, The mapping of the multi-modal spatial data into the three-dimensional grid to obtain the three-dimensional grid data includes: performing semantic understanding on the text description information of the scene to determine entity information contained in the text description information; converting the entity information into a three-dimensional grid element; mapping the three-dimensional grid element into a grid cell of the three-dimensional grid to obtain the three-dimensional grid data.

7. The method of claim 2, wherein, The multi-modal spatial data includes text description information of a scene, and after the construction of the at least one scene model according to the three-dimensional grid data, the method further includes: for any one of the at least one scene model, in a case where first structure information of an entity object existing in the scene model is missing, determining structure information existing in a mapping relationship with the object type of the entity object in a preset database as the first structure information.

8. The method of claim 2, wherein, After the construction of the at least one scene model according to the three-dimensional grid data, the method further includes: determining the type and appearance attribute of each entity model in the at least one scene model; determining the material parameters of each entity model according to the type and appearance attribute of the entity model by a material prediction model; receiving a first input of a user on a first entity model; the first entity model is an entity model in any one of the scene models; in response to the first input, determining the material parameters input by the first input as the material parameters of the first entity model.

9. The method of claim 8, wherein, After the determination of the material parameters input by the first input as the material parameters of the first entity model, the method further includes: according to the base station position and the number of base stations in a first scene, setting P base station models in a first scene model, wherein the first scene model is any one of the scene models, the first scene model is a scene model of the first scene, there are P base stations in the first scene, and P is a positive integer; performing simulation testing according to the first scene model and the P base station models to obtain a second simulation result of wireless signals under the P base stations in the first scene model; obtaining an actual propagation result of the wireless signals in the first scene; adjusting the material parameters of each entity model in the first scene model according to the actual propagation result and the second simulation result.

10. The method of claim 9, wherein, The adjustment of the material parameters of each entity model in the first scene model according to the actual propagation result and the second simulation result includes: obtaining a coverage accuracy weight coefficient, a calculation efficiency weight coefficient, and a parameter stability coefficient; According to the coverage accuracy weight coefficient, the calculation efficiency weight coefficient, and the parameter stability coefficient, a target function is constructed; The actual propagation result and the second simulation result are input into the target function to obtain a target function value; The material parameters are adjusted according to the target function value.

11. A wireless channel twin device for complex scenarios, comprising: The device comprises: A first receiving module configured to receive a simulation instruction input by a user, the simulation instruction comprising scene information of a target scene and base station information; A first obtaining module configured to obtain a scene keyword in the scene information, and determine a scene model matched with the scene keyword in a scene database as a target scene model corresponding to the target scene, wherein the scene database is a database constructed based on multi-modal spatial data associated with at least one scene, and the multi-modal spatial data comprises at least one of actually measured spatial data and text description information; A first setting module configured to set N base station models in the target scene model according to the base station information, wherein N is a positive integer; A first simulation module configured to perform wireless signal simulation testing according to the target scene model and the N base station models to obtain a first simulation result of wireless signals in the target scene model under the N base station models; The target scene model comprises at least one entity model, and the first simulation module is specifically configured to obtain material parameters of each entity model, determine a signal propagation model corresponding to the target scene model according to a type of the target scene model, generate signal source parameters according to the N base station models, and perform simulation testing on a process of propagation of the signal source parameters in the target scene model according to the material parameters of the at least one entity model and the signal propagation model, to obtain the first simulation result of the wireless signals in the target scene model under the N base station models.

Images