Evaluation method and device before building indoor distribution construction and readable storage medium

Abstract

The invention provides an evaluation method and device before building indoor distribution construction and a readable storage medium. The method comprises the steps that a building CAD design plan drawing set is acquired; splitting the drawing set to obtain a CAD (Computer Aided Design) planar graph of each building; according to the CAD design plane graph of each building, identifying special space information of each building by using a pre-trained convolutional neural network model; obtaining a base map of the CAD design planar graph of which the special space information is identified, further dividing the base map into a plurality of closed spaces, and calculating the area of each closed space to obtain a processed base map; and based on the processed base map, generating an indoor distribution point location map corresponding to each building by using a preset large model according to the scene corresponding to each building and a preset point location layout principle. According to the method, the device and the medium, the problems of low evaluation efficiency, insufficient accuracy and lack of service competitiveness of an existing evaluation method before building indoor distribution construction can be solved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular, to an evaluation method, device, and readable storage medium for building indoor distribution construction before construction. Background Art

[0002] With the growth of mobile users and the increasing number of high-rise large buildings in cities, the problem of mobile communication signals inside buildings has become increasingly prominent. At the same time, relevant parties have put forward new requirements for the construction of building mobile communication infrastructure.

[0003] However, the existing evaluation methods for building indoor distribution construction before construction have many drawbacks, such as low evaluation efficiency, insufficient accuracy, lack of business competitiveness, etc., and it is difficult to meet the needs of current business development. Therefore, there is an urgent need for a digital means to assist the evaluation work for building indoor distribution construction before construction. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an evaluation method, device, and readable storage medium for building indoor distribution construction before construction in view of the above deficiencies of the prior art, so as to solve the problems of low evaluation efficiency, insufficient accuracy, and lack of business competitiveness existing in the existing evaluation methods for building indoor distribution construction before construction.

[0005] In a first aspect, the present invention provides an evaluation method for building indoor distribution construction before construction, the

[0006] method includes:

[0007] Obtain a building computer-aided design (CAD) design floor plan drawing set;

[0008] Split the drawing set to obtain CAD design floor plans of each building;

[0009] Based on the CAD design floor plans of each building, use a pre-trained convolutional neural network model to identify special space information of each building;

[0010] Obtain the base map of the CAD design floor plan with the identified special space information, further divide the base map into several enclosed spaces, and calculate the area of each enclosed space to obtain a processed base map;

[0011] Based on the processed base map, use a preset large model to generate a corresponding indoor distribution point map for each building and a signal coverage intensity prediction map for each floor according to the corresponding scenarios of each building and the preset point placement principle;

[0012] Calculate the main equipment investment and distribution system investment of each building according to the positions of antenna point placements in the corresponding indoor distribution point map of each building, and summarize to obtain the overall investment cost estimation result of the building.

[0013] Further, based on the CAD design floor plans of each building, the pre-trained convolutional neural network model is used to identify the special space information of each building, specifically including:

[0014] Preprocess the CAD design floor plans of each building;

[0015] Convert the preprocessed CAD design floor plans into vectorized data;

[0016] Input the vector data into the trained convolutional neural network model to identify the floor information of each building and the information of the special space;

[0017] Among them, the special space includes at least one of the following: walls, stairwells, and elevator shafts.

[0018] Further, obtain the base map of the CAD design floor plan with the identified special space information, further divide the base map into several closed spaces, and calculate the area of each closed space to obtain the processed base map, specifically including:

[0019] Delete the auxiliary lines and annotation information from the CAD design floor plan with the identified special space information to obtain the corresponding base map;

[0020] Use edge recognition technology to identify the external contour of the base map, and outline the overall area perimeter of each floor;

[0021] Based on the overall area perimeter of each floor, draw lines according to the dimension scale to divide the base map into several closed spaces;

[0022] Use optical character recognition (OCR) technology to identify the compartment type of each closed space and calculate the area of each closed space to obtain the processed base map.

[0023] Further, before generating the indoor distribution point map corresponding to each building and the signal coverage intensity prediction map for each floor based on the processed base map using a preset large model according to the corresponding scenarios of each building and the preset point placement principle, the method further includes:

[0024] Calculate and test the antenna intensity for different scenarios according to the free space signal arrival intensity formula to obtain the antenna spacing range for different scenarios;

[0025] Use the antenna spacing range for different scenarios as the basis for the point placement principle for the corresponding scenarios.

[0026] Further, based on the processed base map, using a preset large model, according to the scenes corresponding to each building and the preset principle of point placement, generate the indoor distribution point maps corresponding to each building and the signal coverage intensity prediction maps for each floor, specifically including:

[0027] For each building, input the processed base map into the preset large model, and use the large model to perform the following steps: Place the antenna points on the processed base map according to the floor height of the building and the point placement principle of the corresponding scene, obtain the indoor distribution point map where the signal coverage can reach the expectation, and obtain the signal coverage intensity prediction map for each floor according to the positions of the antenna points placed in the indoor distribution point map.

[0028] Further, calculating the main equipment investment and distribution system investment of each building according to the positions of the antenna points placed in the indoor distribution point maps corresponding to each building, and summarizing to obtain the estimation result of the overall investment cost of the building, specifically including:

[0029] According to the positions of the antenna points placed in the indoor distribution point maps corresponding to each building and the areas of each enclosed space, use the preset cost estimation model to calculate the main equipment investment and distribution system investment of each building respectively;

[0030] Summarize the main equipment investment and distribution system investment of all buildings to obtain the estimation result of the overall investment cost of the building.

[0031] Further, the method further includes:

[0032] According to the user's query conditions, display at least one of the following: the indoor distribution point maps corresponding to each building, the signal coverage intensity prediction map of any floor in a certain building, the estimation result of the overall investment cost of the building.

[0033] In a second aspect, the present invention provides an evaluation device before building indoor distribution in a building, and the device includes:

[0034] A drawing set acquisition module, configured to acquire a building computer-aided design (CAD) design plan drawing set;

[0035] An automatic splitting module, connected to the drawing set acquisition module, and configured to split the drawing set to obtain the CAD design plan drawings of each building;

[0036] A special space recognition module, connected to the automatic splitting module, and configured to identify the special space information of each building according to the CAD design plan drawings of each building by using a pre-trained convolutional neural network model;

[0037] The base map processing module, connected to the special space recognition module, is used to obtain the base map of the CAD design floor plan with the recognized special space information, further divide the base map into several enclosed spaces, calculate the area of each enclosed space, and obtain the processed base map;

[0038] The point map generation module, connected to the base map processing module, is used to generate the indoor distribution point map corresponding to each building and the signal coverage intensity prediction map for each floor based on the processed base map by using a preset large model according to the scenarios corresponding to each building and the preset point placement principle;

[0039] The investment cost estimation module, connected to the point map generation module, is used to calculate the main equipment investment and distribution system investment of each building according to the positions of the antenna point placements in the indoor distribution point map corresponding to each building, and summarize to obtain the estimation result of the overall investment cost of the building.

[0040] In a third aspect, the present invention provides an evaluation device before building indoor distribution, including a memory and a processor. A computer program is stored in the memory, and the processor is configured to run the computer program to implement the evaluation method before building indoor distribution described in the first aspect above.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the evaluation method before building indoor distribution described in the first aspect above.

[0042] The evaluation method, device and readable storage medium before the construction of in-building distribution provided by the present invention. First, obtain the building computer-aided design (CAD) design floor plan drawing set; and split the drawing set to obtain the CAD design floor plans of each building; then, according to the CAD design floor plans of each building, use a pre-trained convolutional neural network model to identify the special space information of each building; and obtain the base map of the CAD design floor plan with the special space information identified, further divide the base map into several closed spaces, and calculate the area of each closed space to obtain the processed base map; then, based on the processed base map, use a preset large model to generate the in-building distribution point map corresponding to each building and the signal coverage intensity prediction map of each floor according to the scene corresponding to each building and the preset point placement principle; finally, calculate the main equipment investment and distribution system investment of each building according to the position of the antenna point placement in the in-building distribution point map corresponding to each building, and summarize to obtain the overall investment cost estimation result of the building. The present invention significantly improves the efficiency and accuracy of the evaluation before the construction of in-building distribution by automatically processing the building CAD design floor plan, accurately identifying special space information using a convolutional neural network model, and combining the in-building distribution point map and signal coverage intensity prediction map generated by the large model. This solution can quickly provide accurate evaluation information, greatly enhancing the business competitiveness, and solving the problems of low evaluation efficiency, insufficient accuracy, and lack of business competitiveness in the existing evaluation methods before the construction of in-building distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flowchart of an evaluation method before the construction of in-building distribution according to Embodiment 1 of the present invention;

[0044] Figure 2 It is an architecture diagram of an evaluation system before the construction of in-building distribution according to an embodiment of the present invention;

[0045] Figure 3 It is a schematic diagram of the building area according to an embodiment of the present invention;

[0046] Figure 4 It is a schematic diagram of the in-building distribution point map in the elevator scenario according to an embodiment of the present invention;

[0047] Figure 5 It is a schematic diagram of the visualization interface according to an embodiment of the present invention;

[0048] Figure 6 It is a schematic diagram of the wireless signal coverage situation according to an embodiment of the present invention;

[0049] Figure 7 It is a structural schematic diagram of an evaluation device before the construction of in-building distribution according to Embodiment 2 of the present invention;

[0050] Figure 8Schematic structural diagram of an evaluation device before the construction of in-building distribution for Embodiment 3 of the present invention. Detailed implementation manners

[0051] To enable those skilled in the art to better understand the technical solutions of the present invention, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0052] It can be understood that the specific embodiments and accompanying drawings described herein are only for explaining the present invention, rather than limiting the present invention.

[0053] It can be understood that, without conflict, the various embodiments in the present invention and the features in the embodiments can be combined with each other.

[0054] It can be understood that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings of the present invention, and the parts not related to the present invention are not shown in the accompanying drawings.

[0055] It can be understood that each unit and module involved in the embodiments of the present invention may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple units and modules may also be integrated into one entity structure.

[0056] It can be understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of the present invention may occur in a different order from that marked in the accompanying drawings.

[0057] It can be understood that in the flowcharts and block diagrams of the present invention, the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the embodiments of the present invention are shown. Among them, each block in the flowchart or block diagram may represent a unit, module, program segment, or code, which contains executable instructions for implementing the specified function. Moreover, each block or combination of blocks in the block diagram and flowchart can be implemented by a hardware-based system for implementing the specified function, or can be implemented by a combination of hardware and computer instructions.

[0058] It can be understood that the units and modules involved in the embodiments of the present invention can be implemented in software or in hardware. For example, the units and modules can be located in the processor.

[0059] Application overview

[0060] The existing evaluation methods before the construction of in-building distribution have the following disadvantages:

[0061] 1. Low evaluation efficiency

[0062] Relying on on-site surveys and manual design drawings and calculating investment evaluations is a cumbersome process that consumes a large amount of time and labor costs. For example, when conducting pre-construction evaluations of in-building distribution for large commercial office buildings or residential communities, manual labor is required to measure each area one by one, count the number of rooms and other information, and then draw drawings and calculate costs. The entire process may take several days or even longer, and it is impossible to quickly respond to business needs. There is a lack of unified principles and efficient processing procedures for investment evaluations in different scenarios, resulting in low work efficiency, making it difficult to provide accurate evaluation results in a short period of time, and affecting the speed of business progress.

[0063] 2. Insufficient accuracy

[0064] Estimation based on manual experience is highly subjective, and there may be significant differences in the evaluations of the same building by different personnel, resulting in poor accuracy in aspects such as area calculation, antenna point planning, and investment cost estimation. For example, when estimating the area, manual measurements may produce errors due to different measurement methods and tools, which in turn affect the overall cost estimation and signal coverage effect evaluation. Traditional methods are difficult to accurately identify various detailed information in complex building drawings, such as the accurate location and size of special spaces (stairwells, elevator shafts, etc.), and the impact of different material walls on signals, thus affecting the quality of the final evaluation and planning.

[0065] 3. Lack of business competitiveness

[0066] Due to low efficiency and poor accuracy, it is impossible to quickly and accurately provide persuasive pre-construction evaluation information for in-building distribution in business negotiations, making it difficult to meet the needs of customers for quick decision-making and being unfavorable to business competition. It is unable to provide personalized and precise evaluation solutions according to different scenarios and requirements, unable to create a differentiated competitive advantage, and being in a passive position in the market competition, restricting the expansion of the enterprise in the field of in-building distribution construction business.

[0067] In view of the above technical problems, the concept of this application is to provide a pre-construction evaluation method, device, and readable storage medium for in-building distribution. By automatically processing the building CAD design floor plan, using a convolutional neural network model to accurately identify special space information, and combining the in-building distribution point map and signal coverage intensity prediction map generated by a large model, the efficiency and accuracy of the pre-construction evaluation of in-building distribution are significantly improved. This solution can quickly provide accurate evaluation information and greatly enhance business competitiveness.

[0068] After introducing the basic principle of this application, the following will specifically introduce various non-limiting embodiments of this application with reference to the accompanying drawings.

[0069] Embodiment 1:

[0070] This embodiment provides a pre-construction evaluation method for in-building distribution, as Figure 1 shown, this method includes:

[0071] Step S101: Obtain the building CAD (Computer-Aided Design) design floor plan drawing set;

[0072] In this embodiment, the user uploads the building CAD design floor plan drawing set to the system through the upload interface provided by the system, and the system obtains the building CAD design floor plan drawing set.

[0073] Step S102: Split the drawing set to obtain the CAD design floor plans of each building.

[0074] In this embodiment, the system automatically identifies the drawings of each building in the drawing set and splits them to facilitate subsequent separate processing of the drawings of each building. Among them, the building includes one or more buildings. Usually, each CAD design floor plan is a floor plan of a building (including all its floors).

[0075] Step S103: According to the CAD design floor plans of each building, use a pre-trained convolutional neural network model to identify the special space information of each building.

[0076] In this embodiment, the convolutional neural network model is pre-trained with a large amount of training set data labeled with special space information and can identify the special space information in the CAD design floor plan. Among them, the convolutional neural network model is preferably a U-Net model. The U-Net model is a deep learning model for image segmentation. Its architecture is in the shape of a U and consists of an encoder (downsampling path) and a decoder (upsampling path). The encoder gradually extracts the high-dimensional features of the image through convolution and pooling operations, while the decoder restores the features to the same size as the input image through upsampling and deconvolution operations and restores the spatial details of the image. In addition, U-Net directly connects the corresponding layers of the encoder and the decoder through skip connections, thus effectively retaining the low-level feature information, making the model perform excellently in the image segmentation task and achieving pixel-level accurate segmentation, especially suitable for small sample data scenarios.

[0077] Optionally, the step of using a pre-trained convolutional neural network model to identify the special space information of each building according to the CAD design floor plans of each building specifically includes:

[0078] Preprocess the CAD design floor plans of each building;

[0079] Convert the preprocessed CAD design floor plan into vectorized vector data;

[0080] Input the vector data into the trained convolutional neural network model to identify the floor information of each building and the information of the special space;

[0081] Among them, the special space includes at least one of the following: wall, stairwell, elevator shaft.

[0082] In this embodiment, the preprocessing includes operations such as removing noise and correcting graphic deviation to improve the quality and recognizability of the drawing. Then, planar graph vectorization is performed to convert the CAD design planar graph into vector data that can be processed by the convolutional neural network, facilitating subsequent recognition and calculation.

[0083] In this embodiment, the special space includes at least one of the following: wall, stairwell, elevator shaft. The information of the identified special space may include its position, size, or contour, etc. Among them, for the elevator scenario, while the elevator shaft is identified using the convolutional neural network model, the model automatically reads the floor information of the current building, and according to the standard of arranging one antenna every 3 floors, the number of floors is divided by 3 (rounded up) to calculate the number of antennas required for elevator shaft coverage. It should be noted that because there are clear rules for antenna placement in the elevator scenario (such as arranging one antenna every 3 floors), and the number of elevator shafts and floor information can be directly used to calculate the total number of required antennas, therefore, the total number of antennas can be calculated when the positions and numbers of the elevator shafts are identified.

[0084] Specifically, for the identification of special spaces, a convolutional neural network model can be used to simultaneously identify all special spaces, or multiple convolutional neural network models can be selected to identify them separately according to the characteristic differences between different special spaces. For example, a convolutional neural network model can be used to simultaneously identify walls, stairwells, and elevator shafts, or a convolutional neural network model can be used to identify walls and stairwells, and then another convolutional neural network model can be used to identify elevator shafts.

[0085] Step S104: Obtain the base map of the CAD design planar graph with the identified special space information, further divide the base map into several closed spaces, and calculate the area of each closed space to obtain the processed base map.

[0086] In this embodiment, the closed spaces include special spaces and other non-special spaces (such as office areas, balconies, etc.). If the drawing itself contains area information, it can also be directly read.

[0087] Optionally, the obtaining the base map of the CAD design planar graph with the identified special space information, further dividing the base map into several closed spaces, and calculating the area of each closed space to obtain the processed base map specifically includes:

[0088] Delete the auxiliary lines and dimension information from the CAD design floor plan with the identified special space information to obtain the corresponding base map;

[0089] Use edge recognition technology to identify the external contour of the base map and outline the overall area perimeter of each floor;

[0090] Based on the overall area perimeter of each floor, draw lines according to the size scale to divide the base map into several closed spaces;

[0091] Use OCR (Optical Character Recognition) technology to identify the compartment type of each closed space and calculate the area of each closed space to obtain the processed base map.

[0092] In this embodiment, a professional CAD software can be used to process the CAD design floor plan with the identified special space information. First, delete redundant elements such as auxiliary lines and dimension information in the drawing to generate a base map. Then, use edge recognition technology to outline the external contour of the base map and determine the overall area perimeter of each floor. Next, draw lines on the base map based on the size scale to divide the overall area into several closed spaces and label each closed space. Then, use OCR technology to identify the text in each closed area of the base map to identify the compartment type (such as office area, balcony, elevator, etc.). If the drawing itself contains area information, it is directly read. If not, the area of the closed space is calculated by manually drawing lines, and the overall area is further obtained.

[0093] Step S105: Based on the processed base map, use a preset large model to generate the corresponding indoor distribution point map for each building and the predicted signal coverage intensity map for each floor according to the scenarios corresponding to each building and the preset point placement principle.

[0094] In this embodiment, the preset large model refers to an industry large model developed specifically for the application scenarios of building indoor distribution construction. This model has been trained in advance based on a large amount of industry-specific data, aiming to improve the accuracy of antenna point placement and the accuracy of predicted signal coverage intensity. Through learning and training, the large model can process the input base map to automatically output the indoor distribution point map and the predicted signal coverage intensity map for each floor. This process not only improves the efficiency and accuracy of the evaluation, but also provides strong support for wireless planning and optimization in the application scenarios of building indoor distribution construction.

[0095] Optionally, before the step of generating the corresponding indoor distribution point map for each building and the predicted signal coverage intensity map for each floor based on the processed base map using a preset large model according to the scenarios corresponding to each building and the preset point placement principle, the method further includes:

[0096] Perform antenna intensity calculation tests on different scenarios according to the free space signal arrival intensity formula to obtain the antenna spacing ranges under different scenarios;

[0097] Use the antenna spacing ranges under different scenarios as the basis for the point placement principle in the corresponding scenarios.

[0098] In this embodiment, the free space signal arrival intensity formula is:

[0099] Rss = Pt + Gt + Gr - Lc - Lbf

[0100] Wherein, Rss is the signal arrival intensity, with the unit of dbm; Pt is the transmit power, with the unit of dBm; Gt is the transmit antenna gain, with the unit of dBi; Gr is the receive antenna gain, with the unit of dBi; Lc is the loss of the cable and connector, with the unit of dB; Lbf is the free space loss, with the unit of dB.

[0101] Specifically, perform antenna intensity calculation tests on different scenarios in an ideal environment to obtain the antenna spacing ranges under different scenarios as the theoretical basis for point placement. Among them, different scenarios include local space scenarios such as office buildings, factories, industrial parks, rooftops, underground parking lots, large shopping malls, residences, and elevators. For example, in the office building scenario of an industrial park, in the case of a single-channel passive distribution system, the antenna spacing is usually 12 - 20 meters, and ceiling antennas are used; while in a residential community in the residential scenario, the antenna spacing is 50 - 150 meters, and spotlights or log periodic antennas can be selected, etc.

[0102] It should be noted that since there are clear rules for antenna placement in the elevator scenario, the scenarios can also be divided into elevator scenarios and non-elevator scenarios. For the elevator scenario, the total number of antennas can be calculated when the positions and quantities of elevator shafts are identified.

[0103] Optionally, based on the processed base map, use a preset large model to generate the indoor distribution point maps corresponding to each building and the signal coverage intensity prediction maps for each floor according to the scenarios corresponding to each building and the preset point placement principle, specifically including:

[0104] For each building, input the processed base map into the preset large model, and use the large model to perform the following steps: perform antenna point placement on the processed base map according to the floor height of the building and the point placement principle of the corresponding scenario to obtain an indoor distribution point map where the signal coverage can reach the expectation, and obtain the signal coverage intensity prediction map for each floor according to the positions of the antenna points placed in the indoor distribution point map.

[0105] In this embodiment, the antenna spacing is dynamically adjusted. The large model first makes a preliminary layout of the antenna positions on the processed base map and determines the corner positioning points. Subsequently, the large model automatically places the remaining antenna positions and maps these positions to the original drawing. To solve the problem of antennas overlapping with the wall, an antenna position optimization mechanism is integrated inside the large model. After the preliminary layout is completed, the model will perform signal simulation optimization. According to the results of field strength calculation and display, if it is found that the signal coverage does not meet the expectations (for example, the proportion of the area with a signal strength above -90 dBm is less than 95%), the model will automatically adjust the antenna spacing and re-place the antennas until the coverage requirements are met. Finally, a complete indoor distribution point map and a signal coverage intensity prediction map are output.

[0106] Step S106: Calculate the main equipment investment and distribution system investment for each building according to the positions of the antenna points placed in the indoor distribution point map corresponding to each building, and summarize to obtain the overall investment cost estimation result of the building.

[0107] In this embodiment, the investment cost estimation is mainly used for the cost assessment of site construction, which is mainly divided into main equipment investment and distribution system investment. For the self-built scenario, it can effectively evaluate its cost-benefit, and for the owner investment scenario (the distribution system is invested by the developer), it can effectively balance the interests of both parties.

[0108] Optionally, the calculating the main equipment investment and distribution system investment for each building according to the positions of the antenna points placed in the indoor distribution point map corresponding to each building, and summarizing to obtain the overall investment cost estimation result of the building specifically includes:

[0109] According to the positions of the antenna points placed in the indoor distribution point map corresponding to each building and the areas of each enclosed space, use a preset cost estimation model to calculate the main equipment investment and distribution system investment for each building respectively;

[0110] Summarize the main equipment investment and distribution system investment of all buildings to obtain the overall investment cost estimation result of the building.

[0111] In this embodiment, the formula for the distribution system investment can be shown as follows:

[0112] S 1 =X 1 *(K 1 +K 2 +K 3 )+X 2 *K 4

[0113] Among them, X 1 is the unit price of indoor antenna construction, X 2 is the unit price of outdoor antenna construction, K 1 、K2 、K 2 、K 3 are pre-set parameters, which are specific numerical values and are set differently according to different scenarios.

[0114] Optionally, the method further includes:

[0115] displaying at least one of the following according to the user's query conditions: the in-building distribution point maps corresponding to each building, the signal coverage intensity prediction map of any floor in a certain building, and the estimation result of the overall investment cost of the building.

[0116] In this embodiment, the system can provide a visual interface for users to query. The user inputs query conditions (such as viewing the signal coverage of a certain floor, cost estimation) through the visual interface, and the interface displays the results in an intuitive chart.

[0117] It should be noted that the evaluation method provided by the present invention for in-building distribution construction before construction has the following beneficial effects:

[0118] a) Improving the evaluation efficiency: By using the technology of automatically processing drawings, the automatic processing of CAD design floor plans is realized, including drawing acquisition, indoor space information recognition, and drawing optimization processing, etc., to quickly and accurately obtain building information, such as floor information, area of each region, etc., and shorten the evaluation cycle. For example, by automatically identifying and splitting the drawing set by the system and performing rapid vectorization processing, accurate data basis can be provided for subsequent evaluation in a short time. With the training and optimization of the large model, the in-building distribution point maps and investment cost estimation results are intelligently output, reducing manual intervention and improving the efficiency of the overall work process. For example, the antenna point layout in different scenarios is quickly calculated through the pre-trained model, avoiding repeated manual trial and error.

[0119] b) Improving the evaluation accuracy: Deep learning algorithms are used for image recognition and feature recognition to accurately obtain key information in the building graphics, such as the accurate positions and dimensions of walls, stairwells, elevator shafts, etc., providing an accurate basis for area calculation and antenna point planning. For example, the number and position of elevator shafts are accurately identified through convolutional neural networks, providing accurate data for the antenna layout for elevator coverage. A refined cost estimation model is established, comprehensively considering various factors, such as the area of different regions, floor structure, in-building distribution point drawing information, etc., and using big data and machine learning algorithms to improve the prediction accuracy of investment costs and reduce errors.

[0120] c) Enhance business competitiveness: It can quickly and accurately provide pre-construction assessment information for the business line for the construction of in-building distribution systems, serving as support for project negotiation, helping the enterprise gain the upper hand in market competition, quickly plan projects, create differentiated competitive advantages, and attract more customers to cooperate. According to different scenarios (such as residential, commercial office buildings, factories, etc.) and customer needs, personalized and precise assessment solutions and high-quality services are provided to enhance the enterprise's core competitiveness in the field of wireless communication, expand market share, and promote the sustainable development of the business.

[0121] In a specific embodiment, the pre-construction assessment method for in-building distribution systems is applied to a pre-construction assessment system for in-building distribution systems. The corresponding architecture diagram of the system is as Figure 2 shown, including a data layer, a model layer, an application layer, and an interaction layer.

[0122] 1) Data layer: Responsible for collecting, storing, and managing various types of data, such as wireless signal data, CAD design floor plan data, geographic information data, etc. These data are the foundation for the operation of the entire system and provide data support for subsequent model training and applications. For example, CAD design floor plan data contains detailed information such as the structure and layout of the building and is a key data source for identifying indoor space information.

[0123] 2) Model layer: Includes a large model (i.e., an industry large model) and a deep learning model. The large model is used to optimize indicators such as the accuracy of area calculation, the accuracy of antenna point positions, and the prediction accuracy of investment costs; the deep learning model is used to implement functions such as image recognition and feature recognition, such as identifying special spaces such as walls, stairwells, and elevator shafts through convolutional neural networks, as well as identifying drawing legends, characteristics of enclosed spaces, and contour features. The model layer learns and trains on the data in the data layer, continuously optimizing model parameters to improve the accuracy and performance of the model.

[0124] 3) Application layer: Applies the models in the model layer to actual wireless planning to achieve functions such as automatic processing of drawings, automatic output of in-building distribution point maps, and intelligent investment assessment. The application layer works closely with the data layer and the model layer, inputs data into the model for processing, and integrates and applies the processing results.

[0125] 4) Interaction layer: Provides an interaction interface between the user and the system, including a visualization interface and a feedback mechanism. Users can upload CAD design floor plans, input query conditions, view results, and adjust parameters through the interaction layer. At the same time, the interaction layer transmits the user's operations and requirements to the application layer and the model layer, and displays the processing results to the user in an intuitive manner.

[0126] The working process of the system is as follows: First, the data layer collects and organizes relevant data; then, the model layer uses this data to train and optimize the model; next, the application layer calls the trained model to process the input data (such as CAD design floor plans), realizing functions such as automatically processing drawings, generating indoor distribution point maps, and conducting investment evaluations; finally, the interaction layer displays the processing results to the user and receives the user's feedback for further optimization and adjustment of the system.

[0127] Specifically, based on the above system, the corresponding evaluation method before building indoor distribution in a building can include the following steps:

[0128] 1. Automatically process drawings

[0129] (1) Drawing acquisition and preprocessing

[0130] Condition: The user needs to have the CAD design floor plan of the building and upload it to the system through the upload interface provided by the system. The uploaded drawings should meet certain format requirements to ensure that the system can correctly identify and process them.

[0131] Steps: The system automatically identifies the drawings of each building in the drawing set and splits them, and processes each building's drawings separately. When uploading the drawing set, the user needs to ensure the accuracy and integrity of the drawing set to avoid missing or incorrect drawings affecting subsequent processing. Preprocess the split CAD design floor plan (i.e., process the relevant factors affecting drawing recognition), including operations such as removing noise and correcting graphic deviations, to improve the quality and recognizability of the drawings. Then perform planar vectorization to convert the image file into vector data that can be processed by a convolutional neural network for subsequent recognition and calculation.

[0132] It should be noted that a building includes one or more buildings. By splitting the drawing set of the building, all the drawings of each building are obtained. Usually, each CAD design floor plan is the floor plan of a building (including all its floors), and each input to the convolutional neural network is the floor plan of a building.

[0133] (2) Identify indoor space information

[0134] Condition: A large amount of accurately labeled training set data is required for model training, including the labeled information of special spaces such as walls, stairwells, and elevator shafts. At the same time, a computing device with good performance and a deep learning framework (such as TensorFlow or PyTorch) are required to support the training and operation of the model.

[0135] Steps: Use a convolutional neural network (such as the U-Net model) to identify special spaces such as walls and stairwells. First, download a training set containing annotation information of walls, stairwells, etc. in a large number of different types of architectural drawings, and use this training set to train the model. During the training process, by adjusting the parameters of the model (such as the convolution kernel size, learning rate, etc.), the model can accurately identify these special spaces. After training, use the test set to evaluate the model, and the evaluation metrics include accuracy, precision, recall, and F1 score, etc. For example, accuracy is used to measure the proportion of samples predicted correctly by the model in the total samples; precision represents the proportion of the number of samples predicted as positive samples and actually being positive samples in the number of samples predicted as positive samples by the model; recall is the proportion of the number of samples predicted as positive samples and actually being positive samples in the number of samples actually being positive samples; the F1 score is the harmonic mean that comprehensively considers precision and recall. According to these evaluation metrics, if the model performance is poor (such as the accuracy is lower than the set threshold), then adjust the hyperparameters of the model (such as increasing the number of training epochs, adjusting the learning rate, etc.) or reselect the training set for training until the model performance meets the requirements.

[0136] Similarly, use a convolutional neural network to identify elevator shafts. Download a training set with elevator shaft annotation information to train the model, evaluate the model performance through the test set, and optimize the model according to the evaluation results so that the model can accurately identify the location of the elevator shaft and draw the outline, and at the same time automatically count the number of elevator shafts. Then, automatically read the floor information of this building. Since the drawings of each floor of the same building are drawn on the same.dwg drawing, the system can know the floor information by automatically identifying each floor's drawing. For elevator coverage, according to the standard that one antenna needs to be arranged every 3 floors, divide the number of floors by 3 (round up) to calculate the number of antennas required for elevator shaft coverage, that is, the total antenna number calculation formula is: total Among them, is the symbol for rounding up. Assuming a building has 8 floors and each floor has 3 elevator shafts, then the total number of antennas = 3×3 = 9 antennas.

[0137] (3) Processing of the drawings

[0138] Condition: Install a professional CAD software or have a corresponding drawing processing library to operate on the original CAD drawings.

[0139] Steps: Open the original CAD drawing, find the "Layer" or "Layer Properties" option in the CAD software, uncheck the unwanted layers so that the elements of these layers are no longer displayed in the base map. Then use the selection tool to select the elements to be deleted (such as extra lines, annotations, etc.), press the "Delete" key or use the "Clear" function to delete the selected elements to obtain the base map after the lines are removed. Finally, save the base map as a new CAD file for subsequent use. When calculating the area, first use edge recognition technology to identify the outer contour of the base map after the lines are removed, and outline the outer periphery of the overall area of ​​the floor (for example, with white lines). Then, according to the size ruler, draw lines (for example, with gray lines), divide the overall area into several closed spaces, and label each closed space. Then use OCR technology to identify the text in each closed area in the base map, identify the compartment type, correspond the text information of each compartment to each number, and output the overall floor information, including the type of each compartment, area, number of closed spaces and the outline of the overall floor plan. If the drawing itself contains area information, it is read directly; if not, the area of ​​the enclosed space is calculated by self-drawing to obtain the overall area. For example, the building area diagram can be as follows Figure 3 As shown, Figure 3 There are 25 encapsulated spaces in total, among which 1-22 are office areas, 16 includes the bathroom; 23 is the smoke exhaust shaft, weak current shaft, strong current shaft, cold coal pipe shaft, corridor and stairs; 24 is the shared vestibule, elevator, vestibule and stairs; 25 is the balcony. The total building area of ​​this floor is 824 square meters.

[0140] It should be noted that the most important thing to pay attention to when placing indoor substations is the overall external contour of the building, and some closed spaces in the drawings need to be avoided for placement. In order to improve the recognition accuracy, we deleted the auxiliary lines, annotation information and other dispensable elements on the original CAD drawings, and obtained a base map that only includes the external contours of each floor in the building and the contours of the internal closed spaces.

[0141] 2. Automatic output of indoor point map

[0142] (1) Point placement principle

[0143] The signal strength at the corresponding distance can be calculated based on the free space signal arrival strength formula:

[0144] Rss=Pt+Gt+Gr-Lc-Lbf

[0145] Among them, Rss is the signal arrival strength, the unit is dBm; Pt is the transmit power, the unit is dBm; Gt is the transmit antenna gain, the unit is dBi; Gr is the receive antenna gain, the unit is dBi; Lc is the loss of the cable and connector, the unit is dB; Lbf is the free space loss, the unit is dB.

[0146] The free space loss formula is as follows:

[0147] Lbf = 20lg(F) + 20lg(D) + 32.4

[0148] Where F is the signal frequency in MHz, and D is the propagation distance in km.

[0149] Under ideal conditions, antenna strength calculation tests are carried out for different scenarios to obtain the antenna spacing ranges for different scenarios, which serve as the theoretical basis for point placement. For example, in the office building scenario in an industrial park, in the case of a single-channel passive distribution system, the antenna spacing is usually 12 - 20 meters, and ceiling-mounted antennas are used; while in a residential community in the residential scenario, the antenna spacing is 50 - 150 meters, and spotlights or log-periodic antennas can be selected.

[0150] It should be noted that different scenarios refer to local space scenarios such as office buildings, factories, industrial parks, rooftops, underground parking lots, large shopping malls, residences, and elevators.

[0151] (2) Model establishment

[0152] Elevator scenario:

[0153] Conditions: Obtain accurate floor height information and a model (i.e., a large model) that can calculate the antenna placement position based on the floor height after training.

[0154] Steps: Calculate and place according to the floor height. Taking the principle of placing one antenna every about 10 meters (3 floors), round up to determine the number of antennas. Use the elevator-related data (such as elevator shaft position, number of floors, etc.) obtained from preliminary drawing processing to automatically label the elevators and generate a model, generating a room distribution point map for the elevator scenario. For example, the room distribution point map for the elevator scenario can be as Figure 4 shown, where the elevator shaft numbers (such as 1#, 2#, etc.) correspond to different elevator shafts. The DT1 - DT3 or DT1, DT2 marked under each elevator shaft represent the number and positions of the room distribution equipment installed in that elevator shaft.

[0155] It should be noted that the model can change the antenna layout interval by adjusting parameters, etc., and according to different scenarios, this parameter can also be determined using past experience. For example, in the elevator scenario, one antenna is arranged every 3 floors, but it can be changed.

[0156] (3) Model output, optimization, and application feedback

[0157] After the model processes the drawings one by one, it integrates all the drawings of a single station (referring to the entire building), arranges them according to the building plan, and finally outputs a complete point location drawing (i.e., marks the antenna point locations on the drawing). After the model training is completed, the cross-validation technique is used to divide the dataset into a training set, a validation set, and a test set. By evaluating the performance of the model (such as accuracy, mean squared error, etc.) under different hyperparameter combinations on the validation set, the optimal hyperparameter combination is selected to improve the accuracy and generalization ability of the model. The reinforcement learning algorithm can also be adopted to enable the model to automatically select the optimal model architecture and parameter settings during the interaction with the environment. In addition, in practical applications, the differences between the predicted results output by the model (such as antenna point locations, signal coverage strength, etc.) and the actual results (obtained through on-site testing or measurements after actual construction) are collected to provide feedback and adjustment to the model. If it is found that the model's predictions are inaccurate in certain scenarios (such as unreasonable antenna point locations under specific building structures resulting in poor signal coverage), the reasons are further analyzed, which may be insufficient training data, inappropriate model structure, etc. The model is improved accordingly, such as adding training data for relevant scenarios, adjusting the model structure, etc., to continuously improve the performance and applicability of the model.

[0158] It should be noted that the model finally outputs the indoor distribution point location map of each building, and at the same time also outputs the signal coverage strength prediction map of each floor of the building. The coverage strength is more estimated based on the location of the placement points.

[0159] 3. Intelligent investment evaluation

[0160] (1) Data collection and preprocessing

[0161] Condition: After successfully completing the steps of automatically processing the drawings, accurate CAD drawing vector data is obtained.

[0162] Steps: Directly read the CAD drawing vector data processed in the previous step (i.e., the indoor distribution point location map of a building) as the basic data for subsequent investment evaluation, ensuring the accuracy and integrity of the data, and avoiding evaluation result deviations caused by data missing or errors.

[0163] (2) Automatically obtain the areas of each region of the building

[0164] Condition: A deep learning model that can accurately identify and extract key information such as walls and doors and windows is trained, and there is sufficient computing resources to run the model.

[0165] Steps: Automatically read vector data using a trained deep learning model, understand the characteristics of each area by identifying and extracting key information such as walls, doors, and windows, and thus automatically calculate the areas of each area in the building. For example, the model can accurately distinguish different types of rooms (such as offices, bathrooms, stairwells, etc.), calculate the area of each room, and summarize the area data of each area in the entire building.

[0166] (3) Establish a cost estimation model:

[0167] Investment cost estimation is mainly used for the cost assessment of site construction, which is mainly divided into main equipment investment and distribution system investment. For self-built scenarios, it can effectively evaluate the cost-effectiveness, and for the owner investment scenario (the distribution system is invested by the developer), it can effectively balance the interests of both parties.

[0168] Distribution system investment:

[0169] Based on the antenna point data output from the indoor distribution point map, preliminary calculations are carried out according to the 451 quota. Assume that the construction unit price of indoor antennas is X 1 (total construction cost of a single antenna, including corresponding materials), and the construction unit price of outdoor antennas is X 2 Then the total construction cost S 1 Should be:

[0170] S 1 = X 1 *(K 1 + K 2 + K 3 ) + X 2 * K 4

[0171] Among them, K 1 、K 2 、K 2 、K 3 Are pre-set parameters, which are specific numerical values and are set differently according to different scenarios.

[0172] It should be noted that for the entire building, the main equipment investment and distribution system investment can be calculated separately for each building first, and then the cost estimation of the entire building can be obtained.

[0173] (4) Model optimization, application, and feedback adjustment:

[0174] Use cross-validation technology to optimize the model hyperparameters, and knowledge distillation and transfer learning can also be used to improve the generalization ability. Collect the differences between the estimation results and the actual results and feedback to adjust the model.

[0175] 4. Information visualization

[0176] User interface design: Such as Figure 5As shown, a clear and intuitive visualization interface can be provided to display the wireless signal coverage and cost estimation results in the form of charts or graphs. The interface includes components such as text boxes and drop-down menus, facilitating users to input query conditions, view results, and adjust parameters. Meanwhile, user authentication and permission management functions are implemented to ensure the security and stability of the system. It communicates with the backend service through the HTTP (Hyper Text Transfer Protocol) protocol to ensure the coordinated operation of the front and back ends. For example, the schematic diagram of the wireless signal coverage can be as Figure 6 shown, where different colors or texts can be used to mark areas with good signal and areas with poor signal.

[0177] In another specific embodiment, taking a commercial office building as an example, first, the user uploads the CAD design floor plan of the office building to the data layer of the system and performs steps to automate the processing of the drawings. The system automatically splits the drawing set, preprocesses and vectorizes the drawings. Through a convolutional neural network model, special spatial information such as walls, stairwells, and elevator shafts can be accurately identified, the floor information is automatically read, and the number of antennas required for elevator shaft coverage is calculated. When processing the drawings, a base map with unnecessary lines removed is obtained, and information such as the area of each region (i.e., enclosed space) and partition type is calculated.

[0178] Next, the step of automatically outputting the indoor distribution point diagram is executed. According to the office building scenario, based on an antenna spacing of about 25 meters (assumed) and corresponding placement principles, first, antenna points are placed on the contour map (i.e., the base map). After determining the corner positioning (assumed), other points are automatically placed, and then mapped to the original drawing. The problem of antenna-wall coincidence is processed through an antenna position optimization model, and then signal simulation optimization is carried out. According to the field strength calculation and display results, if it is found that the signal coverage does not meet the expectations (such as the proportion above -90dbm is less than 95%), the antenna spacing is automatically reduced and re-placed until the requirements are met, and finally a complete point diagram is output.

[0179] In the step of intelligent investment evaluation, the system automatically obtains the area data of each area of the building. Based on the cost estimation model, combined with information such as antenna point data and equipment unit price, the distributed system investment and main equipment investment are calculated respectively (assuming the RRU + fiber optic repeater method is adopted, and the number of RRU, proximal and distal ends of the fiber optic repeater is calculated according to the actual number of antennas, and then the main equipment cost is obtained), and finally the total cost is obtained.

[0180] The user inputs query conditions (such as viewing the signal coverage and cost estimation of a certain floor) through the visualization interface of the interaction layer, and the interface displays the results in an intuitive chart. Meanwhile, during the operation of the system, the differences between the model output results and the actual situation are continuously collected and fed back to the model layer to optimize and adjust the model to improve the system performance and evaluation accuracy.

[0181] It should be noted that the evaluation method before the construction of in-building distribution provided by the present invention has the following characteristics:

[0182] 1. Integrated application of large model and deep learning technology

[0183] Utilize deep learning technology, especially convolutional neural network, to perform precise image recognition and feature recognition on CAD design floor plans. For example, accurately identify special spaces such as walls, stairwells, elevator shafts, as well as various drawing legends, characteristics of enclosed spaces, and contour features. This is the basis for subsequent accurate area calculation, antenna point positioning, and investment cost evaluation. Through the training of a large amount of labeled data, the model can learn the characteristic patterns of various elements in the architectural drawings, thereby achieving automated precise recognition.

[0184] The application of the large model optimizes key indicators such as the accuracy rate of area calculation, the accuracy of antenna point positioning, and the prediction accuracy of investment costs. By learning a large amount of data and processing complex algorithms, the large model can calculate the areas of different regions in the building more accurately, reasonably plan the antenna points, accurately predict the investment costs, and improve the accuracy and reliability of the entire evaluation system.

[0185] 2. Scenario-based precise prediction and evaluation model

[0186] Establish targeted models and deployment principles according to different scenarios (such as residential buildings, commercial office buildings, factories, elevators, underground parking lots, etc.). In the step of automated output of in-building distribution point maps, determine the appropriate antenna type, spacing, and deployment location according to the characteristics of each scenario, such as building structure, spatial layout, signal propagation characteristics, etc. For example, in the factory building scenario of an industrial park, considering its large open area and high floor height, use corner wall-mounted antennas and deploy them according to area division; in the residential scenario, determine different antenna spacings and types for different types of residences (such as residential communities, apartments) to achieve the best signal coverage effect.

[0187] In the intelligent investment evaluation step, combine various factors such as area, floor structure, and in-building distribution point map information in different scenarios to establish a refined cost estimation model. For the investment in main equipment (RRU + fiber optic repeater or RRU) and distribution system investment, respectively consider factors such as equipment carrying capacity and power distribution in different scenarios, accurately calculate the investment costs, and provide strong support for project decision-making.

[0188] 3. Automated processing and intelligent optimization process

[0189] An automated processing flow from drawing acquisition to the output of the final evaluation result is realized. During the automated processing of drawings, the system automatically identifies and splits the drawing set, performs preprocessing, vectorization, indoor space information recognition, and drawing optimization processing, quickly and accurately obtains building information, reduces manual intervention, and improves processing efficiency and accuracy.

[0190] Introduce an intelligent optimization mechanism, such as antenna position optimization and signal simulation optimization in the automated output step of the indoor distribution point map. The machine learning model is used to identify the coincidence situation between the antenna and the wall and automatically adjust the antenna position, and the antenna is automatically adjusted according to the field strength calculation and simulation results of the signal penetrating the medium, continuously optimizing the signal coverage effect to ensure meeting different requirements. At the same time, in the intelligent investment evaluation module, by collecting the differences between the actual results and the estimated results, the model is feedback-adjusted to improve the accuracy and applicability of the model.

[0191] The evaluation method for pre-construction of building indoor distribution provided by the embodiments of the present invention first obtains the building computer-aided design (CAD) design floor plan drawing set; splits the drawing set to obtain the CAD design floor plans of each building; then, according to the CAD design floor plans of each building, uses a pre-trained convolutional neural network model to identify the special space information of each building; obtains the base map of the CAD design floor plan with the identified special space information, further divides the base map into several closed spaces, and calculates the area of each closed space to obtain the processed base map; then, based on the processed base map, uses a preset large model to generate the indoor distribution point map corresponding to each building and the signal coverage intensity prediction map for each floor according to the corresponding scenarios of each building and the preset point placement principle; finally, calculates the main equipment investment and distribution system investment of each building according to the positions of the antenna points placed in the indoor distribution point map corresponding to each building, and summarizes to obtain the overall investment cost estimation result of the building. The present invention significantly improves the efficiency and accuracy of the pre-construction evaluation of building indoor distribution by automatically processing the building CAD design floor plan, accurately identifying special space information using a convolutional neural network model, and combining the indoor distribution point map and signal coverage intensity prediction map generated by the large model. This solution can quickly provide accurate evaluation information, greatly enhance business competitiveness, and solve the problems of low evaluation efficiency, insufficient accuracy, and lack of business competitiveness in the existing pre-construction evaluation methods for building indoor distribution.

[0192] Embodiment 2:

[0193] As Figure 7 shown, this embodiment provides an evaluation device for pre-construction of building indoor distribution, which is used to execute the above-mentioned evaluation method for pre-construction of building indoor distribution, and includes:

[0194] A drawing set acquisition module 11, which is used to acquire the building computer-aided design (CAD) design floor plan drawing set;

[0195] The automatic splitting module 12, connected to the drawing set acquisition module 11, is used to split the drawing set to obtain the CAD design floor plans of each building;

[0196] The special space recognition module 13, connected to the automatic splitting module 12, is used to recognize the special space information of each building according to the CAD design floor plans of each building by using a pre-trained convolutional neural network model;

[0197] The base map processing module 14, connected to the special space recognition module 13, is used to obtain the base map of the CAD design floor plan with the recognized special space information, further divide the base map into several closed spaces, and calculate the area of each closed space to obtain the processed base map;

[0198] The point map generation module 15, connected to the base map processing module 14, is used to generate the indoor distribution point map corresponding to each building and the signal coverage intensity prediction map of each floor based on the processed base map by using a preset large model according to the scenes corresponding to each building and the preset point placement principle;

[0199] The investment cost estimation module 16, connected to the point map generation module 15, is used to calculate the main equipment investment and distribution system investment of each building according to the positions of the antenna point placements in the indoor distribution point map corresponding to each building, and summarize to obtain the overall investment cost estimation result of the building.

[0200] Optionally, the special space recognition module 13 includes:

[0201] A preprocessing unit, used to preprocess the CAD design floor plans of each building;

[0202] A vectorization unit, used to convert the preprocessed CAD design floor plan into vectorized vector data;

[0203] An identification unit, used to input the vector data into the trained convolutional neural network model to identify the floor information of each building and the information of the special space;

[0204] Among them, the special space includes at least one of the following: wall, stairwell, elevator shaft.

[0205] Optionally, the base map processing module 14 includes:

[0206] An element deletion unit, used to delete the auxiliary lines and annotation information from the CAD design floor plan with the recognized special space information to obtain the corresponding base map;

[0207] An outer contour recognition unit, configured to recognize the external contour of the base map by using edge recognition technology, and outline the periphery of the overall area of each floor;

[0208] A closed space division unit, configured to perform line tracing based on the periphery of the overall area of each floor according to a dimension scale, and divide the base map into several closed spaces;

[0209] An area calculation unit, configured to recognize the compartment type of each closed space by using optical character recognition (OCR) technology, calculate the area of each closed space, and obtain a processed base map.

[0210] Optionally, the device further includes:

[0211] A spacing range acquisition module, configured to perform antenna intensity calculation tests on different scenarios according to the free space signal arrival strength formula, and obtain the antenna spacing ranges in different scenarios;

[0212] A point placement basis acquisition module, configured to use the antenna spacing ranges in different scenarios as the basis for the point placement principle in the corresponding scenarios.

[0213] Optionally, the point map generation module 15 is specifically configured to:

[0214] For each building, input the processed base map into a preset large model, and use the large model to perform the following steps: perform antenna point placement on the processed base map according to the floor height of the building and the point placement principle of the corresponding scenario, obtain an in-building distribution point map with signal coverage reaching the expectation, and obtain a signal coverage intensity prediction map for each floor according to the positions of the antenna point placements in the in-building distribution point map.

[0215] Optionally, the investment cost estimation module 16 includes:

[0216] A building investment calculation unit, configured to calculate the main equipment investment and distribution system investment of each building respectively by using a preset cost estimation model according to the positions of the antenna point placements in the in-building distribution point map corresponding to each building and the areas of each closed space;

[0217] A building investment summary unit, configured to summarize the main equipment investment and distribution system investment of all buildings to obtain the estimation result of the overall investment cost of the building.

[0218] Optionally, the device further includes:

[0219] A visualization module, configured to display at least one of the following according to the user's query conditions: the in-building distribution point map corresponding to each building, the signal coverage intensity prediction map of any floor in a certain building, and the estimation result of the overall investment cost of the building.

[0220] Embodiment 3:

[0221] Reference Figure 8 , this embodiment provides an evaluation device before building a building in-building distribution system, including a memory 21 and a processor 22. A computer program is stored in the memory 21, and the processor 22 is configured to run the computer program to execute the evaluation method before building a building in-building distribution system in Embodiment 1.

[0222] Among them, the memory 21 is connected to the processor 22. The memory 21 can adopt flash memory, read-only memory or other memories, and the processor 22 can adopt a central processing unit or a single-chip microcomputer.

[0223] Embodiment 4:

[0224] This embodiment provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, it implements the evaluation method before building a building in-building distribution system in the above Embodiment 1.

[0225] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program modules or other data). The computer-readable storage medium includes, but is not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), flash memory or other memory technologies, CD-ROM (Compact Disc Read-Only Memory), digital versatile disc (DVD) or other optical disc storage, magnetic cassette, tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0226] In summary, the evaluation method, device and readable storage medium provided by the embodiments of the present invention for the construction of in-building distributed antenna systems first obtain the building computer-aided design (CAD) design floor plan drawing set; and split the drawing set to obtain the CAD design floor plans of each building; then, according to the CAD design floor plans of each building, use a pre-trained convolutional neural network model to identify the special space information of each building; and obtain the base map of the CAD design floor plan with the special space information identified, further divide the base map into several closed spaces, and calculate the area of each closed space to obtain the processed base map; then, based on the processed base map, use a preset large model to generate the in-building distributed antenna system point map corresponding to each building and the signal coverage intensity prediction map for each floor according to the scenarios corresponding to each building and the preset point placement principle; finally, calculate the main equipment investment and distribution system investment of each building according to the positions of the antenna points placed in the in-building distributed antenna system point map corresponding to each building, and summarize to obtain the estimated result of the overall investment cost of the building. The present invention significantly improves the efficiency and accuracy of the evaluation before the construction of in-building distributed antenna systems by automatically processing the building CAD design floor plans, accurately identifying special space information using a convolutional neural network model, and combining the in-building distributed antenna system point map and signal coverage intensity prediction map generated by the large model. This solution can quickly provide accurate evaluation information, greatly enhancing business competitiveness, and solving the problems of low evaluation efficiency, insufficient accuracy, and lack of business competitiveness in the existing evaluation methods for the construction of in-building distributed antenna systems.

[0227] It can be understood that the above embodiments are merely exemplary embodiments adopted to illustrate the principle of the present invention, and the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered within the protection scope of the present invention.

Claims

1. A building evaluation method before construction, characterized in that: The method comprises: Get a set of building computer-aided design (CAD) design floor plan drawings; The drawing set is split to obtain the CAD design plan drawings of each building; According to the CAD design plan of each building, the pre-trained convolutional neural network model is used to identify the special spatial information of each building; Acquire a base map of a CAD design plan drawing in which special spatial information has been identified, further divide the base map into a plurality of closed spaces, and calculate the area of ​​each closed space to obtain a processed base map; Based on the processed base map, the preset large model is used to generate the indoor point map corresponding to each building and the signal coverage strength prediction map of each floor according to the corresponding scenes of each building and the preset point layout principle; The main equipment investment and distribution system investment of each building are calculated according to the positions of the antenna points in the indoor distribution point map corresponding to each building, and the overall investment cost estimation result of the building is summarized.

2. The method according to claim 1, characterized in that The method uses a pre-trained convolutional neural network model to identify the special spatial information of each building based on the CAD design plan of each building, specifically including: Pre-process the CAD design drawings of each building; Convert the pre-processed CAD design plan into vectorized vector data; Inputting the vector data into a trained convolutional neural network model to identify the floor information of each building and the information of the special space; Wherein, the special space includes at least one of the following: a wall, a stairwell, and an elevator shaft.

3. The method according to claim 1, characterized in that The step of obtaining a base map of a CAD design plan view in which special spatial information has been identified, further dividing the base map into a plurality of closed spaces, and calculating the area of ​​each closed space to obtain a processed base map specifically includes: Deleting auxiliary lines and annotation information from the CAD design plan view in which special spatial information has been identified, and obtaining a corresponding base map; Using edge recognition technology to identify the outer contour of the base map, outlining the overall area perimeter of each floor; Based on the overall area perimeter of each floor, the base map is divided into a number of closed spaces by drawing lines according to a dimension ruler; Optical character recognition (OCR) technology is used to identify the compartment type of each enclosed space, and the area of ​​each enclosed space is calculated to obtain a processed base map.

4. The method according to claim 1, characterized in that: Before generating the indoor point map corresponding to each building and the signal coverage strength prediction map of each floor based on the processed base map and using the preset large model according to the scene corresponding to each building and the preset point layout principle, the method further includes: The antenna strength calculation test is performed for different scenarios according to the free space signal arrival strength formula to obtain the antenna spacing range in different scenarios; The antenna spacing range in different scenarios is used as the basis for the point deployment principle in the corresponding scenarios.

5. The method according to claim 4, characterized in that Based on the processed base map, the preset large model is used to generate the indoor point map corresponding to each building and the signal coverage strength prediction map of each floor according to the scene corresponding to each building and the preset point layout principle, which specifically includes: For each of the buildings, the processed base map is input into a preset large model, and the large model is used to perform the following steps: antenna points are placed on the processed base map according to the floor height of the building and the point placement principle of the corresponding scene, so as to obtain a room distribution point map in which the signal coverage can achieve the expected level, and a signal coverage strength prediction map for each floor is obtained according to the positions of the antenna point placements in the room distribution point map.

6. The method according to claim 1, characterized in that The main equipment investment and distribution system investment of each building are calculated according to the positions of the antenna points in the indoor distribution point map corresponding to each building, and the overall investment cost estimation result of the building is obtained by summarizing, specifically including: According to the location of antenna points in the indoor distribution point map corresponding to each building, as well as the area of ​​each enclosed space, the main equipment investment and distribution system investment of each building are calculated using the preset cost estimation model; The main equipment investment and distribution system investment of all buildings are summarized to obtain the overall investment cost estimation result of the building.

7. The method according to claim 1, characterized in that The method further comprises: According to the user's query conditions, at least one of the following is displayed: a room distribution point map corresponding to each building, a signal coverage strength prediction map of any floor in a building, and an estimation result of the overall investment cost of the building.

8. A building evaluation device before construction, characterized in that: The device comprises: A drawing set acquisition module is used to acquire a building computer-aided design (CAD) design plan drawing set; An automatic splitting module, connected to the drawing set acquisition module, is used to split the drawing set to obtain the CAD design plan drawings of each building; A special space identification module is connected to the automatic splitting module and is used to identify the special space information of each building based on the CAD design plan of each building using a pre-trained convolutional neural network model; A base map processing module is connected to the special space identification module, and is used to obtain a base map of the CAD design plan view in which special space information has been identified, further divide the base map into a plurality of closed spaces, and calculate the area of ​​each closed space to obtain a processed base map; A point map generation module is connected to the base map processing module and is used to generate a room point map corresponding to each building and a signal coverage strength prediction map for each floor based on the processed base map and the preset large model according to the scene corresponding to each building and the preset point layout principle; The investment cost estimation module is connected to the point map generation module and is used to calculate the main equipment investment and distribution system investment of each building according to the positions of the antenna points in the indoor point map corresponding to each building, and summarize the overall investment cost estimation result of the building.

9. A building evaluation device before construction, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to implement the pre-construction assessment method for building compartments as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the building room evaluation method before construction is implemented as described in any one of claims 1-7.

Images