Site deployment planning method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a site deployment planning method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring scene data of a target scene, and constructing a three-dimensional scene model of the target scene based on the scene data; semantic recognition is carried out on the model data of the three-dimensional scene model, region information of at least one region of the three-dimensional scene model and preliminary suggestion information corresponding to each region are obtained, and the preliminary suggestion information comprises wireless coverage level information and capacity demand information; the model data, the scene recognition result and the RAG knowledge base are input into an AI agent for site deployment prediction processing, and an initial site deployment planning scheme and a planning target are obtained; and performing simulation iteration processing on the initial site deployment planning scheme and the planning target based on the three-dimensional scene model to obtain a site deployment planning scheme corresponding to the target scene. By adopting the method, the site deployment planning efficiency and precision can be improved.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a site deployment planning method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] In the field of wireless communication network construction, indoor network planning is a key step in ensuring users receive stable, high-speed network services. Compared to the open and relatively regular outdoor environment, the complex and diverse structure of the indoor environment leads to complex and variable signal coverage.

[0003] Currently, indoor network planning relies heavily on on-site surveys and testing. Technicians must measure site dimensions, document obstacle distribution, and measure existing signal strength, all of which consumes significant manpower and time. Furthermore, planning schemes are often based on engineering experience, relying on past cases and operational habits, and lack precise quantitative analysis and scientific calculations, resulting in poor indoor network planning results. Summary of the Invention

[0004] Based on this, it is necessary to provide a site deployment planning method, apparatus, computer equipment, computer-readable storage medium and computer program product that can improve indoor network planning in response to the above technical problems.

[0005] In a first aspect, the present application provides a site deployment planning method, comprising:

[0006] Acquiring scene data of a target scene, constructing a three-dimensional scene model of the target scene based on the scene data, and outputting model data of the three-dimensional scene model;

[0007] performing semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, the scene recognition result including area information of at least one area and preliminary recommendation information corresponding to each of the areas, the preliminary recommendation information including wireless coverage level information and capacity requirement information;

[0008] Input the model data, the scenario recognition results, and the RAG knowledge base into the AI ​​agent to perform site deployment prediction processing to obtain an initial site deployment planning scheme and planning objectives;

[0009] A simulation iteration process is performed based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain a site deployment planning scheme corresponding to the target scene.

[0010] In one embodiment, the process of the AI ​​agent performing site deployment prediction processing includes:

[0011] For any of the areas, dividing the area into grids according to the wireless coverage level information corresponding to the area;

[0012] The RAG knowledge base is used as constraint information, and site settings are performed according to the capacity requirements corresponding to each of the areas and the grid division results of each of the areas to generate the initial site deployment planning solution.

[0013] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0014] For any of the areas, association rules are queried from the RAG knowledge base according to the area type label, wireless coverage level information and capacity requirement information corresponding to the area, and the planning target corresponding to the area is calculated according to the association rules.

[0015] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0016] According to the region type labels corresponding to the regions, simulation weights corresponding to the regions are determined and output.

[0017] In one embodiment, the performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain the site deployment planning scheme corresponding to the target scene includes:

[0018] Adopting a low-granularity simulation strategy, performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning goal, to obtain an iterative site deployment planning scheme;

[0019] A high-granularity simulation strategy is adopted to perform simulation iteration processing based on the three-dimensional scene model, the iterative site deployment planning scheme and the planning target to obtain the iterative site deployment planning scheme corresponding to the target scene.

[0020] In one embodiment, the simulation iterative processing process includes:

[0021] performing simulation evaluation based on the three-dimensional scenario model and the target site deployment planning scheme, and if the simulation results do not meet the planning objectives, adjusting the target site deployment planning scheme and continuing the simulation evaluation until the simulation results meet the planning objectives;

[0022] The target site deployment planning scheme includes an initial site deployment planning scheme and an iterative site deployment planning scheme.

[0023] In one embodiment, the adjusting the target site deployment plan and continuing the simulation evaluation includes:

[0024] After adjusting the target site deployment plan, determining an adjusted first site and an unadjusted second site;

[0025] The associated simulation data of the second site is retained, and the simulation operation is continued for the first site, and a simulation evaluation is performed based on the simulation result of the first site and the associated simulation data of the second site.

[0026] In a second aspect, the present application further provides a site deployment planning device, comprising:

[0027] A construction module, configured to acquire scene data of a target scene, construct a three-dimensional scene model of the target scene based on the scene data, and output model data of the three-dimensional scene model;

[0028] an identification module, configured to perform semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, wherein the scene recognition result includes area information of at least one area and preliminary recommendation information corresponding to each area, wherein the preliminary recommendation information includes wireless coverage level information and capacity requirement information;

[0029] A prediction module is configured to input the model data, the scenario recognition results, and the RAG knowledge base into an AI agent to perform site deployment prediction processing to obtain an initial site deployment planning scheme and planning objectives;

[0030] A simulation module is used to perform simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme and the planning target to obtain the site deployment planning scheme corresponding to the target scene.

[0031] In one embodiment, the process of the AI ​​agent performing site deployment prediction processing includes:

[0032] For any of the areas, dividing the area into grids according to the wireless coverage level information corresponding to the area;

[0033] The RAG knowledge base is used as constraint information, and site settings are performed according to the capacity requirements corresponding to each of the areas and the grid division results of each of the areas to generate the initial site deployment planning solution.

[0034] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0035] For any of the areas, association rules are queried from the RAG knowledge base according to the area type label, wireless coverage level information and capacity requirement information corresponding to the area, and the planning target corresponding to the area is calculated according to the association rules.

[0036] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0037] According to the region type labels corresponding to the regions, simulation weights corresponding to the regions are determined and output.

[0038] In one embodiment, the performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain the site deployment planning scheme corresponding to the target scene includes:

[0039] Adopting a low-granularity simulation strategy, performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning goal, to obtain an iterative site deployment planning scheme;

[0040] A high-granularity simulation strategy is adopted to perform simulation iteration processing based on the three-dimensional scene model, the iterative site deployment planning scheme and the planning target to obtain the iterative site deployment planning scheme corresponding to the target scene.

[0041] In one embodiment, the simulation iterative processing process includes:

[0042] performing simulation evaluation based on the three-dimensional scenario model and the target site deployment planning scheme, and if the simulation results do not meet the planning objectives, adjusting the target site deployment planning scheme and continuing the simulation evaluation until the simulation results meet the planning objectives;

[0043] The target site deployment planning scheme includes an initial site deployment planning scheme and an iterative site deployment planning scheme.

[0044] In one embodiment, the adjusting the target site deployment plan and continuing the simulation evaluation includes:

[0045] After adjusting the target site deployment plan, determining an adjusted first site and an unadjusted second site;

[0046] The associated simulation data of the second site is retained, and the simulation operation is continued for the first site, and a simulation evaluation is performed based on the simulation result of the first site and the associated simulation data of the second site.

[0047] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements any of the above site deployment planning methods when executing the computer program.

[0048] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements any of the above site deployment planning methods when executed by a processor.

[0049] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements any of the above site deployment planning methods when executed by a processor.

[0050] The above-mentioned site deployment planning method, apparatus, computer device, computer-readable storage medium, and computer program product obtain scene data of the target scene, construct a three-dimensional scene model of the target scene based on the scene data, output model data of the three-dimensional scene model, perform semantic recognition on the model data of the three-dimensional scene model, and obtain a scene recognition result of the three-dimensional scene model. The scene recognition result includes regional information of at least one area and preliminary recommendation information corresponding to each area. The preliminary recommendation information includes wireless coverage level information and capacity requirement information. The model data, the scene recognition result, and the RAG knowledge base are input into the AI ​​intelligent agent for site deployment prediction processing to obtain an initial site deployment planning scheme and planning targets. Simulation iterative processing is performed based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning targets to obtain a site deployment planning scheme corresponding to the target scene. Using the site deployment planning method, apparatus, computer device, computer-readable storage medium, and computer program product provided in the embodiments of the present application, the initial site deployment planning scheme and planning targets for indoor three-dimensional scenes can be automatically generated by the AI ​​intelligent agent through semantic recognition and spatial perception of indoor three-dimensional scenes and combined with the RAG knowledge base. Based on this initial site deployment planning scheme and planning objectives, a digital twin of the indoor environment is created through three-dimensional scene models and simulation technology. Wireless signal propagation is accurately simulated, and automatic site deployment planning is achieved through simulation calculations and iterations. This effectively reduces manpower input and significantly improves the planning efficiency and accuracy of site deployment. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 A schematic diagram of a flow chart of a site deployment planning method in one embodiment;

[0053] Figure 2 A schematic diagram of a process flow for an AI agent to perform site deployment prediction processing in one embodiment;

[0054] Figure 3 108 is a flow chart of step 108 in one embodiment;

[0055] Figure 4 A schematic diagram of a site deployment planning method according to an embodiment;

[0056] Figure 5 A schematic diagram of an AI agent execution in one embodiment;

[0057] Figure 6 is a schematic diagram of a scene recognition result in one embodiment;

[0058] Figure 7 It is a structural block diagram of a site deployment planning device in one embodiment;

[0059] Figure 8 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0061] In one embodiment, Figure 1 As shown, a site deployment planning method is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understood that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps 102 to 108, wherein:

[0062] Step 102: Acquire scene data of a target scene, construct a three-dimensional scene model of the target scene based on the scene data, and output model data of the three-dimensional scene model.

[0063] In this embodiment of the present application, the target scene may be an indoor space where network deployment is to be implemented. The scene data for the target scene may include at least one of the following: an architectural plan, point cloud data, and CAD (Computer Aided Design) drawing data. The architectural plan provides the two-dimensional layout of the target scene, including information such as room layout, wall locations, and door and window orientations. Point cloud data, acquired through technologies such as laser scanning, more accurately represents the three-dimensional form of the target scene space. CAD drawing data, as a professional design product, contains detailed parameters such as scene structure design and dimensioning, and can be used in conjunction with the above two to build a comprehensive foundation for scene recognition.

[0064] After acquiring the scene data of the target scene, three-dimensional modeling can be performed based on the scene data to construct a three-dimensional scene model of the target scene, and further output model data corresponding to the three-dimensional scene model. The model data can include spatial information, such as the length, width, and height of the scene, the three-dimensional coordinates of each region, and the relative positional relationship; it can also include material information to specify the material type and electromagnetic characteristic parameters of each region or structural element. The material information can be data automatically associated with the three-dimensional scene model based on the scene data, or it can be information manually bound to the structural elements in the three-dimensional scene model after the three-dimensional scene model is constructed. This is not specifically limited in the embodiments of the present application. The electromagnetic characteristic parameters corresponding to the material can be used for subsequent wireless coverage simulation calculations.

[0065] It should be noted that the embodiments of the present application do not specifically limit the method of constructing a three-dimensional scene model. Any method that can construct a three-dimensional scene model based on scene data such as building floor plans, point cloud data, CAD drawing data, etc. is applicable to the embodiments of the present application. For example: the scene data can be processed based on a pre-trained network model for three-dimensional modeling, and the scene data can be feature extracted and spatially reconstructed through a deep learning algorithm to obtain a corresponding three-dimensional scene model, or a modeling algorithm (such as a Poisson reconstruction algorithm) or tool can be used to convert a two-dimensional drawing or discrete point cloud data into a three-dimensional model.

[0066] Step 104, perform semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, the scene recognition result includes area information of at least one area and preliminary recommendation information corresponding to each area, the preliminary recommendation information includes wireless coverage level information and capacity requirement information.

[0067] In the embodiment of the present application, after the three-dimensional scene model is constructed, semantic recognition can be further performed on the model data of the three-dimensional scene model. By analyzing the characteristics of each area in the three-dimensional scene model, the functions and characteristics of different areas can be determined, such as identifying areas such as conference rooms, offices, and corridors. Further, based on the functional differences of these areas, the wireless coverage level information and capacity level information required for each area are determined. For example, the conference room requires high-capacity coverage due to the dense population and high network usage demand, while the corridor only requires basic coverage. This process can be implemented by extracting regional functional attributes and network demand parameters through algorithms, models, or tools. The specific implementation process is not specifically limited in the embodiment of the present application.

[0068] Exemplarily, a three-dimensional scene model and / or model data of a three-dimensional scene model can be identified and processed by a spatial clustering algorithm or a deep learning model to output regional information of at least one area. The regional information may include the identification of the area, the area type label (such as conference room, office, corridor, staircase, etc. type labels), spatial coordinates, dimensions, materials, etc. In addition, the RAG (Retrieval-augmented Generation) knowledge base rules can be combined to assist in the judgment. For example, an area with an area greater than 200 square meters and containing projection equipment is preferentially identified as a conference room, and a long and narrow area connecting multiple rooms is determined to be a corridor. Among them, the RAG knowledge base is a knowledge base system that applies the Retrieval-Augmented Generation (RAG) technology. It is a structured information set designed specifically for the RAG system and is used to support the data source of the Retrieval-Augmented Generation technology.

[0069] After obtaining the regional information, the wireless coverage level and capacity requirement information for each region can be further determined. For example, the wireless coverage level and capacity requirement information can be determined based on the region type tag. For example, the region type tag can be used to analyze the spatial importance of the region, user behavior, and propagation environment complexity, and then set the corresponding wireless coverage level information for the region. For example, an office may require high coverage to support multiple people accessing the network, while a corridor may only require basic coverage.

[0070] Furthermore, user density and service bandwidth requirements can be analyzed based on the area type tag, and corresponding capacity requirement information can be set for the area. For example, corridors have a low user density and low service bandwidth requirements, so relatively small capacity requirement information can be set for them. Conference rooms, on the other hand, often have a high user density and require services such as video conferencing, so relatively large capacity requirement information can be set. Capacity requirement information can be determined by user density and single-user bandwidth requirements. For example, the user density of an area can be determined based on the area type tag and the area size, and the single-user bandwidth requirements can be determined based on the area type tag. Furthermore, the corresponding capacity requirement information can be determined based on the user density and single-user bandwidth.

[0071] The final output of scene recognition results is presented in the form of structured data, including area information (area ID, type, spatial coordinates, size, material) and preliminary recommendation information (wireless coverage level information (which can be expressed as signal strength) and capacity requirement information (which can be expressed as bandwidth requirement)). For example:

[0072] Region ID: R001

[0073] Area Type: Meeting Room

[0074] Position coordinates: (10, 20, 0)

[0075] Dimensions: 12m long x 8m wide x 3m high

[0076] Material: concrete + glass

[0077] Wireless coverage level information: Coverage rate 95%

[0078] Capacity requirement information: user density 0.5 people / ㎡, single-user bandwidth 20Mbps, total capacity 960Mbps.

[0079] In one embodiment, the recognition results can be superimposed on the three-dimensional scene model in the form of a heat map, and the wireless coverage level information can be intuitively displayed through color coding, or the capacity demand information can be marked at the center of the area to provide a visual reference for subsequent planning decisions.

[0080] In step 106, the model data, scene recognition results, and RAG knowledge base are input into the AI ​​agent for site deployment prediction processing to obtain the initial site deployment planning scheme and planning goals.

[0081] In an embodiment of the present application, the AI ​​agent can use the scene recognition results (area type labels, wireless coverage level information, and capacity level information of each area) and the RAG knowledge base to determine the preliminary location, number, and equipment type of wireless sites, and obtain an initial site deployment planning plan. At the same time, it can query the association rules between area types and coverage levels, capacity, etc. through the knowledge graph, calculate basic KPIs (Key Performance Indicators) such as RSRP (Received Signal Reference Power) and SINR (Signal-to-Interference-plus-Noise Ratio) based on the queried association rules, and reserve a certain amount of redundant calculation throughput to obtain the planning target.

[0082] Step 108 : Perform simulation iteration based on the three-dimensional scenario model, the initial site deployment plan, and the planning target to obtain a site deployment plan corresponding to the target scenario.

[0083] In embodiments of the present application, simulation software can be used to perform a simulation evaluation based on a three-dimensional scenario model and an initial site deployment plan. Exemplarily, the simulation evaluation can include both wireless coverage simulation evaluation and capacity simulation evaluation. A wireless coverage simulation evaluation can be performed first to prioritize the number of sites and their basic distribution that meet coverage requirements. A capacity simulation evaluation can then be performed. If the capacity simulation results do not meet the capacity requirements, engineering parameters such as site locations and equipment parameters can be adjusted, and then either the wireless coverage simulation evaluation or the capacity simulation evaluation can be performed again.

[0084] Both wireless coverage and capacity simulation assessments can be performed using simulation software or tools. For example, by building a digital twin of the indoor environment using a high-precision three-dimensional scene model and integrating advanced simulation techniques such as ray tracing, the propagation path, attenuation characteristics, and interference distribution of wireless signals within the target scenario can be accurately simulated. During the wireless coverage simulation process, by simulating signal propagation and coverage scenarios, core indicators such as signal strength and interference level are deeply analyzed and compared with pre-set planning targets. If the evaluation results indicate that the current solution fails to meet the target, targeted adjustments are made to engineering elements such as deployment location and equipment parameters. For example, antenna types with different gain and radiation patterns are replaced, and parameter configurations such as transmit power are optimized. The simulation evaluation process is then restarted. Through multiple rounds of iterative optimization, wireless signal coverage efficiency is gradually improved, approaching the ideal state, and ultimately achieving the planning target.

[0085] The capacity simulation evaluation phase follows the same logic as above. After wireless coverage meets the requirements, the system capacity is further verified to meet actual requirements. If the capacity simulation results fall short of expectations, the plan is refined by adjusting engineering parameters and conducting another wireless coverage or capacity simulation evaluation. After multiple rounds of simulation iterations, the final site deployment plan is output, ensuring the optimal configuration of the wireless communication system in terms of both coverage and capacity.

[0086] In one example, the initial site deployment plan and planning objectives generated by the AI ​​agent can be manually reviewed. Using their expertise and practical experience, the human can determine whether the plan meets actual requirements and engineering standards. If the plan is suitable, it is submitted for subsequent simulation evaluation. If there are any issues with the plan, the AI ​​agent can be re-executed by modifying parameters to regenerate the initial site deployment plan and planning objectives, or manual modifications can be made before submitting it for subsequent simulation evaluation. This human intervention can further ensure the rationality and feasibility of the plan and avoid significant deviations due to the limitations of AI decision-making.

[0087] The above-mentioned site deployment planning method obtains the scene data of the target scene, constructs a three-dimensional scene model of the target scene based on the scene data, outputs the model data of the three-dimensional scene model, performs semantic recognition on the model data of the three-dimensional scene model, and obtains the scene recognition result of the three-dimensional scene model. The scene recognition result includes the regional information of at least one area and the preliminary recommendation information corresponding to each area. The preliminary recommendation information includes wireless coverage level information and capacity demand information. The model data, the scene recognition result and the RAG knowledge base are input into the AI ​​intelligent agent for site deployment prediction processing, and the initial site deployment planning scheme and planning goals can be obtained. The simulation iterative processing is performed based on the three-dimensional scene model, the initial site deployment planning scheme and planning goals to obtain the site deployment planning scheme corresponding to the target scene. The site deployment planning method provided in the embodiment of the present application can be used to automatically generate the initial site deployment planning scheme and planning goals for the indoor three-dimensional scene by performing semantic recognition and spatial perception on the indoor three-dimensional scene and combining the RAG knowledge base with the help of the AI ​​intelligent agent. Based on this initial site deployment planning scheme and planning objectives, a digital twin of the indoor environment is created through three-dimensional scene models and simulation technology. Wireless signal propagation is accurately simulated, and automatic site deployment planning is achieved through simulation calculations and iterations. This effectively reduces manpower input and significantly improves the planning efficiency and accuracy of site deployment.

[0088] In an exemplary embodiment, referring to Figure 2 As shown, the process of the AI ​​agent performing site deployment prediction processing may include the following steps 202 to 204, wherein:

[0089] Step 202: For any area, divide the area into grids according to the wireless coverage level information corresponding to the area;

[0090] Step 204 : Using the RAG knowledge base as constraint information, perform site configuration according to the capacity requirements of each area and the grid division results of each area, and generate an initial site deployment plan.

[0091] In this embodiment of the present application, multi-dimensional feature data can first be collected based on model data, scene recognition results, and the RAG knowledge base, including scene features, demand features, and constraint features. Scene features can include area type labels (such as conference room, office), material (such as concrete, glass), spatial dimensions, and obstacle distribution, which directly affect wireless signal propagation. Demand features are used to accurately define network usage standards for each area, including wireless coverage level information (such as -90dBm@95%), meaning that 95% of the area must have a signal strength of -90dBm) and capacity requirements (such as 50Mbps / user). Constraint features introduce frequency band propagation characteristics (such as the 3.5GHz penetration loss model) and antenna pattern parameters, limiting wireless signal propagation conditions and device performance. These data are interrelated and together form the input conditions for the initial point generation algorithm.

[0092] The initial site generation algorithm is based on the concept of spatial segmentation and employs a heuristic strategy to quickly generate a preliminary wireless site layout. For any area, a grid is first created based on the corresponding wireless coverage level information. The grid resolution is positively correlated with the wireless coverage level, meaning that areas with high coverage levels have higher grid resolution than areas with low coverage levels. For example, a fine grid is used in conference rooms to ensure planning accuracy, while a coarse grid is used in corridors to improve efficiency.

[0093] During the segmentation process, 3D voxelization technology is used to decompose each region into regular cubic grid cells. For each grid cell, electromagnetic characteristics such as signal attenuation and reflection are pre-calculated based on the material properties of the 3D scene model and electromagnetic wave propagation theory.

[0094] The RAG knowledge base serves as the core constraint information source, extracting specialized knowledge about frequency band propagation characteristics, antenna performance parameters, and the impact of building structures on signals. When configuring sites, appropriate frequency bands and equipment types can be selected based on capacity requirements and the frequency band propagation characteristics in the RAG knowledge base (e.g., the 2.4 GHz band has strong signal diffraction but low transmission rate, while the 5 GHz band has high transmission rate but high penetration loss). Antenna installation locations and orientations are also determined based on antenna performance parameters (e.g., antenna gain, horizontal beamwidth, and vertical beamwidth). During site selection, open areas within the grid area are prioritized to minimize signal obstruction. The distance between sites must meet the minimum inter-site spacing constraint (e.g., at least half the carrier wavelength) to avoid co-channel interference. It should be noted that the RAG knowledge base includes not only specialized knowledge related to site deployment planning but also historical test data and case studies from other scenarios. This embodiment of the present application does not specifically limit the RAG knowledge base.

[0095] Through an iterative optimization algorithm, site locations and equipment parameters are continuously adjusted until the capacity requirements of all grid areas are met, ultimately generating an initial site deployment plan.

[0096] By adopting the site deployment planning method provided in the embodiment of the present application, the initial site deployment planning scheme is generated by combining the AI ​​agent with the RAG knowledge base, and subsequent simulation iterations are performed based on this, which can effectively improve the convergence speed and the accuracy of the final generated site deployment planning method.

[0097] In an exemplary embodiment, the region information includes a region type label, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0098] For any area, according to the area type label, wireless coverage level information and capacity requirement information corresponding to the area, the association rules are queried from the RAG knowledge base, and the planning target corresponding to the area is calculated according to the association rules.

[0099] In this embodiment, association rules are precisely retrieved from the RAG knowledge base based on area type, wireless coverage level information, and capacity requirements. These association rules reflect industry experience and technical standards. Based on these association rules, basic KPIs such as RSRP and SINR are calculated to measure signal strength and quality. A 20% redundant calculation throughput is reserved to ensure network stability during peak usage. The resulting planning targets are ultimately returned in dictionary form.

[0100] For example, taking any area as an example, the AI ​​agent uses the area type label (such as conference room, office area, corridor), wireless coverage level information (such as -90dBm @ 95% coverage), and capacity requirement information (such as 50Mbps / user) as core search criteria to accurately retrieve association rules from the RAG (Retrieval Enhancement Generation) knowledge base. This knowledge base integrates the experience of industry experts, communication technology standards, and a large amount of real-world scenario data, covering multi-dimensional knowledge such as frequency band propagation characteristics, antenna selection principles, and the impact of different building structures on signals.

[0101] The AI ​​agent first matches the predefined scenario templates in the knowledge base using the area type tag. For example, when the area type is "Conference Room," the system retrieves the "High Capacity, High Reliability Scenario" rule set, which contains typical configuration and performance requirements for conference rooms. Subsequently, the rules are further filtered and refined based on the wireless coverage level and capacity requirement information. For example, if the wireless coverage level of a conference room is -90dBm@95% and the capacity requirement is 50Mbps / user, specific rules such as signal strength, interference tolerance, and device selection that match these requirements are extracted from the rule set.

[0102] Based on the extracted association rules, the AI ​​agent can use built-in algorithms to calculate the corresponding planning objectives for the area. For example, based on the signal propagation model (such as the free space path loss formula) in the association rules, combined with the area's material properties (obtained from 3D model data) and frequency band characteristics (retrieved from the knowledge base), it can calculate basic KPI indicators such as reference signal received power (RSRP) and signal-to-interference-plus-noise ratio (SINR). For example, if the area's walls are made of concrete and use the 3.5 GHz frequency band, the system automatically substitutes the concrete's penetration loss parameter for that frequency band to calculate the RSRP threshold that meets the required coverage level.

[0103] Considering the dynamic nature of network usage and adhering to the "20% redundancy" rule, the target throughput for the area is calculated based on the user concurrency rate and service type (such as video conferencing or general office use). For example, if an office area is expected to accommodate 100 people, with a basic bandwidth requirement of 50 Mbps per person and a concurrency rate of 60%, the target throughput is calculated as 100 × 50 Mbps × 60% × 1.2 = 3600 Mbps.

[0104] Based on the antenna selection and deployment principles in the regulations, and taking into account the area dimensions and obstacle distribution, determine device parameters such as antenna type (omnidirectional / directional), gain, mounting height, and tilt angle. For example, in narrow corridors, directional antennas are recommended, and the antenna tilt angle needs to be adjusted based on the corridor length and ceiling height.

[0105] In this way, the planning target of the target scenario site deployment planning can be obtained, and then simulation iterations can be performed based on the planning target and the initial site deployment planning scheme to finally obtain a site deployment planning scheme that achieves the planning target.

[0106] In an exemplary embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing may further include the following steps:

[0107] According to the region type label corresponding to the region, the simulation weight corresponding to each region is determined and output.

[0108] In the embodiment of the present application, the analytic hierarchy process can be used to determine the simulation weight of each area in the simulation evaluation. First, a hierarchical structure including factors such as wireless coverage quality, capacity requirements and scene importance is constructed. Each main factor can be further subdivided into sub-factors. For example, wireless coverage quality can be subdivided into signal strength uniformity and edge coverage effect. Capacity requirements can be subdivided into user density adaptability, peak bandwidth carrying capacity, and multi-service concurrent support. Scene importance can be subdivided into business sensitivity of associated areas (such as conference rooms > ordinary corridors), traffic fluctuation coefficient, future scalability requirements, etc. In the embodiment of the present application, the division of sub-factors is not specifically limited. Those skilled in the art can divide sub-factors based on demand.

[0109] By querying expert experience from the RAG knowledge base, a weight matrix can be set to quantify the importance of each factor. For example, office areas have high requirements for capacity and wireless coverage quality, so these two factors can be assigned higher simulation weights. This allows for differentiated consideration of different regions and factors during the simulation evaluation phase based on weights. The simulation weights serve as weighting coefficients for each region's performance indicators, prioritizing key issues such as signal coverage blind spots and capacity bottlenecks in high-weighted areas. This ensures that optimization resources are allocated to core scenarios and core areas. For example, during the iterative process of simulation, adjustment, and re-simulation, parameter adjustments (such as increasing site density or replacing high-gain antennas) are prioritized for high-weighted areas until their performance meets the target. After optimizing the core areas, adjustments are made to non-core areas (low-weighted areas) based on remaining resources, prioritizing key areas and important indicators. Alternatively, during the simulation evaluation process, the simulation weights can be positively correlated with the simulation accuracy of the algorithm used. For example, a relatively high-precision algorithm can be used for simulation evaluation of high-weighted areas, while a relatively low-precision algorithm can be used for simulation evaluation of low-weighted areas. In this way, the simulation process can be made more accurate and efficient based on the simulation weights.

[0110] In an exemplary embodiment, referring to Figure 3As shown, in step 108, a simulation iteration process is performed based on the three-dimensional scenario model, the initial site deployment plan, and the planning target to obtain a site deployment plan corresponding to the target scenario. The process may include the following steps 302 to 304, wherein:

[0111] Step 302: Using a low-granularity simulation strategy, perform simulation iteration based on the three-dimensional scenario model, the initial site deployment plan, and the planning objectives to obtain an iterative site deployment plan.

[0112] Step 304 , adopting a high-granularity simulation strategy, performing simulation iteration based on the three-dimensional scenario model, the iterative site deployment planning scheme and the planning target, and obtaining the iterative site deployment planning scheme corresponding to the target scenario.

[0113] In the embodiments of the present application, a hierarchical simulation strategy can be employed during the simulation iteration process. This includes a low-granularity simulation strategy and a high-granularity simulation strategy. The low-granularity simulation strategy uses a lightweight simulation algorithm to perform simulations to preliminarily determine the number of sites and their basic distribution, quickly establishing a rough outline for the site deployment plan. The high-granularity simulation strategy uses a high-precision simulation algorithm to perform simulations to adjust the edge coverage of the site deployment plan and refine the details of the site deployment plan.

[0114] For example, the low-granularity simulation phase uses simplified computational models and efficient algorithms to achieve rapid iteration. During the simulation, based on the basic geometric information of the 3D scene model (such as room layout and wall location) and the initial site deployment plan (site location and device type), empirical propagation models (such as the COST-HATA model) are used, combined with frequency band propagation characteristic parameters, to provide a preliminary estimate of signal coverage.

[0115] This phase performs calculations on a macro-area basis, ignoring details such as small obstacles and material non-uniformity. Each area is abstracted as a unified signal propagation unit, significantly reducing computational complexity. By rapidly simulating signal strength distribution, the algorithm can initially determine the rationality of the number and layout of sites, identifying obvious areas of weak coverage and potential interference points. For example, in an office area, if simulation results show that signal strength in certain rooms is generally below the planned target, the algorithm will flag that area as a priority for subsequent adjustments.

[0116] After multiple rounds of iterative adjustments to parameters such as site location and transmit power, an iterative site deployment plan that meets basic coverage requirements is generated.

[0117] The high-granularity simulation phase utilizes a precise computational model based on ray tracing, combining high-precision details of the 3D scene model (such as material electromagnetic properties and specific door and window dimensions), iterative site deployment plans, and planning objectives to conduct in-depth simulation analysis. This phase divides the scene into a dense grid, performing point-by-point calculations of the signal propagation path for each grid cell, accurately simulating physical phenomena such as reflection, refraction, diffraction, and penetration loss in complex indoor environments. Furthermore, a multipath effect model is introduced to account for the overlapping effects of wireless signals reaching the receiver via different paths, more realistically reproducing signal propagation characteristics.

[0118] High-granularity simulation focuses on analyzing problem areas identified during the low-granularity simulation phase. For example, signal blind spots at corridor corners are optimized by adjusting antenna orientation and adding relay equipment. During the optimization process, various KPIs (such as RSRP, SINR, and throughput) are continuously monitored to ensure that the adjusted solution not only resolves local issues but also maintains the stability of overall network performance.

[0119] After multiple rounds of meticulous parameter fine-tuning and solution iteration, a target scenario iterative site deployment plan that meets all planning objectives is ultimately generated. This plan includes detailed information such as the precise location of each site, equipment selection, and installation method, providing clear and specific guidance for subsequent engineering construction and network deployment. Furthermore, based on supplementary simulation, all relevant planning parameters and simulation results can be output, forming a specific plan and simulation renderings, presenting the planning results in a visual form.

[0120] Through a layered simulation iteration strategy, a feasible solution is quickly built at the macro level, and then the details are deeply optimized at the micro level. This not only ensures planning efficiency, but also significantly improves the accuracy and reliability of the solution, thereby achieving precise and efficient deployment of indoor wireless sites.

[0121] In an exemplary embodiment, the simulation iterative process may include the following steps:

[0122] A simulation evaluation is performed based on the 3D scenario model and the target site deployment plan. If the simulation results do not meet the planning objectives, the target site deployment plan is adjusted and the simulation evaluation is continued until the simulation results meet the planning objectives. The target site deployment plan includes an initial site deployment plan and an iterative site deployment plan.

[0123] In order to achieve accurate optimization of the wireless site deployment planning scheme, a closed-loop iterative simulation evaluation mechanism is adopted in the embodiment of the present application. Relying on the high-precision digital twin capabilities of the three-dimensional scene model, the site deployment plan is verified and adjusted multiple times in combination with the planning goals.

[0124] For example, based on the complete spatial structure information (such as room layout and wall dimensions) and material electromagnetic parameters (concrete penetration loss of 10dB / m, glass 3dB / m) in the 3D scene model, a ray tracing simulation engine is used to perform a detailed simulation of the target site deployment plan. This stage uses a grid to calculate the propagation path of wireless signals in complex environments point by point, accurately simulating physical phenomena such as reflection, refraction, diffraction, and penetration loss. It also considers multipath effects and interference superposition, and outputs visualization results such as signal strength heat maps and SINR (signal to interference and noise ratio) distribution maps.

[0125] During the evaluation process, the simulation data planning targets were compared item by item. For example, the RSRP (reference signal received power) in each area was verified to ensure it met the -90dBm@95% coverage standard, marking areas where the signal strength fell below the threshold. The system simulated multi-user concurrent scenarios to calculate whether the actual throughput met the 50Mbps / user requirement and detect any bandwidth bottlenecks. Furthermore, the system analyzed co-channel interference between adjacent sites to ensure the SINR quality requirement of 15dB or higher.

[0126] When the simulation results do not meet the planning goals, a multi-dimensional solution adjustment mechanism is triggered, including site parameter optimization, site location migration, and adding / deleting sites.

[0127] Exemplary site parameter optimization includes: for weak coverage areas, prioritizing adjustment of transmit power (e.g., from 10dBm to 15dBm), antenna tilt (downtilt increased from 5° to 8°), or replacement of high-gain antennas; if there is co-channel interference, switching frequency bands (e.g., 2.4GHz→5GHz) or adjusting the antenna's horizontal direction angle to avoid the interference direction.

[0128] Site location migration involves using genetic algorithms or particle swarm optimization algorithms to optimize the locations of sites near signal blind spots, and moving sites to areas with weak signals while meeting the minimum site spacing (half-wavelength constraint, such as 6.25 cm for 2.4 GHz).

[0129] Adding / deleting sites includes: adding relay sites to high-concurrency areas (such as conference rooms) based on capacity demand calculation results; reducing the number of sites in coverage-redundant areas to optimize resource allocation.

[0130] In an exemplary embodiment, continuing the simulation evaluation after adjusting the target site deployment plan may include:

[0131] After adjusting the target site deployment plan, determine the adjusted first site and the unadjusted second site; retain the associated simulation data of the second site, and continue to perform simulation operations on the first site, and perform simulation evaluation based on the simulation results of the first site and the associated simulation data of the second site.

[0132] In an embodiment of the present application, after the adjustment plan, an incremental simulation strategy can be used to improve computing efficiency, that is, only the changed site (first site) and the area within its influence range are re-simulated, and the historical simulation data (such as signal propagation path, loss parameters) of the unadjusted site (second site) are reused in the evaluation phase, that is, simulation evaluation is performed based on the simulation results of the first site and the associated simulation data retained by the second site. In an embodiment of the present application, by comparing the changes in the grid of the three-dimensional scene model before and after the adjustment, the area that needs to be recalculated can be automatically identified to reduce the amount of repeated calculations. For example, when the position of a site moves, only the grid within a 50m radius covered by the site is updated and calculated, and other areas directly call cached data, which significantly shortens the iteration cycle.

[0133] Continue the iterative cycle of simulation evaluation, solution adjustment, and re-evaluation until convergence conditions are met. For example, convergence conditions may include: full regional compliance: 100% of KPI indicators such as coverage, capacity, and interference in all functional areas meet the planned targets; parameter stability: site location adjustments are less than 1 meter and device parameter changes are less than 5% over three consecutive iterations; and cost optimization: while meeting performance requirements, the solution achieves the global optimal solution for cost indicators such as device quantity and energy consumption.

[0134] After the iterations converge, the final site deployment plan is output, including a detailed site list (coordinates, equipment models, antenna parameters), a simulation evaluation report (heat map, KPI compliance rate), and an optimization log.

[0135] In order to enable those skilled in the art to better understand the embodiments of the present application, the embodiments of the present application are described below with reference to specific examples.

[0136] This embodiment of the application proposes a site deployment planning method based on digital twins and AI agents. Through this embodiment of the application, semantic recognition and spatial perception of indoor three-dimensional scenes can be performed. Combined with RAG technology and a local knowledge base, the AI ​​agent is used to automatically provide initial site locations and planning KPI recommendations for the scene. 3D modeling and ray tracing simulation are used to create a digital twin of the indoor environment, accurately simulating indoor wireless signal propagation conditions. Automatic planning is achieved through simulation calculations and iterations, improving the algorithm convergence speed and computing power efficiency.

[0137] This embodiment of the application utilizes semantic recognition and spatial perception of three-dimensional scenes, combined with RAG technology and a local knowledge base, and uses an AI agent to automatically provide initial deployment locations and planning KPI recommendations for the scenario. This changes the manual process of initial deployment in traditional wireless planning, improves the convergence speed of the automatic planning algorithm and the efficiency of computing power, and is highly innovative. Incorporating expert experience, the AI ​​agent's output reflects sound engineering experience and lays the foundation for rapid convergence of iterative simulations. This approach is particularly applicable to operators' deep indoor coverage scenarios and is highly practical.

[0138] For example, refer to Figure 4 As shown, the site deployment planning in the embodiment of the present application may include the following steps:

[0139] Step 4.1, obtain building floor plan or point cloud data:

[0140] The building floor plan provides the two-dimensional layout of the site, including information such as room distribution, wall location, door and window orientation; point cloud data is obtained through technologies such as laser scanning, which can more accurately present the three-dimensional form of the space.

[0141] Step 4.2, build a 3D scene model:

[0142] Based on building floor plans or point cloud data, a 3D scene model is constructed, and the structural elements in the model are bound to materials. Material properties are associated with electromagnetic characteristics such as electromagnetic wave penetration loss and reflection coefficient, and are used for subsequent simulation calculations. Using this data to generate a 3D scene not only outputs spatial information such as length, width, height, and the relative positions of each area, but also clarifies material information, such as whether the wall is concrete, wood, or glass. Different materials attenuate wireless signals to varying degrees, affecting signal coverage.

[0143] Step 4.3, scene analysis and information extraction:

[0144] AI algorithms perform scene semantic recognition and determine the functions and characteristics of each area in a 3D scene. For example, it can identify different areas such as conference rooms, offices, and corridors. Based on this, it determines the required wireless coverage and capacity levels for each area. For example, a conference room may require high capacity to support multiple people with simultaneous high-speed internet access, while a corridor may only require basic coverage. It also generates initial recommendations for key performance indicators (KPIs) for coverage in each area, such as signal strength and bandwidth requirements.

[0145] Step 4.4, knowledge integration and reference:

[0146] The RAG local knowledge base, containing planning expert experience amassed through long-term practical experience, guides site deployment planning. Frequency band information determines the propagation characteristics of wireless signals, with different frequency bands exhibiting varying penetration and coverage. Antenna information, including antenna type and gain, influences signal transmission and reception. This knowledge provides professional support for the development of initial site deployment plans and simulation iterations.

[0147] Step 4.5, AI agent performs:

[0148] Through the execution of the AI ​​agent, an initial site deployment plan and planning objectives are generated. Based on the scenario analysis results and knowledge base information, the preliminary locations, number, and device types of wireless sites are determined. Simulation algorithm weights are also set for each scenario area, as different areas have different signal sensitivities and requirements. For example, an office area may have a higher weight and require more precise simulation.

[0149] Reference Figure 5 The figure shows the process of the AI ​​agent performing site deployment prediction processing. The AI ​​agent adopts a multi-stage hybrid algorithm framework, combining expert knowledge and intelligent optimization strategy. The specific implementation steps are as follows:

[0150] Input feature engineering: Collect multiple features related to the scenario, requirements, and constraints. Scenario features include the semantic labels of the 3D model (such as area type and material properties), spatial dimensions, and obstacle distribution. Requirement features include wireless coverage levels in each area (e.g., -90dBm @ 95%) and capacity requirements (e.g., 50Mbps / user). Constraint features include frequency band propagation characteristics (such as the 3.5GHz penetration loss model) and antenna pattern parameters. These features serve as the foundational input for subsequent algorithms.

[0151] Initial site generation algorithm: A heuristic algorithm based on spatial segmentation is used. First, the scene is gridded according to semantic regions. The grid resolution is dynamically adjusted based on the coverage level, with finer meshes in high-level areas and coarser meshes in low-level areas. Next, sites are selected, prioritizing the center of open areas while avoiding strong obstructions (such as reinforced concrete walls) and meeting minimum site spacing constraints (e.g., ≥λ / 2, where λ is the carrier wavelength). Finally, site density is set based on capacity requirements, for example, one wireless site per 200 square meters in an office area.

[0152] KPI target generation model: Leveraging a multi-objective optimization method based on a knowledge graph, the knowledge graph is searched for relevant rules based on area type, wireless coverage level, and capacity requirements. Basic KPIs, such as RSRP (reference signal received power) and SINR (signal to interference and noise ratio), are then calculated based on these rules, reserving 20% ​​redundant throughput. Finally, a dictionary containing these KPIs is returned.

[0153] Simulation Weight Allocation Strategy: The Analytic Hierarchy Process (AHP) was used to determine regional weights. A hierarchy of factors, including coverage quality, capacity requirements, and scenario importance, was constructed, with each factor further broken down into sub-factors. Expert experience was used to establish a weight matrix to clarify the importance of each factor in the overall assessment.

[0154] Step 4.6, manual assisted review:

[0155] The initial site deployment planning scheme and planning objectives output by the AI ​​agent are manually reviewed. If suitable, they are submitted for subsequent simulation evaluation. If not suitable, the parameters are modified and the AI ​​agent is re-executed, or the output results are manually modified and submitted for subsequent simulation evaluation.

[0156] Step 4.7, simulation evaluation:

[0157] Based on 3D scenario modeling and the initial site deployment plan, simulation evaluations are conducted to simulate wireless signal propagation and coverage in real-world scenarios. Using simulation software, metrics such as signal strength, interference, and throughput are analyzed. Based on the simulation results, engineering parameters such as site locations and equipment specifications are adjusted. Evaluations are then repeated, and iterations are continuously conducted until pre-defined KPIs are met.

[0158] Step 4.8, iterative simulation to see if it meets the requirements:

[0159] The initial site deployment plan is checked for compliance. If it does, the results are output. If not, engineering parameters such as site or antenna parameters are modified and simulation evaluation is continued. During simulation iterations, only the relevant data for the modified sites is simulated. Simulation data for unchanged sites is retained and not recalculated, reducing global computing pressure.

[0160] During the overall iteration process, low-granularity simulations were used to initially determine the number of sites and their basic distribution. High-granularity simulations were then used to adjust the edge coverage. Finally, supplementary simulations were used to refine the details for a final evaluation, identifying and addressing any missed areas with weak coverage. Through task decomposition and layered simulations, the overall computational workload was effectively reduced, accelerating iteration efficiency.

[0161] Step 4.9: Output the final layout plan

[0162] After multiple rounds of optimization, the final wireless site deployment plan is output, including detailed information such as the precise location of each site, equipment selection, and installation method, providing clear guidance for subsequent engineering construction and network deployment. Planning plan and simulation output: Based on the supplementary simulation, all planning plan-related parameters and simulation results are output, forming a detailed plan and simulation renderings.

[0163] The following describes an embodiment of the present application by taking the wireless site deployment planning of a large indoor office scene as an example.

[0164] Construct a 3D scene model: Based on the building plan, a 3D scene model is constructed to obtain output space information, such as length, width, height, and the relative positions of each area, and to clarify the material information of the main walls, doors, windows, etc.

[0165] Perform scene analysis and information extraction: Use AI algorithms to perform scene semantic recognition and determine the functions and characteristics of each area in the three-dimensional scene. Identify different areas of area AG, and on this basis, determine the wireless coverage level and capacity level recommendations required for each area. The semantic recognition results can be referred to Figure 6 shown.

[0166] Expert experience and local knowledge base: For example, in the 3.5 GHz frequency band, the site spacing in area B is about 30-40 meters, while the site spacing in area C is about 20-30 meters.

[0167] AI agent execution: Combining the entire plane area, regional divisions, and semantic recognition results, the preliminary number of deployment points is determined to be 12, and the initial locations are given, resulting in the initial site deployment plan and planning goals.

[0168] After manual review of the preliminary results, they were submitted for simulation evaluation and, based on the simulation results, iterated. Finally, a site deployment plan consisting of 15 sites and location recommendations was developed.

[0169] Through digital twin modeling and scene semantic recognition, the initial deployment plan and simulation algorithm weights for each scene area are automatically given, and a method for automatic wireless site planning and iterative evaluation is performed.

[0170] The site deployment planning method provided by the embodiments of this application can effectively reduce the total amount of computing tasks while ensuring the effectiveness of the final solution, providing a strong basis for wireless network planning. Furthermore, this method can also be used to automatically generate standardized planning solution data, providing training data for machine learning methods.

[0171] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0172] Based on the same inventive concept, embodiments of the present application also provide a site deployment planning device for implementing the aforementioned site deployment planning method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations in one or more of the following site deployment planning device embodiments can be found in the aforementioned limitations on the site deployment planning method and will not be further elaborated here.

[0173] In an exemplary embodiment, Figure 7 As shown, a site deployment planning device 700 is provided, comprising: a construction module 702, an identification module 704, a prediction module 706 and a simulation module 708, wherein:

[0174] A construction module 702 is configured to obtain scene data of a target scene, construct a three-dimensional scene model of the target scene based on the scene data, and output model data of the three-dimensional scene model;

[0175] an identification module 704 configured to perform semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, wherein the scene recognition result includes area information of at least one area and preliminary recommendation information corresponding to each area, wherein the preliminary recommendation information includes wireless coverage level information and capacity requirement information;

[0176] Prediction module 706, configured to input the model data, the scenario recognition results, and the RAG knowledge base into an AI agent to perform site deployment prediction processing to obtain an initial site deployment planning scheme and planning objectives;

[0177] The simulation module 708 is configured to perform simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain a site deployment planning scheme corresponding to the target scene.

[0178] The site deployment planning device provided in the embodiments of this application can automatically generate an initial site deployment plan and planning objectives for indoor 3D scenes by performing semantic recognition and spatial perception of the indoor 3D scene, combined with a RAG knowledge base, using an AI agent. Based on this initial site deployment plan and planning objectives, a digital twin of the indoor environment is created using 3D scene models and simulation technology. This accurately simulates wireless signal propagation conditions, and automated site deployment planning is achieved through simulation calculations and iterations. This effectively reduces human effort and significantly improves the efficiency and accuracy of site deployment planning.

[0179] In one embodiment, the process of the AI ​​agent performing site deployment prediction processing includes:

[0180] For any of the areas, dividing the area into grids according to the wireless coverage level information corresponding to the area;

[0181] The RAG knowledge base is used as constraint information, and site settings are performed according to the capacity requirements corresponding to each of the areas and the grid division results of each of the areas to generate the initial site deployment planning solution.

[0182] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0183] For any of the areas, association rules are queried from the RAG knowledge base according to the area type label, wireless coverage level information and capacity requirement information corresponding to the area, and the planning target corresponding to the area is calculated according to the association rules.

[0184] In one embodiment, the region information includes a region type tag, and the process of the AI ​​agent performing site deployment prediction processing further includes:

[0185] According to the region type labels corresponding to the regions, simulation weights corresponding to the regions are determined and output.

[0186] In one embodiment, the performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain the site deployment planning scheme corresponding to the target scene includes:

[0187] Adopting a low-granularity simulation strategy, performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning goal, to obtain an iterative site deployment planning scheme;

[0188] A high-granularity simulation strategy is adopted to perform simulation iteration processing based on the three-dimensional scene model, the iterative site deployment planning scheme and the planning target to obtain the iterative site deployment planning scheme corresponding to the target scene.

[0189] In one embodiment, the simulation iterative processing process includes:

[0190] performing simulation evaluation based on the three-dimensional scenario model and the target site deployment planning scheme, and if the simulation results do not meet the planning objectives, adjusting the target site deployment planning scheme and continuing the simulation evaluation until the simulation results meet the planning objectives;

[0191] The target site deployment planning scheme includes an initial site deployment planning scheme and an iterative site deployment planning scheme.

[0192] In one embodiment, the adjusting the target site deployment plan and continuing the simulation evaluation includes:

[0193] After adjusting the target site deployment plan, determining an adjusted first site and an unadjusted second site;

[0194] The associated simulation data of the second site is retained, and the simulation operation is continued for the first site, and a simulation evaluation is performed based on the simulation result of the first site and the associated simulation data of the second site.

[0195] Each module in the above-mentioned site deployment planning device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0196] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. 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 internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means. The wireless means can be implemented via Wi-Fi, a mobile cellular network, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a site deployment planning method. The display unit of the computer device is used to produce a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0197] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0198] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0199] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0200] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0201] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0202] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0203] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0204] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A site deployment planning method, characterized in that: The method comprises: Acquiring scene data of a target scene, constructing a three-dimensional scene model of the target scene based on the scene data, and outputting model data of the three-dimensional scene model; performing semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, the scene recognition result including area information of at least one area and preliminary recommendation information corresponding to each of the areas, the preliminary recommendation information including wireless coverage level information and capacity requirement information; Input the model data, the scenario recognition results, and the RAG knowledge base into the AI ​​agent to perform site deployment prediction processing to obtain an initial site deployment planning scheme and planning objectives; A simulation iteration process is performed based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain a site deployment planning scheme corresponding to the target scene.

2. The method according to claim 1, characterized in that The process of the AI ​​agent performing site deployment prediction processing includes: For any of the areas, dividing the area into grids according to the wireless coverage level information corresponding to the area; The RAG knowledge base is used as constraint information, and site settings are performed according to the capacity requirements corresponding to each of the areas and the grid division results of each of the areas to generate the initial site deployment planning solution.

3. The method according to claim 2, characterized in that The region information includes a region type label. The process of the AI ​​agent performing site deployment prediction processing further includes: For any of the areas, association rules are queried from the RAG knowledge base according to the area type label, wireless coverage level information and capacity requirement information corresponding to the area, and the planning target corresponding to the area is calculated according to the association rules.

4. The method according to claim 3, characterized in that The region information includes a region type label. The process of the AI ​​agent performing site deployment prediction processing further includes: According to the region type labels corresponding to the regions, simulation weights corresponding to the regions are determined and output.

5. The method according to any one of claims 1 to 4, characterized in that The performing simulation iteration processing based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning target to obtain the site deployment planning scheme corresponding to the target scene includes: Adopting a low-granularity simulation strategy, performing simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme, and the planning goal, to obtain an iterative site deployment planning scheme; A high-granularity simulation strategy is adopted to perform simulation iteration processing based on the three-dimensional scene model, the iterative site deployment planning scheme and the planning target to obtain the iterative site deployment planning scheme corresponding to the target scene.

6. The method according to claim 5, characterized in that The simulation iterative processing process includes: performing simulation evaluation based on the three-dimensional scenario model and the target site deployment planning scheme, and if the simulation results do not meet the planning objectives, adjusting the target site deployment planning scheme and continuing the simulation evaluation until the simulation results meet the planning objectives; The target site deployment planning scheme includes an initial site deployment planning scheme and an iterative site deployment planning scheme.

7. The method according to claim 6, characterized in that Continuing the simulation evaluation after adjusting the target site deployment planning scheme includes: After adjusting the target site deployment plan, determining an adjusted first site and an unadjusted second site; The associated simulation data of the second site is retained, and the simulation operation is continued for the first site, and a simulation evaluation is performed based on the simulation result of the first site and the associated simulation data of the second site.

8. A site deployment planning device, characterized in that: The device comprises: A construction module, configured to acquire scene data of a target scene, construct a three-dimensional scene model of the target scene based on the scene data, and output model data of the three-dimensional scene model; an identification module, configured to perform semantic recognition on the model data of the three-dimensional scene model to obtain a scene recognition result of the three-dimensional scene model, wherein the scene recognition result includes area information of at least one area and preliminary recommendation information corresponding to each area, wherein the preliminary recommendation information includes wireless coverage level information and capacity requirement information; A prediction module is configured to input the model data, the scenario recognition results, and the RAG knowledge base into an AI agent to perform site deployment prediction processing to obtain an initial site deployment planning scheme and planning objectives; A simulation module is used to perform simulation iteration based on the three-dimensional scene model, the initial site deployment planning scheme and the planning target to obtain the site deployment planning scheme corresponding to the target scene.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images