An antenna beam visualization intelligent agent interaction display method and system

Abstract

This application discloses an antenna beam visualization intelligent interactive display method and system, belonging to the interdisciplinary field of communication antenna marketing and intelligent visualization technology. The system adopts a three-layer architecture: infrastructure layer, core technology layer, and application display layer. The core technology layer integrates five major modules, including beam visualization and digital twin scene construction. It achieves precise beam visualization through technologies such as cone parametric modeling, scene editing, and skeleton binding animation. Combined with multimodal AI interaction and a structured knowledge base, it realizes a closed loop of technical parameters, scene effects, and intelligent interaction. This solves the problems of abstract beam effects and lack of scene linkage in traditional antenna marketing, improves customer cognitive efficiency, reduces marketing reliance on professional personnel, and supports global market expansion.

Description

Technical Field

[0001] This application relates to the intersection of communication antenna marketing and intelligent visualization technology, and in particular to an interactive display method and system for antenna beam visualization intelligent agents. Background Technology

[0002] The core competitiveness of communication antenna products is typically reflected in technical indicators such as beam coverage characteristics and multi-scenario adaptability. Especially in complex communication scenarios such as high-speed rail, tunnels, and mountain viaducts, the coverage range, propagation direction, and scenario adaptability of antenna products directly affect their application effectiveness. As the marketing methods in the communications industry gradually shift towards digitalization and intelligence, using 3D displays, digital twin displays, and intelligent interactive methods to demonstrate product performance has become an important development direction for related product marketing and presentation.

[0003] Currently, the marketing and display methods for antenna products mainly rely on paper brochures, static presentations, 3D product models, and general digital showroom displays. While related technologies include multimedia interactive displays, virtual showroom displays, 3D engine-based product displays, and digital twin exhibition displays for product model display, scene construction, and basic interaction, these solutions typically focus on general product display, virtual showroom construction, or basic 3D interaction. They lack a dedicated display mechanism for the beam parameter characteristics of communication antenna products, and have not yet formed an integrated technical solution for beam coverage effects, scene-linked displays, and intelligent call-to-action displays.

[0004] Under the aforementioned technical conditions, the current marketing and demonstration of antenna products still suffers from the following problems: First, beam parameters are usually presented in the form of textual descriptions or static parameters, making it difficult to intuitively reflect the beam coverage and propagation effect; second, existing demonstration methods lack effective linkage between scenarios, solutions, products, and beam effects, making it difficult to present corresponding beam coverage characteristics according to different application scenarios; third, existing intelligent interaction methods mostly remain at the level of general question and answer, lacking identification and linkage mechanisms for antenna product demonstration and call requirements; fourth, the beam parameters, scenario adaptation cases, and related technical information of antenna products are relatively scattered, making it difficult to provide unified data support for marketing and demonstration, etc.

[0005] Therefore, how to construct a marketing and display solution for antenna products that can achieve beam parameter visualization, precise scene linkage, and multimodal intelligent interaction has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the problems of abstract and difficult-to-understand beam effects and the lack of linkage mechanisms between scenarios, products, and beams in existing antenna product marketing, this application provides a method and system for interactive display of antenna beam visualization intelligent agents. The technical solution is as follows: According to one aspect of this application, an antenna beam visualization intelligent agent interactive display system is provided, characterized in that the system adopts a three-layer architecture design, including a facility layer, a core technology layer and an application display layer, with each layer working together to realize antenna beam visualization and intelligent agent display; The facility layer provides hardware and computing power support for system operation, including 3D rendering computing power resources, large model inference computing power resources, data storage resources, network communication resources, as well as front-end display terminals, directional microphones, and voice pickup devices. The facility layer also supports local independent deployment of the system. The core technology layer is the core function implementation layer of the system, including a beam visualization module, a digital twin scene construction module, a multimodal AI interaction module, a structured knowledge base module, and a message collaboration module. These modules work together to achieve antenna beam parameter modeling, digital twin scene construction, intelligent interaction, and system linkage control. Specifically, the beam visualization module is configured to realize 3D visualization modeling and rendering of the antenna beam; the digital twin scene construction module is configured to construct a digital twin 3D scene of the antenna deployment environment and bind the antenna installation pose to the scene's world coordinate system; the multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition; the structured knowledge base module is configured to store structured data related to the antenna product; and the message collaboration module is configured to realize linkage communication between the modules of the core technology layer and between the core technology layer and the application display layer. The application display layer is the system's visual interaction entry layer, including a digital twin scene roaming module, a beam visualization display module, a 3D product details module, an AI intelligent question module, and a product matrix navigation module; wherein, the AI ​​intelligent question module is configured to provide intelligent question-and-answer interaction functions, and the AI ​​intelligent question module has the function of collecting all dialogue questions.

[0007] Optionally, the beam visualization module is configured to achieve three-dimensional visualization modeling and rendering of the antenna beam in the following manner: Using a cone as the core basic geometry, a quantitative mapping relationship between antenna beam parameters and cone geometric parameters is established, and beam parameterization modeling is completed. The cone model is converted into an editable polygon model, and customized scene editing is performed on the bottom coverage area of ​​the beam according to the contour features of different application scenarios. Configure skeletal rigging animation for the edited beam model to achieve beam tracking display, and use antenna tracking parameters as the animation driving source to achieve precise linkage between animation and antenna tracking actions; A voxel grid is superimposed within the cone-shaped coverage space, and a quantitative mapping relationship between beam gain and voxel density is established through a voxel density gradient algorithm, thereby achieving a refined expression of beam coverage intensity. Through a full-process rendering process including geometry optimization, texture mapping, and lighting calculation, a high-fidelity 3D beam coverage effect model is generated.

[0008] Optionally, the digital twin scene construction module is configured to construct a digital twin 3D scene and bind the antenna installation pose to the scene's world coordinate system in the following manner: A virtual city base was constructed, along with three core sub-scenes: high-speed rail plains, mountain viaducts, and tunnels, to recreate the actual antenna deployment environment at a 1:1 scale. An internal antenna installation pose calibration algorithm is provided, which enables precise binding of the antenna installation pose to the scene's world coordinate system. The system pre-defines the fusion rules between the beam model and the scene environment, establishes and stores an association mapping library of scenes and products, beam parameters, scene calibration coefficients, and editing rules, and automatically matches parameters and rules when switching scenes based on the association mapping library.

[0009] Optionally, the multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition in the following ways: It integrates an ASR speech recognition unit, a TTS speech synthesis unit, and a large model inference engine, supporting multimodal interaction with Chinese and English text input and voice input; Customized training is conducted based on keywords related to the scenarios, solutions, beam parameters, and performance indicators of antenna products to optimize the accuracy of intent recognition; The system incorporates a standardized intent recognition process, which enables accurate differentiation between knowledge-based question-and-answer requests and display / call requests. This standardized intent recognition process includes three core steps: keyword extraction, feature matching, and request type determination.

[0010] Optionally, the structured knowledge base module is used to build an enterprise knowledge base that combines vectorization and structure, integrating antenna product beam parameters, scene adaptation cases, technical documents and solution description information, while storing technical data such as scene calibration coefficients, editable polygon editing rules, skeleton binding animation parameters and voxel mesh mapping rules, and organizing them according to a four-level structure of scene, solution, product and beam parameters, to provide data support for AI intent recognition and beam parameter calling.

[0011] Optionally, the message collaboration module implements real-time communication between the front-end display layer and the core technology layer based on the WebSocket protocol. It is used to receive display call instructions issued by the multimodal AI interaction module, send scene jump instructions to the digital twin scene construction module, and send beam rendering update instructions to the beam visualization module, so as to realize the synchronous execution between AI interaction, scene switching and beam rendering.

[0012] On the other hand, a method for implementing interactive display of antenna beam visualization intelligent agents is provided, based on the aforementioned antenna beam visualization intelligent agent interactive display system, including: Step 101: Construct an association system for antenna products, beam parameters, and scenarios; collect structured data of antenna products; establish a four-level association data structure for products, beam parameters, solutions, and scenarios; and construct a digital twin scenario system and preset scenario roaming paths. Step 102: Implement 3D visualization rendering and animation of antenna beam parameters, including parametric modeling of the cone, scene editing of editable polygons, skeleton binding and beam tracking animation, voxel mesh overlay intensity expression, beam and scene fusion, and scene switching linkage. Step 103: Construct a multimodal AI interaction system to perform multimodal recognition and intent determination on user input, and process knowledge-based question-and-answer requirements and display-and-call requirements separately; Step 104: Implement product matrix navigation and multilingual adaptation, and perform statistics and analysis based on user interaction behavior data.

[0013] Optionally, the mapping relationship for the parametric modeling of the cone in step 102 is as follows: Horizontal cone angle = Antenna horizontal coverage angle × Scene calibration coefficient k1; Vertical cone angle = antenna vertical coverage angle × scene calibration coefficient k2; Cone height = maximum beam propagation distance.

[0014] Optionally, the skeleton rigging beam tracking animation in step 102 includes: Arrange a main skeleton node along the central axis of the beam model; Arrange multiple child bone nodes at the vertices of the polygon covering the bottom area and establish a parent-child hierarchical relationship; Weight binding is applied between the polygonal faces of the beam model and the bone nodes; Using the antenna beam tracking angle as the animation driver, the main skeleton node drives the sub-skeleton nodes to move, thereby achieving beam tracking animation.

[0015] Optionally, the intent recognition process in step 103 includes: Extract core keywords from user-input text and filter out stop words; The extracted keywords are matched with the feature word library in the knowledge base to determine the relevance of the requirements. When identified as a parameter query, technical document consultation, or performance indicator query, it is determined to be a knowledge-based question and answer request. When identified as scene display, beam effect viewing, solution visualization, or beam tracking animation demonstration, it is determined to be a display call type requirement.

[0016] This invention discloses an antenna beam visualization intelligent interactive display system and its implementation method. The system adopts a three-layer architecture: infrastructure layer, core technology layer, and application display layer. The core technology layer integrates five modules, including beam visualization and digital twin scene construction. It achieves precise beam visualization through technologies such as conical parametric modeling, scene-based editing, and skeletal animation binding. Combined with multimodal AI interaction and a structured knowledge base, it realizes a closed loop from technical parameters and scene effects to intelligent interaction. This system solves the problems of abstract beam effects and lack of scene linkage in traditional antenna marketing, improves customer cognitive efficiency, reduces marketing reliance on professional personnel, and supports global market expansion. Attached Figure Description

[0017] Figure 1 This is a three-layer architecture diagram of the intelligent agent display system of the present invention; Figure 2 A schematic diagram of the parametric modeling and editable polygon editing process for antenna beam cones; Figure 3 Logic diagram for beam model skeleton binding and tracking animation implementation; Figure 4 A schematic diagram of the composite structure of a beam model cone and a voxel mesh; Figure 5 This is a schematic diagram illustrating the fusion effect of a digital twin scene and beam visualization, using a tunnel scene as an example. Figure 6 This is a schematic diagram illustrating the fusion effect of a digital twin scene and beam visualization, using a plain scene as an example. Figure 7 A flowchart of the multimodal AI interaction workflow; Figure 8 A schematic diagram of a four-level navigation system consisting of scenarios, solutions, products, and beams; Figure 9 This is a flowchart illustrating the overall steps of the method for implementing the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0019] In this article, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship. Example 1

[0020] According to one aspect of this application, an antenna beam visualization intelligent agent interactive display system is provided. The system adopts a three-layer architecture design, including a facility layer, a core technology layer, and an application display layer. Each layer works together to realize antenna beam visualization and intelligent agent display.

[0021] The infrastructure layer provides hardware and computing power support for system operation, including 3D rendering computing power resources, large model inference computing power resources, data storage resources, network communication resources, as well as front-end display terminals, directional microphones, and voice pickup devices. The infrastructure layer also supports local independent deployment of the system.

[0022] Specifically, the hardware and computing power support at the infrastructure layer includes 3D rendering computing power resources to ensure real-time rendering of beam models, large model inference computing power resources to support intent recognition calculations for multimodal AI interactions, data storage resources to store product data, scene data, knowledge base data, etc., and network communication resources to ensure communication and transmission between modules and between the front-end and back-end.

[0023] The core technology layer is the core function implementation layer of the system, including the beam visualization module, the digital twin scene construction module, the multimodal AI interaction module, the structured knowledge base module, and the message collaboration module. These modules work together to realize antenna beam parameter modeling, digital twin scene construction, intelligent interaction, and system linkage control.

[0024] Among them, the beam visualization module is configured to realize the three-dimensional visualization modeling and rendering of antenna beams; the digital twin scene construction module is configured to construct a digital twin three-dimensional scene of the antenna deployment environment and realize the binding of the antenna installation pose with the scene world coordinate system; the multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition; the structured knowledge base module is configured to store structured data related to antenna products; and the message collaboration module is configured to realize the linkage communication between various modules of the core technology layer and between the core technology layer and the application display layer.

[0025] The application presentation layer serves as the system's visual interaction entry layer, including a digital twin scene roaming module, a beam visualization display module, a 3D product details module, an AI-powered inquiry module, and a product matrix navigation module.

[0026] The AI-powered question-and-answer module is configured to provide intelligent question-and-answer interaction functions and has the ability to collect a full range of dialogue questions.

[0027] The front-end display terminal can use 4K touch all-in-one machine, PAD, monitoring large screen, etc., and directional microphones and voice pickup devices are used to accurately collect user voice input. The localized independent deployment mode can meet the needs of special marketing scenarios such as exhibitions without network.

[0028] The modules in the core technology layer have clear divisions of labor and work together, as detailed below.

[0029] The beam visualization module focuses on the 3D modeling and rendering of antenna beams, transforming abstract beam parameters into intuitive visualizations; the digital twin scene construction module builds virtual scenes consistent with the actual environment, providing a realistic environmental carrier for beam display; the multimodal AI interaction module builds an intelligent interaction bridge between users and the system, supporting Chinese and English text and voice input; the structured knowledge base module provides data support for the system, ensuring the accuracy of interactive responses and beam modeling; and the message collaboration module acts as a linkage hub, ensuring real-time communication and collaborative operation between various modules. The application presentation layer provides users with diverse interactive entry points: the digital twin scene roaming module allows users to autonomously roam virtual scenes and explore antenna deployment and beam coverage in different scenarios; the beam visualization display module loads and displays the 3D beam coverage effect generated by the beam visualization module; the 3D product details module displays the 3D model and detailed parameters of antenna products; the AI ​​intelligent question module provides a human-like intelligent question-and-answer function, allowing users to inquire about beam parameters, scene adaptation, and other issues. This module also collects all dialogue questions to provide data support for knowledge base iteration and product optimization; and the product matrix navigation module enables quick navigation and related display of scenes, solutions, products, and beam parameters.

[0030] like Figure 1 The diagram shows the three-layer architecture of the intelligent agent display system of the present invention. The diagram shows the hierarchical structure of the facility layer, the core technology layer, and the application display layer, as well as the core components contained in each layer. The layers work together from top to bottom to form a complete technical system, ensuring the stable operation and functional realization of the system.

[0031] As can be seen, the rational design of the three-layer architecture achieves layered collaboration between hardware support, core functions, and interactive entry points. Each module performs its own function and works closely together, breaking through the limitations of traditional antenna marketing display systems that are single in function and loose in structure. The localized deployment capability of the facility layer meets the usage needs of different marketing scenarios, the modular design of the core technology layer improves the system's scalability and maintainability, and the diversified interactive entry points of the application display layer enhance the user experience. The overall architecture provides stable and efficient technical support for antenna beam visualization and intelligent marketing display. Example 2

[0032] Furthermore, the beam visualization module is configured to achieve 3D visualization modeling and rendering of the antenna beam in the following ways.

[0033] Using a cone as the core geometric model, a quantitative mapping relationship between antenna beam parameters and cone geometric parameters is established, and beam parametric modeling is completed. The cone model is converted into an editable polygon model, and customized scene editing is performed on the bottom coverage area of ​​the beam according to the contour features of different application scenarios. Skeletal binding animation is configured for the edited beam model to achieve beam tracking display, and the antenna tracking parameters are used as the animation driving source to achieve precise linkage between animation and antenna tracking actions. Voxel meshes are superimposed in the cone coverage space, and a quantitative mapping relationship between beam gain and voxel density is established through a voxel density gradient algorithm to achieve a refined expression of beam coverage intensity. Through a full-process rendering process of geometry optimization, texture mapping, and lighting calculation, a high-fidelity 3D beam coverage effect model is generated.

[0034] The cone is chosen as the core geometric shape because its spatial form closely matches the propagation characteristics of the antenna beam, enabling precise simulation of the beam's coverage. Establishing a quantitative mapping relationship between the antenna beam parameters and the cone's geometric parameters is crucial for achieving accurate beam modeling, ensuring that the 3D model accurately reflects the antenna's actual beam characteristics.

[0035] Furthermore, converting the cone model into an editable polygon model is to meet the customized editing needs of different scenarios. The editable polygon model can flexibly adjust the shape of the bottom coverage area to adapt to the contour features of different scenarios, such as the narrow contour of a tunnel scene or the complex terrain contour of a mountain viaduct scene. Configuring skeletal rigging animation for the edited beam model is to dynamically demonstrate the antenna beam tracking function. Through the coordinated movement of the main skeleton nodes and sub-skeleton nodes, the beam tracking trajectory is accurately simulated. The antenna tracking parameters serve as the driving source, ensuring that the animation effect is consistent with the actual working state of the antenna.

[0036] Superimposing voxel meshes within the conical coverage space enables a refined representation of beam coverage intensity. The voxel density distribution is correlated with beam gain, revealing the strength distribution of beam coverage and helping users better understand the antenna beam coverage performance. Through a full-process rendering workflow encompassing geometry optimization, texture mapping, and lighting calculations, geometry optimization eliminates invalid model units and merges continuous units with the same attributes, improving rendering efficiency. Texture mapping distinguishes beam coverage intensity using different colors, enhancing visualization. Lighting calculations are customized to the ambient light characteristics of different scenes, improving the model's integration with the scene, ultimately generating a high-fidelity 3D beam coverage effect model.

[0037] like Figure 2 The diagram illustrates the parametric modeling and editable polygon editing process for an antenna beam cone. It clearly demonstrates the complete workflow from parametric modeling of the cone to scene-based editing of the editable polygon. Figure 4The diagram shows a composite structure of a beam model cone and a voxel mesh, demonstrating the superposition effect of the cone base model and the voxel mesh.

[0038] This demonstrates that by employing a technical chain encompassing parametric modeling, scenario-based editing, animation configuration, intensity representation, and full-process rendering, the antenna beam has been transformed from abstract parameters into a high-fidelity 3D visualization model. This solves the problem that existing 3D visualization technologies cannot accurately reproduce the spatial propagation and tracking characteristics of the beam. Furthermore, the customized scene editing and animation display functions enable the beam model to adapt to different application scenarios and dynamically present its working status, enhancing the professionalism and intuitiveness of the beam display and helping users quickly understand the beam performance of antenna products. Example 3

[0039] Furthermore, the digital twin scene construction module is configured to construct a digital twin 3D scene and bind the antenna installation pose to the scene's world coordinate system in the following way.

[0040] The system constructs a virtual city base and three core sub-scenes: high-speed rail plains, mountain viaducts, and tunnels, recreating the actual antenna deployment environment at a 1:1 scale. It incorporates an antenna installation pose calibration algorithm to accurately bind the antenna installation pose to the scene's world coordinate system. It also pre-sets fusion rules between the beam model and the scene environment, establishes and stores a mapping library of associations between scenes and products, beam parameters, scene calibration coefficients, and editing rules, and automatically matches parameters and rules when switching scenes based on this mapping library.

[0041] Among them, a virtual city base is constructed as the basic scenario, and on this basis, three core sub-scenarios are built: high-speed rail plains, mountain viaducts, and tunnels. These three scenarios are the main application scenarios of antenna products. The actual deployment environment of the antenna is restored at a 1:1 scale, which can provide users with a realistic scenario experience and help users intuitively feel the deployment and coverage effect of the antenna in the actual environment.

[0042] Furthermore, the built-in antenna installation pose calibration algorithm achieves precise binding between the antenna installation pose and the scene's world coordinate system through coordinate calibration and pose parameter matching steps. This ensures that the antenna model's position and orientation in the scene are consistent with the actual deployment, laying the foundation for the accurate fusion of the beam model and the scene.

[0043] Among them, the preset beam model and scene environment fusion rules include wall-hugging rules for tunnel scenes, unobstructed propagation rules for viaduct scenes, and long-distance coverage rules for plain scenes. The fusion rules for different scenes ensure that the beam model can truly reflect the coverage effect in the scene. A mapping library is established to associate scenes with products, beam parameters, scene calibration coefficients, and editing rules. Various types of data and rules are associated and stored. When the scene is switched, the system can quickly retrieve the corresponding parameters and rules from the mapping library to achieve automatic matching without manual intervention.

[0044] like Figure 5 and Figure 6 As shown, this is a schematic diagram of the fusion effect of digital twin scene and beam visualization (taking tunnel scene and plain scene as examples). From the figure, we can see the fusion effect of beam model with different scenes. In the tunnel scene, the beam model fits the inner wall of the tunnel, while in the plain scene, the beam model shows a long-distance coverage state.

[0045] Therefore, by constructing a 1:1 scale digital twin scene, a realistic environmental carrier is provided for antenna beam display, solving the problem of existing 3D display being disconnected from the actual environment; the antenna installation posture calibration algorithm ensures the accurate deployment of the antenna model in the scene, and the association mapping library between the scene and various parameters and rules realizes automatic matching when switching scenes, improving the system's intelligence level and ease of operation, enabling users to quickly view the antenna beam coverage effect in different scenes and efficiently match the optimal antenna solution for their own scene. Example 4

[0046] Furthermore, the multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition in the following ways.

[0047] This system integrates an ASR speech recognition unit, a TTS speech synthesis unit, and a large-scale model inference engine, supporting multimodal interaction with both Chinese and English text and voice input. Customized training is performed on keywords related to antenna product scenarios, solutions, beam parameters, and performance indicators to optimize intent recognition accuracy. A standardized intent recognition process is built-in, enabling precise differentiation between knowledge-based question-and-answer requests and display / call requests. This standardized process includes three core steps: keyword extraction, feature matching, and request type determination. The integrated ASR speech recognition unit converts user voice input into text in real time, the TTS speech synthesis unit converts system feedback into natural and fluent speech output, and the large-scale model inference engine provides powerful computational support for intent recognition. It supports multimodal interaction with both Chinese and English text and voice input, meeting the operating habits of different users and global marketing needs. Customized training on keywords related to antenna product scenarios, solutions, beam parameters, and performance indicators is designed to improve intent recognition accuracy, allowing the system to accurately understand the professional needs of the antenna industry and avoid intent recognition bias caused by training with generic keywords.

[0048] The built-in standardized intent recognition process involves several steps: the keyword extraction stage extracts core keywords from user input text and filters out stop words, focusing on the core information of user needs; the feature matching stage accurately matches the extracted keywords with the feature word library in the knowledge base to determine the relevance of the needs; and the needs type determination stage distinguishes between knowledge question-and-answer type needs and display and call type needs based on the matching results, ensuring that the system can provide accurate responses to different types of needs.

[0049] like Figure 7 The diagram shows a multimodal AI interaction workflow, illustrating the complete process from user input to demand response, including multimodal input processing, intent recognition, demand classification, and response execution.

[0050] This demonstrates that by employing multimodal interaction methods and customized keyword training, the system improves the ease of interaction between the system and users, as well as the accuracy of intent recognition, thus solving the problem of inaccurate intent recognition in existing AI interaction technologies used in the antenna industry. The standardized intent recognition process enables precise differentiation between knowledge-based question-and-answer requests and display / call requests, ensuring that the system can provide targeted responses to different types of requests, thereby enhancing the user experience and reducing the reliance of marketing on professionals. Example 5

[0051] In addition, the structured knowledge base module is used to build an enterprise knowledge base that combines vectorization and structure, integrating antenna product beam parameters, scene adaptation cases, technical documents and solution description information. It also stores technical data such as scene calibration coefficients, editable polygon editing rules, skeleton binding animation parameters and voxel mesh mapping rules, and organizes them according to a four-level structure of scene, solution, product and beam parameters, providing data support for AI intent recognition and beam parameter calling.

[0052] The system comprises a corporate knowledge base that combines vectorization and structured methods. Structured storage ensures data standardization and readability, facilitating rapid querying and retrieval. Vectorization improves data retrieval efficiency, especially when dealing with large amounts of similar data, enabling rapid matching of relevant information to meet the high-efficiency data requirements of AI intent recognition and beam parameter retrieval. Integrated antenna product beam parameters, scenario adaptation cases, technical documents, and solution descriptions form the foundation for intelligent interaction and beam modeling. Technical data such as scenario calibration coefficients, editable polygon editing rules, skeleton binding animation parameters, and voxel mesh mapping rules ensure the accuracy and professionalism of beam modeling, editing, and animation production. Organized according to a four-level structure of scenario, solution, product, and beam parameters, the knowledge base has a clear logical hierarchy and tight data connections, enabling rapid tracing and retrieval from scenario requirements to beam parameters. This provides accurate data support for AI intent recognition and facilitates maintenance and updates of the knowledge base content for the enterprise.

[0053] Therefore, by constructing an enterprise knowledge base that combines vectorization and structuring, the scattered antenna product-related data is integrated and standardized, solving the problem of fragmented and difficult-to-reuse knowledge assets in traditional antenna marketing. The four-level organizational structure improves the efficiency of data retrieval and retrieval, providing efficient and accurate data support for AI intent recognition and beam parameter retrieval, while also providing a reliable technical carrier for the accumulation and reuse of enterprise knowledge assets. Example 6

[0054] In addition, the message collaboration module uses the WebSocket protocol to realize real-time communication between the front-end presentation layer and the core technology layer. It is used to receive display call instructions from the multimodal AI interaction module, send scene jump instructions to the digital twin scene construction module, and send beam rendering update instructions to the beam visualization module, so as to realize the synchronous execution between AI interaction, scene switching and beam rendering.

[0055] The reason for using the WebSocket protocol for real-time communication is that the WebSocket protocol supports full-duplex communication, which can establish a persistent connection between the client and the server, ensuring the real-time performance and stability of data transmission, and meeting the communication requirements for AI interaction, scene switching and beam rendering synchronous execution.

[0056] The message collaboration module receives the display call instruction from the multimodal AI interaction module, parses the instruction to clarify the target scene, product, and beam display requirements, and then sends a scene jump instruction to the digital twin scene construction module to ensure that the scene can be quickly switched to the target scene. At the same time, it sends a beam rendering update instruction to the beam visualization module to ensure that the beam model can be rendered and updated according to the parameters and rules of the target scene, so as to realize the synchronous execution of scene switching and beam rendering.

[0057] Therefore, by using the real-time communication link built through the WebSocket protocol and the command forwarding and coordination functions of the message collaboration module, synchronous execution between AI interaction, scene switching and beam rendering is achieved, solving the problems of poor coordination and delayed response of various modules in the existing system; ensuring that users can quickly see the target scene and the corresponding beam display effect after issuing a display call command, improving the system's interactive smoothness and user experience. Example 7

[0058] Therefore, based on Embodiments 1 to 6, this application also provides a method for implementing an antenna beam visualization intelligent agent interactive display system, including: Step 101: Construct an association system for antenna products, beam parameters, and scenarios; collect structured data of antenna products; establish a four-level association data structure for products, beam parameters, solutions, and scenarios; and construct a digital twin scenario system and preset scenario roaming paths. A system linking antenna products, beam parameters, and scenarios is constructed. Structured data of antenna products is collected, and a four-level data structure linking products, beam parameters, solutions, and scenarios is established. A digital twin scenario system and preset scenario roaming paths are also built. The collected structured data includes product 3D model files (obj / fbx / gltf formats), beam parameters (horizontal coverage angle, vertical coverage angle, gain, beamwidth, etc.), electrical performance, mechanical performance, and scenario adaptation specifications. Installation pose parameters and scenario calibration coefficients for each product in different scenarios are also collected. This data forms the foundation for the linking system, ensuring its comprehensiveness and accuracy. The four-level data structure links products, beam parameters, solutions, and scenarios. Each antenna product is bound to a corresponding beam effect. Each product and its corresponding beam parameter set are associated with an adaptation solution. Each solution corresponds to a specific application scenario. Dedicated antenna installation pose parameters, scenario calibration coefficients, editable polygon editing rules, and beam model fusion rules are configured for each scenario. All relationships, parameters, and rules are stored in a mapping library to ensure automatic matching of parameters and rules during scenario switching. Building a digital twin scenario system includes creating a virtual city base, meticulously constructing three sub-scenes: high-speed rail plains, mountain viaducts, and tunnels, modeling and simulating antenna deployment environments, supporting automatic scene roaming, with the roaming path following a preset order of "plains → viaducts → tunnels," and simultaneously displaying scene names, typical product identifiers, and solution names, providing users with a clear scene browsing path.

[0059] Step 102: Implement 3D visualization rendering and animation of antenna beam parameters, including parametric modeling of the cone, scene editing of editable polygons, skeleton binding and beam tracking animation, voxel mesh overlay intensity expression, beam and scene fusion, and scene switching linkage. Achieve 3D visualization rendering and animation of antenna beam parameters, including parametric modeling of cones, scene-based editing of editable polygons, beam tracking animation with skeleton binding, intensity expression of voxel mesh overlay, beam and scene fusion, and scene switching linkage. Among them, the parametric modeling of the cone is based on the quantitative mapping relationship between antenna beam parameters and cone geometric parameters, transforming antenna horizontal coverage angle, vertical coverage angle, propagation distance and other parameters into corresponding geometric parameters of the cone to generate a basic beam model; the editable polygon scene-based editing customizes the bottom surface of the cone according to the contour features of different scenes to adapt it to scene requirements; the skeleton binding beam tracking animation production realizes the dynamic display of beam tracking through steps such as arranging bone nodes, binding weights, and setting driving sources; the voxel mesh superposition intensity expression uses the voxel density gradient algorithm to correlate beam gain with voxel density to present the beam coverage intensity; the beam and scene fusion binds the beam model to the scene coordinate system through the antenna installation pose calibration algorithm, and realizes the coverage effect display according to the scene fusion rules; the scene switching linkage ensures that when the scene changes, the beam model can automatically match the parameters and rules of the new scene to achieve seamless linkage.

[0060] Step 103: Construct a multimodal AI interaction system to perform multimodal recognition and intent determination on user input, and process knowledge-based question-and-answer requirements and display-and-call requirements separately; A multimodal AI interaction system is constructed to perform multimodal recognition and intent determination on user input, and to process knowledge-based question-and-answer and display-based request needs separately. Specifically, multimodal recognition processing uses an ASR speech recognition component to convert user speech input into text, and combined with text input functionality, supports multimodal interaction in both Chinese and English. Intent determination uses a standardized intent recognition process to differentiate between knowledge-based question-and-answer and display-based request needs. For knowledge-based question-and-answer requests, corresponding information is retrieved from a structured knowledge base and provided with simultaneous text and speech feedback. For display-based request needs, a message collaboration module triggers the simultaneous execution of scene transitions and beamforming.

[0061] Step 104: Implement product matrix navigation and multilingual adaptation, and perform statistics and analysis based on user interaction behavior data.

[0062] It enables product matrix navigation and multilingual adaptation, and performs statistical analysis based on user interaction behavior data.

[0063] Among them, the product matrix navigation constructs a four-level navigation system of scenarios, solutions, products, and beam parameters, enabling rapid navigation and related display of content at each level; multi-language adaptation enables bilingual switching between Chinese and English throughout the system interface, knowledge base, and interactive feedback, meeting the needs of global marketing; user interaction behavior data statistics and analysis collects user behavior data such as scenario clicks, product queries, and parameter inquiries through tracking points, generating analysis reports to provide data support for product planning and precision marketing.

[0064] like Figure 8 The diagram shows a four-level navigation system consisting of scenario, solution, product, and beam parameters, illustrating the hierarchical structure of the product matrix navigation.

[0065] like Figure 9 The diagram shown is a flowchart illustrating the overall steps of the implementation method of this invention, showing the complete process from the construction of the association system to data statistical analysis.

[0066] Thus, through the orderly execution of four steps, a methodology for building a related system, realizing beam visualization, establishing AI interaction, and adapting navigation and data has been formed. This further realizes the full-process implementation of intelligent marketing display of antenna products with beam visualization. It can solve problems such as abstract beam effects, disconnect between scenarios and products, and lagging interactive response in traditional antenna marketing. It realizes a full-process intelligent closed loop of consultation, identification, display, and explanation, improves the efficiency of customers' understanding of antenna products, reduces marketing costs, and helps antenna products expand into the global market. Example 8

[0067] Furthermore, the mapping relationship for the parametric modeling of the cone in step 102 is as follows: Horizontal cone angle = Antenna horizontal coverage angle × Scene calibration coefficient k1; Vertical cone angle = antenna vertical coverage angle × scene calibration coefficient k2; Cone height = maximum beam propagation distance.

[0068] For example, the scene calibration coefficients k1 and k2 are set to adapt to the beam propagation attenuation characteristics of different scenes. In tunnel scenes, due to limited space and large signal attenuation, k1=0.95 and k2=0.9 can be set; in mountainous viaduct scenes, due to factors such as terrain obstruction, the signal attenuation is moderate, so k1=0.98 and k2=0.97 can be set; in plain scenes, there is no obstruction and the signal attenuation is small, so k1=1 and k2=1 can be set. By adjusting the scene calibration coefficients, it is ensured that the cone model can truly reflect the beam coverage range under different scenes.

[0069] It should be noted that the height of the cone is directly equal to the maximum propagation distance of the beam because the height of the cone can intuitively reflect the propagation distance of the beam, ensuring that the 3D model is consistent with the actual propagation performance of the antenna.

[0070] It is evident that by establishing a clear quantitative mapping relationship and setting scenario-based calibration coefficients, the accurate conversion of antenna beam parameters to conical geometric parameters is achieved, ensuring that the beam model in different scenarios can truly reflect the actual coverage characteristics of the antenna. This solves the problem of beam modeling being disconnected from the actual scenario in existing technologies and improves the accuracy and professionalism of beam visualization. Example 9

[0071] Furthermore, the skeleton rigging and beam tracking animation in step 102 includes: Arrange a main skeleton node along the central axis of the beam model; Arrange multiple child bone nodes at the vertices of the polygon covering the bottom area and establish a parent-child hierarchical relationship; Weight binding is applied between the polygonal faces of the beam model and the bone nodes; Using the antenna beam tracking angle as the animation driver, the main skeleton node drives the sub-skeleton nodes to move, thereby achieving beam tracking animation.

[0072] The reason for placing a main skeleton node on the central axis of the beam model is that the central axis is the core path of beam propagation, and the movement of the main skeleton node can dominate the tracking direction of the beam. The reason for placing multiple sub-skeleton nodes on the vertices of the polygon at the bottom coverage area is to precisely control the movement trajectory of the bottom coverage area of ​​the beam, so as to ensure the integrity and accuracy of the tracking animation.

[0073] Establish a parent-child hierarchical relationship between the main skeleton node and the child skeleton node so that the child skeleton node can move synchronously with the main skeleton node, ensuring the coordination of the beam tracking animation. Weighted binding is performed between the polygonal face of the beam model and the bone nodes. The central axis area is fully bound to the main bone node (weight=1), and the bottom coverage area is bound to the corresponding child bone node according to the distance weight (the closer the distance, the higher the weight, and the weight range is 0.5-1). This ensures that the model face can follow the movement of the corresponding bone node, achieving a smooth and natural animation effect.

[0074] Using the antenna beam tracking angle as the animation driver, the main skeleton node rotates synchronously with the horizontal / vertical tracking angle, causing the sub-skeleton nodes and the bottom coverage polygon to move in the corresponding direction, accurately simulating the real-time tracking effect of the antenna beam. At the same time, the frame rate of the beam tracking animation is matched with the actual tracking speed of the antenna beam. The frame rate is set to beam tracking angular velocity × 0.1fps / ° to ensure that the animation effect is smooth and without lag.

[0075] like Figure 3 The diagram shown is a logic diagram for the implementation of beam model skeleton binding and tracking animation. The diagram shows the arrangement and hierarchical relationship of the main bone nodes and sub-bone nodes, as well as the animation driving and motion transmission logic.

[0076] As can be seen, by employing a series of technical methods such as skeletal node arrangement, weight binding, and driver source setting, accurate simulation of antenna beam tracking animation has been achieved, solving the problem that existing technologies cannot dynamically display beam tracking characteristics. The animation effect is highly consistent with the actual working state of the antenna, which can present the beam tracking function, help users better understand the dynamic performance of antenna products, and enhance the professionalism and attractiveness of marketing displays. Example 10

[0077] Furthermore, the intent recognition process in step 103 includes: Extract core keywords from user-input text and filter out stop words; The extracted keywords are matched with the feature word library in the knowledge base to determine the relevance of the requirements. When identified as a parameter query, technical document consultation, or performance indicator query, it is determined to be a knowledge-based question and answer request. When identified as scene display, beam effect viewing, solution visualization, or beam tracking animation demonstration, it is determined to be a display call type requirement.

[0078] Extracting core keywords from user input text and filtering stop words aims to focus on the core information of user needs, remove meaningless and redundant words, and improve the efficiency and accuracy of intent recognition. Core keywords include professional terms related to antenna marketing, such as the scenario name, solution type, product model, beam parameters, and performance indicators of antenna products.

[0079] The extracted keywords are matched with the feature word library in the knowledge base. The feature word library contains feature words for knowledge questions and answers and feature words for display and call. By matching, the relevance of the user's needs can be quickly determined, whether it is a knowledge-based need such as pure parameter query or technical document consultation, or a display-based need such as scene display or beam effect viewing.

[0080] When a request is identified as a parameter query, technical document consultation, or performance indicator query, it is determined to be a knowledge-based question and answer request. The system retrieves the corresponding information from the structured knowledge base and provides feedback. When a request is identified as a scene display, beam effect viewing, solution visualization, or beam tracking animation demonstration, it is determined to be a display call request. The system triggers the synchronous execution of scene jump and beam rendering to ensure accurate response to different types of requests.

[0081] This demonstrates that a standardized intent recognition process enables precise differentiation between knowledge-based question-and-answer requests and display / call requests, solving the problem that existing AI interaction technologies cannot accurately identify the types of user needs in the antenna industry. This ensures that the system can provide targeted responses to different types of needs, with knowledge-based question-and-answer requests providing rapid and accurate feedback, and display / call requests triggering simultaneous scene and beam display, thus improving the user interaction experience and the accuracy of system responses.

[0082] This application also provides a computer-readable medium storing at least one instruction, which is loaded and executed by the processor to implement the antenna beam visualization intelligent agent interactive display method described in the above embodiments.

[0083] The above description is merely an optional embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An antenna beam visualization intelligent agent interactive display system, characterized in that, The system adopts a three-layer architecture design, including a facility layer, a core technology layer, and an application display layer. Each layer works together to achieve antenna beam visualization and intelligent agent display. The facility layer provides hardware and computing power support for system operation, including 3D rendering computing power resources, large model inference computing power resources, data storage resources, network communication resources, as well as front-end display terminals, directional microphones, and voice pickup devices. The facility layer also supports local independent deployment of the system. The core technology layer is the core function implementation layer of the system, including a beam visualization module, a digital twin scene construction module, a multimodal AI interaction module, a structured knowledge base module, and a message collaboration module. These modules work together to achieve antenna beam parameter modeling, digital twin scene construction, intelligent interaction, and system linkage control. Specifically, the beam visualization module is configured to realize 3D visualization modeling and rendering of the antenna beam; the digital twin scene construction module is configured to construct a digital twin 3D scene of the antenna deployment environment and bind the antenna installation pose to the scene's world coordinate system; the multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition; the structured knowledge base module is configured to store structured data related to the antenna product; and the message collaboration module is configured to realize linkage communication between the modules of the core technology layer and between the core technology layer and the application display layer. The application display layer is the system's visual interaction entry layer, including a digital twin scene roaming module, a beam visualization display module, a 3D product details module, an AI intelligent question module, and a product matrix navigation module; wherein, the AI ​​intelligent question module is configured to provide intelligent question-and-answer interaction functions, and the AI ​​intelligent question module has the function of collecting all dialogue questions.

2. The antenna beam visualization intelligent agent interactive display system according to claim 1, characterized in that, The beam visualization module is configured to achieve three-dimensional visualization modeling and rendering of the antenna beam in the following ways: Using a cone as the core basic geometry, a quantitative mapping relationship between antenna beam parameters and cone geometric parameters is established, and beam parameterization modeling is completed. The cone model is converted into an editable polygon model, and customized scene editing is performed on the bottom coverage area of ​​the beam according to the contour features of different application scenarios. Configure skeletal rigging animation for the edited beam model to achieve beam tracking display, and use antenna tracking parameters as the animation driving source to achieve precise linkage between animation and antenna tracking actions; A voxel grid is superimposed within the cone-shaped coverage space, and a quantitative mapping relationship between beam gain and voxel density is established through a voxel density gradient algorithm, thereby achieving a refined expression of beam coverage intensity. Through a full-process rendering process including geometry optimization, texture mapping, and lighting calculation, a high-fidelity 3D beam coverage effect model is generated.

3. The antenna beam visualization intelligent agent interactive display system according to claim 1, characterized in that, The digital twin scene construction module is configured to construct a digital twin 3D scene and bind the antenna installation pose to the scene's world coordinate system in the following way: A virtual city base was constructed, along with three core sub-scenes: high-speed rail plains, mountain viaducts, and tunnels, to recreate the actual antenna deployment environment at a 1:1 scale. An internal antenna installation pose calibration algorithm is provided, which enables precise binding of the antenna installation pose to the scene's world coordinate system. The system pre-defines the fusion rules between the beam model and the scene environment, establishes and stores an association mapping library of scenes and products, beam parameters, scene calibration coefficients, and editing rules, and automatically matches parameters and rules when switching scenes based on the association mapping library.

4. The antenna beam visualization intelligent agent interactive display system according to claim 1, characterized in that, The multimodal AI interaction module is configured to receive user text / voice input and complete intent recognition in the following ways: It integrates an ASR speech recognition unit, a TTS speech synthesis unit, and a large model inference engine, supporting multimodal interaction with Chinese and English text input and voice input; Customized training is conducted based on keywords related to the scenarios, solutions, beam parameters, and performance indicators of antenna products to optimize the accuracy of intent recognition; The system incorporates a standardized intent recognition process, which enables accurate differentiation between knowledge-based question-and-answer requests and display / call requests. This standardized intent recognition process includes three core steps: keyword extraction, feature matching, and request type determination.

5. The antenna beam visualization intelligent agent interactive display system according to claim 1, characterized in that, The structured knowledge base module is used to build an enterprise knowledge base that combines vectorization and structure. It integrates antenna product beam parameters, scene adaptation cases, technical documents and solution descriptions, and stores technical data such as scene calibration coefficients, editable polygon editing rules, skeleton binding animation parameters and voxel mesh mapping rules. It is organized according to a four-level structure of scene, solution, product and beam parameters to provide data support for AI intent recognition and beam parameter calling.

6. The antenna beam visualization intelligent agent interactive display system according to claim 1, characterized in that, The message collaboration module is based on the WebSocket protocol to realize real-time communication between the front-end display layer and the core technology layer. It is used to receive display call instructions from the multimodal AI interaction module, send scene jump instructions to the digital twin scene construction module, and send beam rendering update instructions to the beam visualization module, so as to realize the synchronous execution between AI interaction, scene switching and beam rendering.

7. A method for interactive display of antenna beam visualization intelligent agents, characterized in that, The antenna beam visualization intelligent agent interactive display system based on any one of claims 1 to 6 is implemented, including: Step 101: Construct an association system for antenna products, beam parameters, and scenarios; collect structured data of antenna products; establish a four-level association data structure for products, beam parameters, solutions, and scenarios; and construct a digital twin scenario system and preset scenario roaming paths. Step 102: Implement 3D visualization rendering and animation of antenna beam parameters, including parametric modeling of the cone, scene editing of editable polygons, skeleton binding and beam tracking animation, voxel mesh overlay intensity expression, beam and scene fusion, and scene switching linkage. Step 103: Construct a multimodal AI interaction system to perform multimodal recognition and intent determination on user input, and process knowledge-based question-and-answer requirements and display-and-call requirements separately; Step 104: Implement product matrix navigation and multilingual adaptation, and perform statistics and analysis based on user interaction behavior data.

8. The method for interactive display of antenna beam visualization intelligent agent according to claim 7, characterized in that, The mapping relationship for the parametric modeling of the cone in step 102 is as follows: Horizontal cone angle = Antenna horizontal coverage angle × Scene calibration coefficient k1; Vertical cone angle = antenna vertical coverage angle × scene calibration coefficient k2; Cone height = maximum beam propagation distance.

9. The method for implementing interactive display of antenna beam visualization intelligent agent according to claim 7, characterized in that, The skeleton rigging and beam tracking animation in step 102 includes: Arrange a main skeleton node along the central axis of the beam model; Arrange multiple child bone nodes at the vertices of the polygon covering the bottom area and establish a parent-child hierarchical relationship; Weight binding is applied between the polygonal faces of the beam model and the bone nodes; Using the antenna beam tracking angle as the animation driver, the main skeleton node drives the sub-skeleton nodes to move, thereby achieving beam tracking animation.

10. The method for implementing interactive display of antenna beam visualization intelligent agent according to claim 7, characterized in that, The intent recognition process in step 103 includes: Extract core keywords from user-input text and filter out stop words; The extracted keywords are matched with the feature word library in the knowledge base to determine the relevance of the requirements. When identified as a parameter query, technical document consultation, or performance indicator query, it is determined to be a knowledge-based question and answer request. When identified as scene display, beam effect viewing, solution visualization, or beam tracking animation demonstration, it is determined to be a display call type requirement.

Images