A location service system and method based on digital twin technology

Abstract

The embodiment of the present application relates to the technical field of cultural tourism, and discloses a location service method based on digital twin technology, which comprises the following steps: receiving a digital beacon broadcast by an environment perception device in a historical and cultural block; determining the context state information and the position information of the digital beacon in a digital twin model according to the received digital beacon; generating a user abstract request according to the context state information and the historical preference of a user; sending the user abstract request to an edge computing node at the edge of a network of the corresponding historical and cultural block, wherein the edge computing node is used for maintaining a lightweight semantic map derived from the digital twin model; generating a corresponding location service response package at the edge computing node according to the user abstract request and the current state of the lightweight semantic map; and sending the location service response package to a corresponding user terminal, and executing the response package by the user terminal to provide the user with a location service deeply integrated with the historical and cultural block. The method in the embodiment of the present application greatly improves the cultural tourism experience of the user.

Description

Technical Field

[0001] This invention relates to the field of cultural tourism technology, specifically to a location service method and system based on digital twin technology. Background Technology

[0002] Currently, users typically rely on paper maps provided by the scenic area or their own sense of direction when visiting historical and cultural districts. However, this can be inefficient, especially in large areas with high foot traffic, making it easy for users to get lost and wander aimlessly. Therefore, designing a solution that provides efficient location services is a pressing technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0003] To address the aforementioned shortcomings, this invention discloses a location service method based on digital twin technology, which can achieve efficient and accurate location services.

[0004] The first aspect of this invention discloses a location service system based on digital twin technology, comprising: Receiving module: Used to receive digital beacons broadcast by environmental sensing devices installed in the historical and cultural district, wherein the digital beacons include the environmental status of the area in which they are located; The generation module is used to determine the contextual state information and location information of the received digital beacon in the digital twin model; and to generate a corresponding user abstract request based on the contextual state information and the user's historical preferences. Edge computing module: used to send the user abstract request to the edge computing node at the edge of the corresponding historical and cultural block network, the edge computing node is used to maintain a lightweight semantic map derived from the digital twin model; Interactive service module: used to generate a corresponding location service response package at the edge computing node based on the user abstract request and the current state of the lightweight semantic map; send the location service response package to the corresponding user terminal, and have the user terminal execute the response package to provide the user with location services deeply integrated with the historical and cultural blocks.

[0005] A second aspect of this invention discloses a location service method based on digital twin technology, comprising: Receives digital beacons broadcast by environmental sensing devices installed in historical and cultural districts, the digital beacons including the environmental status of their respective areas; The contextual state information and location information of the received digital beacon in the digital twin model are determined; and a corresponding user abstract request is generated based on the contextual state information and the user's historical preferences. The user abstract request is sent to the edge computing node at the edge of the corresponding historical and cultural block network. The edge computing node is used to maintain a lightweight semantic map derived from the digital twin model. The edge computing node generates a corresponding location service response package based on the user abstract request and the current state of the lightweight semantic map; the location service response package is sent to the corresponding user terminal, and the user terminal executes the response package to provide the user with location services deeply integrated with the historical and cultural blocks.

[0006] As an optional implementation, in a second aspect of the present invention, determining the contextual state information of the received digital beacon in the digital twin model includes: The digital beacon is parsed to obtain the environmental attributes encoded therein, including noise level, crowd density estimate, lighting conditions, and point of interest identifiers. Combine data from the inertial measurement unit built into the user terminal to determine the user's activity status; The environmental attributes are fused with the activity state to form a vector representation of the contextual state information in the digital twin model.

[0007] As an optional implementation, in a second aspect of the present invention, generating a corresponding location service response packet based on the user abstract request and the current state of the lightweight semantic map includes: The user's abstract request is parsed into corresponding constraints and optimization objectives; In the lightweight semantic map, find available service resources or paths that can satisfy the constraints; From all the solutions that meet the conditions, select the solution that meets the optimization objective and encapsulate it into the location service response package; wherein, the optimization objective includes one or more of the following: shortest time, lowest energy consumption, highest cultural experience value, or strongest congestion avoidance. The location service method further includes: During the execution of the location service response packet by the user terminal, the deviation between the user's actual interaction feedback and the expected effect of the edge computing node is monitored; The detected deviation data is fed back to the corresponding server; The bias data is used at the server to optimize and train a machine learning model for identifying user context patterns or generating service components.

[0008] As an optional implementation, in a second aspect of the present invention, the location service method further includes: Image frames containing the environment of the historical and cultural district are acquired through the image sensor of the user terminal; From the image frames, visual environment features at different spatial scales are extracted; the extraction of visual environment features at different spatial scales includes: Extract the first scale feature, which is the skyline contour feature vector calculated based on the boundary line between the roof of the building and the sky in the image; Extract second-scale features, which are highly discriminative local texture or structural feature points and their descriptors detected from the building facade area in the image. Extract third-scale features, which are paving material patterns, special markings, or landform features identified from ground areas in the image; The extracted visual environment features are matched with a reference feature library corresponding to different spatial scales, which is pre-rendered from the digital twin model. Based on the matching results, the first position is calculated.

[0009] As an optional implementation, in a second aspect of the present invention, the step of generating a corresponding location service response packet at the edge computing node based on the user abstract request and the current state of the lightweight semantic map includes: Determine the target service area in the digital twin model that matches the user's abstract request; According to the user terminal's expected movement direction or interest diffusion direction within the target service area, the target service area is divided into multiple sequentially associated service segments; For each service segment, a corresponding segment context dataset is extracted from the lightweight semantic map. The segment context dataset includes the segment's real-time congestion level, environmental atmosphere, set of accessible points of interest, and its current state. For the first-order service segment, the corresponding segment scenario dataset is input into the first service adaptation model to obtain the first service suitability score and recommended service content for the corresponding service segment. For each subsequent service segment that is not in the first order, its corresponding segment context dataset and the service suitability score of its immediate upstream service segment are input into the corresponding subsequent service adaptation model to obtain the service suitability score and recommended service content for that subsequent segment. The score of the upstream segment serves as the context constraint for the service recommendation of the downstream segment.

[0010] As an optional implementation, in a second aspect of the present invention, the location service method further includes: After generating the location service response package, monitor the user's actual interaction feedback on each service segment during the execution process; When the actual feedback of a certain service segment is significantly lower than its predicted service suitability score, the segment is marked as a segment to be optimized. Analyze the service suitability scores of the adjacent upstream segments of the segment to be optimized and the actual behavior of users in the upstream segments, and calculate the influence contribution of the upstream segments to the segment to be optimized; If the impact contribution is greater than or equal to the preset contribution threshold, the upstream segment is updated and marked as a new segment to be optimized, and the analysis continues to trace upstream. If the impact contribution is less than the preset contribution threshold, then the current segment to be optimized is determined as the actual optimized service segment.

[0011] As an optional implementation, in a second aspect of the present invention, the location service method further includes: After determining the actual optimized service segments, a multi-dimensional feature vector is extracted for each actual optimized service segment. The feature vector includes the location in the service path, physical space features, real-time context status, and associated user profile tags. The multidimensional feature vector is input into the service strategy matching engine, and dynamic optimization strategies that match the segment features are retrieved from the pre-generated service optimization strategy library; wherein, the strategies in the service optimization strategy library include path detour strategies, service content replacement strategies, service timing adjustment strategies, and guidance enhancement strategies; For the retrieved dynamic optimization strategy, determine whether it contains an instant adjustment strategy that can be executed in real time by the edge computing node; If included, the edge computing node generates corresponding control parameters and updates the location service response packet pushed to the user terminal in real time, guiding the user to execute the optimized path or experience content in the current session; After users complete the optimized experience, new feedback on the service segment to be optimized is collected, and this new feedback is compared with the historical feedback before optimization to generate an optimization effect evaluation report for iterative updates to the service strategy matching engine or the service optimization strategy library.

[0012] As an optional implementation, in a second aspect of the present invention, the segmented context dataset further includes spatiotemporal attribute data, which is determined in the following manner: The deviation between the average pedestrian flow pattern in the same historical period and the current pedestrian flow in this segment; The cultural experience intensity index of each point of interest within the segment is calculated based on its historical events and real-time status; the visual presentation quality score of the segment under current lighting and weather conditions; wherein, when calculating the service suitability score, the service adaptation model comprehensively weighs the real-time factor, the cultural experience factor, and the environmental comfort factor.

[0013] As an optional implementation, in a second aspect of the present invention, after determining the target service area in the digital twin model matching the user abstract request, the method further includes: The edge computing node parses the abstract request to determine its dominant request pattern; based on the determined dominant request pattern, it invokes the service segmentation strategy associated with that pattern to divide the multiple sequentially associated service segments. When the dominant request mode is the first request mode, the optimal path from the starting point to the definite destination is calculated on the lightweight semantic map; topological key points on the path are identified, including road intersections, significant terrain change points, and necessary landmark side points; the path is divided into multiple service segments with the topological key points as boundaries, and the navigation complexity value is calculated for each segment. When the dominant request mode is the second request mode, the real-time interest attraction value of each region in the lightweight semantic map is calculated based on the interest preference vector in the user's abstract request; continuous regions with attraction values ​​higher than a threshold are clustered to form multiple core interest region blocks; a virtual access sequence flow is defined for the multiple core interest region blocks based on geographical proximity and interest correlation; each core interest region block and its surrounding buffer area are defined as a corresponding service segment, and the order of the segments is determined according to the virtual access sequence flow. When the dominant request mode is the third request mode, the multiple subtasks defined in the user abstract request are parsed into a corresponding task dependency graph. In the task dependency graph, nodes represent subtasks, and edges represent execution order dependencies or spatial proximity relationships. The task dependency graph is topologically sorted, and subtask clusters that can be executed in parallel or are logically closely related are divided into the same service segment. Based on the task dependency relationship, the execution order of each service segment is determined.

[0014] A third aspect of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calling the executable program code stored in the memory to execute the location service method based on digital twin technology disclosed in the first aspect of the present invention.

[0015] A fourth aspect of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program causes a computer to execute the location service method based on digital twin technology disclosed in the first aspect of the present invention.

[0016] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: The method in this embodiment of the invention upgrades location services from a tool to a cultural experience carrier through a technical chain of environmental perception, context analysis, edge computing, and precise response. It not only solves the pain points of complex signals and diverse service needs in historical and cultural blocks, but also realizes the digital revitalization of cultural resources through digital twin technology, ultimately achieving the dual goals of improving technical efficiency and deepening cultural experience; thus enhancing the user's tour experience. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the location service method based on digital twin technology disclosed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the process for determining the context state information disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of a process for generating a location service response packet, as disclosed in an embodiment of the present invention. Figure 4 This is another schematic diagram of the process for generating a location service response packet disclosed in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a location service system based on digital twin technology provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be noted that the terms first, second, third, fourth, etc., in the specification and claims of this invention are used to distinguish different objects, not to describe a specific order. The terms used in the embodiments of this invention include and have, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0021] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a location service method based on digital twin technology disclosed in an embodiment of the present invention. The execution entity of the method described in this embodiment is an execution entity composed of software and / or hardware. This execution entity can receive relevant information via wired and / or wireless means and can send certain instructions. It may also have certain processing and storage functions. This execution entity can control multiple devices, such as remote physical servers or cloud servers and related software, or local hosts or servers and related software that perform related operations on devices located at a certain location. In some scenarios, multiple storage devices can also be controlled; these storage devices may be placed in the same location as the devices or in different locations. Figure 1 As shown, the location service method based on digital twin technology includes the following steps: S101: Receive digital beacons broadcast by environmental sensing devices installed in the historical and cultural district, wherein the digital beacons include the environmental status of the area in which they are located; S102: Determine the contextual state information and location information of the received digital beacon in the digital twin model; and generate a corresponding user abstract request based on the contextual state information and the user's historical preferences; S103: The user abstract request is sent to the edge computing node at the edge of the corresponding historical and cultural block network, and the edge computing node is used to maintain a lightweight semantic map derived from the digital twin model; S104: The edge computing node generates a corresponding location service response package based on the user abstract request and the current state of the lightweight semantic map; the location service response package is sent to the corresponding user terminal, and the user terminal executes the response package to provide the user with location services deeply integrated with the historical and cultural blocks.

[0022] The digital beacon in this invention collects real-time environmental data such as temperature, passenger density, and lighting conditions. Combined with the spatial mapping capabilities of the digital twin model, this enables the system to accurately determine the user's context. For example, when tourists encounter congestion in a narrow ancient alley, the system can automatically identify their potential needs for avoiding congestion and enjoying nearby views. This context-awareness based on the real environment solves the problem of traditional location services relying solely on coordinates and being disconnected from the physical scene.

[0023] In practice, by integrating users' historical preferences with real-time contextual information, the system generates abstract requests that enable personalized service customization. For example, it can prioritize pushing on-site guided tours of cultural relic restoration sites to history enthusiasts and recommend accessible sightseeing routes to families with children, shifting the service from standardized supply to anticipating user needs and significantly improving the relevance of the experience.

[0024] The lightweight semantic map maintained by the edge computing nodes in this invention strips away redundant data from the digital twin model, retaining only core service elements such as path planning and scene labels. Combined with acceleration technologies like TensorRT, it reduces data processing time. In scenarios with complex signal obstruction, such as historical and cultural districts, it avoids response delays caused by reliance on cloud computing. From sending the user's abstract request to generating the response packet, everything is completed at the edge nodes of the district, eliminating the need for cross-regional data transmission. This architectural design significantly improves service response speed compared to traditional cloud solutions, making it particularly suitable for unexpected scenarios. For example, when an exhibition hall temporarily closes, the edge nodes can immediately update map data and push alternative routes to nearby users, preventing experience failures due to information delays.

[0025] The semantic map status, synchronized in real time at edge nodes, can be quickly updated in response to dynamic changes in the neighborhood (such as temporary market setups or temporary exhibitions of cultural relics). For example, when an ancient street is closed due to folk activities, the system can instantly adjust the navigation route and push activity guides, enabling location services to deeply collaborate with the neighborhood's cultural activities and management needs, becoming the intelligent nerve endings of the neighborhood's operation. The local storage of the lightweight semantic map reduces the amount of data interaction between terminals and the cloud. This advantage is particularly important in historical districts with high visitor density, avoiding network congestion caused by massive concurrent requests from terminals.

[0026] In practice, sensitive data such as user preferences and real-time location are processed locally at edge nodes without being uploaded to the cloud. Combined with end-to-end encryption and hardware-level authentication, this significantly reduces the risk of data leakage. Meanwhile, the encrypted broadcast mechanism of digital beacons ensures the integrity of environmental awareness data, providing security support for reliable service decisions.

[0027] In this embodiment of the invention, the street itself becomes the sensing subject. Low-cost environmental intelligence modules are deployed on streetlights, benches, and building facades, enabling anonymous sensing and broadcasting of the micro-environmental status to the user's mobile phone via sound waves or light encoding. The user's mobile phone does not need to continuously utilize high-powered cameras and GPS for SLAM; instead, it receives these environmental broadcasts as digital beacons and, combined with low-power inertial sensors, achieves precise positioning and contextual understanding. This resolves the conflict between privacy and power consumption.

[0028] Edge Semantic Twin of this invention: Edge servers are deployed in each historical and cultural district to maintain a lightweight, high-timeliness semantic map. It contains only core entities (buildings, streets, key points of interest) and their current states (e.g., congestion, activity) and basic semantics. It is responsible for real-time intent recognition and instant service recommendations. Cloud-based Deep Twin: A complete, high-precision, historically rich digital twin is maintained in the cloud. It is responsible for complex spatiotemporal analysis, deep learning model training, macro-level pattern mining, and cross-district data fusion.

[0029] Pre-computed scenario-service response package: Based on historical big data and simulation, the cloud pre-computes the service packages (navigation path, information card, AR content) that may be used for various typical scenario modes (such as rainy day tour, festival food search, historical study tour) and pushes them to the edge server.

[0030] Specifically, users create a digital avatar representing their preferences on their local device. All sensitive behavioral analysis is performed locally, generating abstract intent vectors or target states. Users do not upload raw trajectory or visual data to the server; they are only allowed to perform target alignment and resource negotiation between the digital avatar and the edge semantic twin of the neighborhood.

[0031] In practical implementation, user-contributed virtual content (comments, graffiti, photos) can be organized into selectable layers based on themes and spatiotemporal dimensions. For example, layers could be created for 2024 Spring Festival memoirs, architectural photography masters, and local residents' life memories. Users can select and load spatiotemporal dimensions of interest, much like selecting map layers. Utilizing consortium blockchain technology, high-quality user-created virtual memoirs (such as an engaging AR comment interpreting history) can be hashed and lightweightly authorized, granting them uniqueness and credibility. This can even be linked to neighborhood incentive mechanisms (such as points and discounts) to stimulate high-quality UGC.

[0032] Lightweight semantic maps are a subset, simplified, and optimized version of the full city-level digital twin model on edge computing nodes. They are not a complete 3D model containing all details, but rather a semantic knowledge network optimized for real-time decision-making. Because the map is lightweight and deployed on edge nodes close to users, the round-trip latency for server request processing and response generation is low, a prerequisite for smooth AR navigation and real-time interaction. Numerous simple real-time decisions (such as route replanning and information retrieval) are made by edge nodes without needing to be uploaded back to the cloud, significantly reducing the load on core data centers and network transmission costs. Lightweight semantic maps can be updated extremely quickly, maintaining a high degree of synchronization with the physical world, providing accurate data for real-time decision-making.

[0033] In this embodiment of the invention, the user abstract request is a structured data object that has undergone local privacy processing, which includes: a field indicating the core intent type, a vector representing multidimensional interest preferences, and a set of constraints containing time, space, and physical constraints; wherein, the generation process of the abstract request does not disclose the user's precise real-time location, raw sensor data, and personal identification information to the edge computing node or the at least one server.

[0034] The lightweight semantic map in this embodiment of the invention is a real-time synchronized derived instance of the digital twin model on the edge computing node. It includes a simplified topological map of historical and cultural blocks, a list of semantic attributes of key points of interest and their current dynamic status, and a real-time environmental data layer overlaid on the topological map. The data volume and computational complexity of the lightweight semantic map are significantly reduced to adapt to the limited resources of the edge computing node and achieve millisecond-level service response.

[0035] In this embodiment of the invention, the lightweight semantic map and the digital twin model maintain incremental synchronization: the at least one server is responsible for calculating the global state changes and semantic updates of the digital twin model; only incremental update data that affects real-time location service decisions (such as changes in point of interest status and temporary activity information) are pushed to the edge computing node at a high frequency to update the lightweight semantic map maintained by the edge computing node; the edge computing node does not synchronize the complete, high-precision three-dimensional geometric model.

[0036] Specifically, the process of generating the first context state from digital beacons is as follows: The terminal receives and decrypts the digital beacon, obtaining a set of (beacon ID, signal strength, and a list of encoded environmental attributes). Using the IDs and signal strengths of multiple received beacons, a triangulation or fingerprint positioning algorithm is used to calculate the first location locally. At this point, there is no need to activate GPS or the camera. Based on the beacon ID, the detailed environmental attributes of the micro-area represented by the beacon are read from the locally cached (or pre-fetched from the edge server) beacon-environment mapping table, forming an environmental state vector. Local sensor data fusion: Data from the IMU (for determining mobility modality), light sensor, and microphone are fused in real time to update the environmental state (e.g., illumination, noise) and user state (mobility modality). The user's historical trajectory, dwell time, and interaction records for the current session are retrieved from local storage to update the task state and interest curve. All the above information is encoded into a standardized, multi-dimensional context state vector, for example: [Location: (x, y, z), Area type: narrow alley, Illumination: low, Noise: medium, Movement: walking, Attention: possibly confused, Battery level: medium, Task progress: 45%, Peak interest: handicraft, Fatigue level: 0.6]. More preferably, such as Figure 2 As shown, determining the contextual state information of the received digital beacon in the digital twin model includes: S1021: Parse the digital beacon to obtain the environmental attributes encoded therein, including noise level, crowd density estimate, lighting conditions, and point of interest identifiers; S1022: Combine data from the inertial measurement unit built into the user terminal to determine the user's activity status; S1023: The environmental attributes and the activity state are fused to form a vector representation of the contextual state information in the digital twin model.

[0037] The digital beacon of this invention encodes multi-dimensional environmental attributes such as noise level, crowd density estimation, and lighting conditions, overcoming the limitations of traditional location services that rely solely on coordinates and point-of-interest names for shallow perception. For example, crowd density estimation can distinguish between crowded alleys and empty courtyards, lighting conditions can determine whether a user is in an indoor exhibition hall or an outdoor ancient street, and noise level can help identify folk activity sites or quiet cultural relic viewing areas.

[0038] In practical implementation, integrating user terminal IMU data can accurately determine the user's activity status (such as walking, pausing to observe, going up and down stairs, riding in a vehicle), solving the problem that traditional solutions cannot distinguish between stationary waiting and pausing to observe, or between fast walking and slow sightseeing. For example, when the IMU detects that the user's acceleration is close to 0 and their posture is stable, combined with point-of-interest markers (such as a display case of an artifact), it can be determined that the user is pausing to observe; when continuous low-frequency vibration and uniform motion are detected, it can be determined that the user is riding in a vehicle, providing behavioral basis for subsequent personalized services.

[0039] In this embodiment of the invention, environmental attributes and activity states are fused into vector representations, transforming abstract scenarios into structured data that machines can directly process, thus avoiding the ambiguity and vagueness of traditional natural language descriptions. For example, a vector representing high crowd density (0.8), medium noise level (0.5), walking status (1.0), and a point of interest in a folk square (1.0) can be quickly matched by edge computing nodes as the core scenario for participating in folk activities, providing accurate decision input for subsequent service responses (such as pushing activity flow guides), thereby improving the response speed of scenario recognition to the millisecond level.

[0040] Vectorized contextual state information can be directly associated with the spatial coordinates and scene labels of the digital twin model, achieving a dual mapping between physical location and contextual state. For example, if a user's coordinates in the digital twin model are (X, Y, Z), the corresponding contextual vectors are low lighting conditions (0.2), a lingering state (1.0), and a point of interest being an ancient building plaque (1.0). The system can accurately mark the user's spatial location, behavioral intent, and environmental characteristics in the digital twin model, providing a precise three-dimensional index for subsequent semantic map queries and service generation, avoiding service matching deviations caused by relying solely on spatial coordinates. Environmental attributes (such as crowd density and noise level) are broadcast in real time via digital beacons, and IMU data is collected in real time via terminals. The contextual vector formed by the fusion of these two data sources can be updated in seconds and synchronized to the digital twin model. The vectorized contextual state information has completed the fusion and structured processing of multi-source environmental and behavioral data. When generating abstract user requests, edge computing nodes do not need to additionally parse scattered environmental and behavioral data; they can directly perform semantic matching and demand prediction based on the vectors.

[0041] More preferably, such as Figure 3 As shown, generating the corresponding location service response packet based on the user abstract request and the current state of the lightweight semantic map includes: S1041: Parse the user abstract request into corresponding constraints and optimization objectives; S1042: In the lightweight semantic map, find available service resources or paths that can satisfy the constraints; S1043: From all the solutions that meet the conditions, select the solution that meets the optimization objective and encapsulate it into the location service response package; wherein, the optimization objective includes one or more of the following: shortest time, lowest energy consumption, highest cultural experience value, or strongest congestion avoidance. The location service method further includes: S1044: During the execution of the location service response packet by the user terminal, monitor the deviation between the user's actual interaction feedback and the expected effect of the edge computing node; S1045: Feedback the detected deviation data to the corresponding server; S1046: At the server, the bias data is used to optimize and train the machine learning model used to identify user context patterns or generate service components.

[0042] In practical implementation, the abstract user request is broken down into constraints (such as prioritizing accessible access to intangible cultural heritage exhibition halls) and optimization goals (shortest time, highest cultural experience value, strongest crowd avoidance, etc.), solving the pain point of traditional location services being single-goal-oriented and neglecting users' implicit needs. For example, for elderly users, the system can use accessibility as a constraint, while prioritizing the optimization goal of the highest cultural experience value, planning routes that pass through core cultural relics sites without steps; for tourists in a hurry, the shortest time can be the core optimization goal, while also meeting the constraint of crowd avoidance, avoiding peak tourist areas. This two-layer screening mechanism ensures that the service solution fully meets the personalized needs of users.

[0043] The lightweight semantic map pre-integrates the street's service resources (such as guide points, rest areas, and cultural and creative stores), path attributes (such as slope, width, and congestion level), and cultural tags (such as historical value level). Edge computing nodes can quickly retrieve resources or paths that meet the requirements based on the parsed constraints and optimization objectives without traversing redundant data.

[0044] During the execution of response packages on the user terminal, the system monitors deviations between actual interaction feedback and expected results (such as users deviating from planned routes, dwell time far exceeding expectations, and proactively switching service types). This addresses the problem of traditional location-based service solutions becoming unmanageable after generation and failing to adapt to changes in user behavior. Deviation data is fed back to the server and used to optimize machine learning models (including contextual pattern recognition models and service component generation models), enabling the system to learn from practice. For example, for recurring deviations by elderly users from accessible routes, the model can optimize contextual recognition rules, adding features such as walking speed and dwell frequency to more accurately assess the user's physical condition. For deviations where the expected cultural experience value does not match reality, the model can adjust the weights of cultural tags for points of interest, making subsequent solutions more closely aligned with the user's actual experience.

[0045] Constraint-based filtering can preemptively eliminate unsuitable resources or paths, significantly reducing the number of candidate solutions for edge computing nodes. Simultaneously, multi-objective optimization supports on-demand priority selection (e.g., prioritizing cultural experience or avoiding congestion), avoiding meaningless full-objective traversal calculations. For example, when a user explicitly selects cultural experience as the highest-value core objective, the system can ignore detailed calculations of secondary objectives such as energy consumption and time, directly focusing on ranking the weights of cultural tags.

[0046] The analysis of deviation data and the optimization training of machine learning models are both completed on cloud servers. Edge nodes are only responsible for solution generation and deviation collection, without having to undertake complex model training tasks. This layered architecture of edge execution and cloud optimization leverages the low latency advantage of edge computing while avoiding computational overload on edge nodes due to model training. The accumulated deviation data and model optimization results can form a service knowledge base for different user groups (such as history enthusiasts, families with children, and senior tourists), enabling the system to continuously optimize personalized recommendation strategies. For example, deviation data for families with children can reveal patterns such as a preference for highly interactive intangible cultural heritage experiences and frequent rest areas. Subsequent service plans can prioritize routes that include interactive projects and rest points, achieving long-term service optimization.

[0047] More preferably, the location service method further includes: Image frames containing the environment of the historical and cultural district are acquired through the image sensor of the user terminal; From the image frames, visual environment features at different spatial scales are extracted; the extraction of visual environment features at different spatial scales includes: Extract the first scale feature, which is the skyline contour feature vector calculated based on the boundary line between the roof of the building and the sky in the image; Extract second-scale features, which are highly discriminative local texture or structural feature points and their descriptors detected from the building facade area in the image. Extract third-scale features, which are paving material patterns, special markings, or landform features identified from ground areas in the image; The extracted visual environment features are matched with a reference feature library corresponding to different spatial scales, which is pre-rendered from the digital twin model. Based on the matching results, the first position is calculated.

[0048] The ancient building complexes in historical and cultural districts possess unique skyline contours (such as the combination of eaves and bracket sets). These features are unaffected by local obstructions or crowds, serving as a global spatial positioning benchmark. By extracting the skyline contour feature vectors, the approximate area of ​​a user within a digital twin model can be quickly located. This also assists in achieving efficient positioning when digital beacons malfunction.

[0049] The unique textures of building facades, such as brick patterns, carvings, and door and window designs, are key distinguishing features from adjacent ancient buildings. Extracting these features and their descriptors allows for further narrowing of the location to the level of specific buildings or courtyards, building upon the initial macro-level anchoring. Historical and cultural districts often use paving materials with unique patterns, such as bluestone slabs and carved tiles, and may even include special markings like cultural relic signs and drainage ditches. These features can serve as the final location calibration basis, creating a multi-level positioning loop when combined with the first two levels of features.

[0050] A multi-scale reference feature library, pre-generated based on a digital twin model, contains standardized feature data of all buildings and ground surfaces within the block, avoiding the influence of on-site feature collection on lighting, weather, and seasonal changes (such as texture blurring due to cloudy days and ground pattern occlusion due to rain). By matching real-time extracted visual features with the reference feature library, the stability of the positioning results can be maintained under complex environmental interference. The high-precision location information output by this positioning method can be directly fused into the context state vector mentioned above, providing more accurate spatial data support for the generation of user abstract requests and the optimal selection of edge computing node solutions. For example, when locating the position of a cultural relic tile by combining ground paving features, the system can automatically push the historical age and craftsmanship information of the tile, extending the location service from route navigation to precise cultural relic-level guidance, further deepening the cultural experience.

[0051] Traditional visual positioning technologies often rely on single-scale features (such as extracting only building facade textures or ground features), which are easily affected by occlusion in historical and cultural districts (e.g., crowds obscuring the ground or trees obscuring building facades), leading to positioning failure. The solution in this invention employs a multi-scale feature layering extraction strategy. Through the complementarity of features at different scales, it ensures that even when local features fail, other scale features can still guarantee positioning continuity. This layered logic fundamentally solves the robustness problem of single-scale positioning and represents an optimization and reconstruction of the traditional visual positioning technology framework.

[0052] Traditional visual positioning requires the on-site collection and real-time matching of a large number of real-world environmental features, which is not only computationally intensive and slow in response, but also easily affected by environmental changes. The solution in this invention innovatively utilizes the precise replication capability of digital twin models to pre-render and generate a standardized multi-scale reference feature library. This transforms real-time on-site feature collection into pre-generated feature library matching, significantly reducing the terminal's computational load and response time. Simultaneously, the reference feature library can iterate synchronously with the updates to the digital twin model, adapting to the dynamic changes in the street scene.

[0053] Traditional positioning technologies only output coordinate information, which is separate from subsequent service generation. This solution utilizes multi-scale visual features not only for positioning but also rich cultural scene information (e.g., skyline corresponding to ancient building types, facade texture corresponding to architectural techniques, and ground paving corresponding to cultural relic levels). These features can be directly used as the core input to the context state vector, enabling the positioning process and context perception process to be completed simultaneously, achieving efficient coupling where positioning completion equals context recognition completion.

[0054] More preferably, the step of generating a corresponding location service response packet at the edge computing node based on the user abstract request and the current state of the lightweight semantic map includes: S104a: Determine the target service area in the digital twin model that matches the user abstract request; S104b: According to the expected movement direction or interest diffusion direction of the user terminal in the target service area, divide the target service area into multiple sequentially associated service segments; S104c: For each service segment, extract the corresponding segment context dataset from the lightweight semantic map. The segment context dataset includes the segment's real-time congestion level, environmental atmosphere, set of accessible points of interest, and its current state. S104d: For the first-order service segment, input the corresponding segment scenario dataset into the first service adaptation model to obtain the first service suitability score and recommended service content for the corresponding service segment; S104e: For each subsequent service segment that is not in the first order, input its corresponding segment context dataset and the service suitability score of its immediate upstream service segment into the corresponding subsequent service adaptation model to obtain the service suitability score and recommended service content of the subsequent segment. The score of the upstream segment serves as the context constraint for the service recommendation of the downstream segment.

[0055] This invention, based on the user's expected movement direction or interest diffusion direction (such as extending from the street entrance to the core cultural relics area, or transitioning from the folk experience area to the cultural and creative shopping area), divides the target service area into sequentially related service segments. This addresses the pain point of traditional service planning, which takes the entire area as a unit and ignores the contextual differences between areas. For example, the tour route of a historical and cultural block can be divided into an entrance guidance segment, an ancient building viewing segment, an intangible cultural heritage experience segment, and a cultural and creative consumption segment. Each segment corresponds to different service objectives and contextual characteristics, making the service planning more aligned with the user's tour rhythm and changing needs.

[0056] Each service segment extracts a unique segmented contextual dataset (real-time congestion, environmental atmosphere, point of interest status, etc.) to avoid decision-making interference caused by mixed global data. For example, core contextual data such as low noise, low congestion, and the open status of cultural relics are extracted for the ancient building viewing segment, while key information such as queue time for experience projects and the on-duty status of interactive staff are extracted for the intangible cultural heritage experience segment, so that service recommendations for each segment are based on its unique scenario characteristics.

[0057] When generating recommended content, service segments that are not prioritized in the first order use the suitability score of the adjacent upstream segment as a contextual constraint, overcoming the limitations of traditional segmented services where each segment makes independent decisions and lacks a link connection. The sequential association structure of multiple service segments enables location services to switch smoothly as the user moves. For example, when a user moves from the entrance guide segment to the ancient building viewing segment, the system can adjust the level of detail and recommendation priority of the downstream segment's content based on the user's dwell time and interaction feedback in the upstream segment, avoiding the feeling of disjointed service switching and improving the smoothness of the tour experience.

[0058] In practical implementation, the overall service planning is broken down into multiple independent computational tasks. Edge computing nodes do not need to process the massive amount of contextual data across the entire domain at once; they only need to perform modeling and computation for the current and one or two downstream segments. The segmented contextual datasets are extracted directly from the lightweight semantic map, without needing to access the overall data of the digital twin model, significantly reducing the amount of data read and transmitted by edge nodes. At the same time, the structured features of the segmented data are more easily processed by the service adaptation model, further improving the model's computational efficiency and shortening the service response packet generation time.

[0059] Different service segments are configured with corresponding service adaptation models, and the model weights can be optimized for the core service objectives of each segment (such as viewing experience and consumption). For example, the model for the ancient architecture viewing segment emphasizes the weights of cultural experience value and crowd avoidance, while the model for the cultural and creative consumption segment emphasizes the weights of user preference matching and cost-effectiveness, ensuring that the recommended content for each segment accurately matches the core needs of users at that stage. The segmented contextual dataset contains real-time updated information such as congestion levels and point-of-interest status, and the service adaptation model can adjust the recommendation strategy in real time based on this dynamic data. For example, when the congestion level of a certain service segment suddenly increases, the model can immediately lower the service suitability score of that segment and push alternative routes or experience projects to downstream segments, preventing users from falling into unpleasant experience scenarios and improving the dynamic anti-interference capability of the service.

[0060] On the edge computing node side, refined service planning and continuous service recommendation are achieved, which not only solves the technical pain points of high computing power consumption and low adaptation accuracy of traditional full-domain planning, but also ensures the smoothness and continuity of the user's browsing experience through the correlation constraints of upstream and downstream segments.

[0061] More preferably, the location service method further includes: After generating the location service response package, monitor the user's actual interaction feedback on each service segment during the execution process; When the actual feedback of a certain service segment is significantly lower than its predicted service suitability score, the segment is marked as a segment to be optimized. Analyze the service suitability scores of the adjacent upstream segments of the segment to be optimized and the actual behavior of users in the upstream segments, and calculate the influence contribution of the upstream segments to the segment to be optimized; If the impact contribution is greater than or equal to the preset contribution threshold, the upstream segment is updated and marked as a new segment to be optimized, and the analysis continues to trace upstream. If the impact contribution is less than the preset contribution threshold, then the current segment to be optimized is determined as the actual optimized service segment.

[0062] Traditional service optimization only adjusts strategies for single service segments with poor feedback, ignoring the cascading effects of upstream segments on downstream segments (e.g., congestion in upstream segments leads to a decline in user experience, which in turn affects the interaction feedback of downstream segments). This invention introduces an impact contribution calculation, quantifying the impact weight of upstream segments on downstream experiences by analyzing the service suitability scores and actual user behavior of adjacent upstream segments. For example, if the downstream intangible cultural heritage experience segment has poor feedback, and the calculated impact contribution of the upstream ancient architecture viewing segment—causing congestion leading to unexpectedly long waiting times and excessive physical exertion—reaches a threshold, then the upstream segment is marked as a segment to be optimized. This addresses the root cause of the poor service experience problem, avoiding ineffective optimization that only treats the symptoms.

[0063] When the impact contribution falls below a preset threshold, upstream tracing is immediately terminated, and the current segment is identified as the actual optimization target. This ensures accurate identification of the root cause of optimization while avoiding meaningless end-to-end tracing. The sequential correlation of service segments means that the upstream experience directly affects the downstream user state and needs (e.g., if the upstream segment's explanations are dull, leading to decreased user interest, the downstream cultural and creative consumption segment's willingness to purchase will also decrease). This solution can eliminate such hidden interference by tracing the impact of upstream segments. For example, if the recommended route in the upstream entrance guide segment is circuitous, causing user fatigue and resulting in low feedback ratings for the downstream core cultural relic viewing segment, then the route planning of the upstream segment can be optimized to shorten the walking distance. This ensures the continuity of the user experience at the link level and solves the pain points of traditional segmented services where each segment is independent and the experience is fragmented.

[0064] This mechanism expands the scope of optimization from a single segment to the entire service chain, ensuring that the service strategies of each segment synergize with upstream and downstream elements. For example, the optimized ancient architecture viewing segment ensures visitor comfort by controlling visitor density, while the downstream intangible cultural heritage experience segment can recommend more interactive experiences based on positive user feedback, forming a positive experience cycle of upstream preparation and downstream reinforcement.

[0065] The monitored segmented interaction feedback and impact contribution data can serve as training samples for optimizing the service adaptation model. For example, in cases where upstream congestion leads to poor downstream feedback, the system can optimize the upstream segment's visitor flow management strategy and the congestion weight in the service adaptation model; for cases where insufficient upstream content attractiveness leads to a decline in downstream experience, the system can update the upstream segment's guide content and point-of-interest recommendation rules. This iterative approach based on actual feedback enables the system to continuously adapt to changes in the historical and cultural district's scenarios (such as temporary events and seasonal visitor flow fluctuations), achieving dynamic optimization of service strategies.

[0066] Based on precise root cause segmentation, the system only needs to adjust service strategies for the actual optimized segment and its directly related upstream segments, without requiring batch optimization of all segments across the entire chain. This optimization model significantly reduces the computational load on edge nodes for model retraining and policy updates, ensuring stable system operation in high-concurrency scenarios. The calculation of contribution relies solely on the scores and user behavior data of adjacent upstream segments, eliminating the need to retrieve historical feedback data from the entire domain, resulting in a simple and lightweight calculation logic.

[0067] More preferably, the location service method further includes: After determining the actual optimized service segments, a multi-dimensional feature vector is extracted for each actual optimized service segment. The feature vector includes the location in the service path, physical space features, real-time context status, and associated user profile tags. The multidimensional feature vector is input into the service strategy matching engine, and dynamic optimization strategies that match the segment features are retrieved from the pre-generated service optimization strategy library; wherein, the strategies in the service optimization strategy library include path detour strategies, service content replacement strategies, service timing adjustment strategies, and guidance enhancement strategies; For the retrieved dynamic optimization strategy, determine whether it contains an instant adjustment strategy that can be executed in real time by the edge computing node; If included, the edge computing node generates corresponding control parameters and updates the location service response packet pushed to the user terminal in real time, guiding the user to execute the optimized path or experience content in the current session; After users complete the optimized experience, new feedback on the service segment to be optimized is collected, and this new feedback is compared with the historical feedback before optimization to generate an optimization effect evaluation report for iterative updates to the service strategy matching engine or the service optimization strategy library.

[0068] The system extracts multi-dimensional feature vectors (path location, physical space features, real-time contextual state, and user profile tags) for actual service segmentation optimization, overcoming the limitations of traditional single-dimensional matching optimization strategies and achieving a three-dimensional and refined characterization of segments. The service optimization strategy library includes four categories of strategies: path detour, content replacement, timing adjustment, and guidance enhancement, providing differentiated solutions for segmentation problems with different characteristics. For example, for segments with high congestion and core cultural relic areas, a path detour strategy (guiding to alternative viewing routes) and a guidance enhancement strategy (real-time congestion warning push) are matched; for segments with insufficient content appeal and family-oriented users, a service content replacement strategy (replacing static explanations with interactive intangible cultural heritage experiences) is matched. Edge computing nodes can directly identify and execute real-time adjustment strategies from the strategy library without relying on cloud computing power, updating location service response packages in real-time during the user's current browsing session. For example, when a user is browsing a highly congested segment, the edge node can immediately push detour routes and simultaneously update the guide content, preventing the user from missing optimization opportunities while waiting for cloud instructions.

[0069] The control parameters generated by the edge nodes only contain core information such as path adjustment coordinates and content replacement identifiers, eliminating the need to transmit redundant data and allowing for direct parsing and execution by the user terminal. This lightweight design reduces communication costs between the edge nodes and the terminal while avoiding performance overload caused by processing complex data, ensuring the smooth implementation of optimization strategies. The strategy, based on multi-dimensional feature vector matching, fully considers the user's current profile tags and contextual state. For example, optimization strategies for elderly users prioritize accessible detour paths and a slow-paced guided tour, while strategies for parent-child users prioritize interactive content replacement and rest stop guidance, ensuring that the optimization effect closely matches the user's real-time needs and avoiding a decline in experience caused by a one-size-fits-all approach to optimization.

[0070] By comparing user feedback before and after optimization, an evaluation report is generated, which quantifies the effectiveness of the optimization strategy (such as the percentage reduction in congestion, the increase in user satisfaction, and the effect of optimizing dwell time). For example, if user satisfaction increases from 60% to 90% after optimization of a certain segment, it can be determined that the path detour and guidance enhancement strategies are effective. If there is no significant improvement in feedback after optimization, it can be identified as a strategy matching deviation, providing a basis for subsequent strategy library updates. The evaluation report can be used simultaneously to iterate the service strategy matching engine and the service optimization strategy library: for strategies with accurate matching, their matching weight in the engine is increased; for ineffective strategies, they are removed from the strategy library or optimized; and for newly emerging segment characteristics, new optimization strategies are added. This long-term iterative mechanism enables the system to continuously adapt to changes in the historical and cultural district's scenarios and the evolution of user needs, achieving a spiral increase in service capabilities.

[0071] More preferably, the segmented context dataset further includes spatiotemporal attribute data, which is determined in the following manner: The deviation between the average pedestrian flow pattern in the same historical period and the current pedestrian flow in this segment; The cultural experience intensity index of each point of interest within the segment is calculated based on its historical events and real-time status; the visual presentation quality score of the segment under current lighting and weather conditions; wherein, when calculating the service suitability score, the service adaptation model comprehensively weighs the real-time factor, the cultural experience factor, and the environmental comfort factor.

[0072] This invention introduces deviation data between historical average pedestrian flow patterns and current pedestrian flow in different segments, overcoming the limitations of traditional solutions that rely solely on real-time passenger density and enabling the prediction of passenger flow trends. For example, if a segment has a low passenger flow pattern in the same historical period, but the current passenger flow deviation rate reaches 150%, the system can determine that the area is in a phase of rapid passenger influx, triggering congestion warnings and route adjustment strategies in advance, rather than passively responding after passenger flow saturation, significantly improving the forward-looking nature of the service.

[0073] The cultural experience intensity index, calculated based on historical events of points of interest (such as the exhibition cycle of cultural relics and the frequency of intangible cultural heritage activities) and real-time status (such as open / closed status and the on-duty status of interactive staff), transforms abstract cultural values ​​into quantifiable model input parameters. For example, if a section of ancient architecture hosts a temporary special exhibition of cultural relics, its cultural experience intensity index increases from the normal value of 0.5 to 0.9. When calculating the suitability score, the service adaptation model will prioritize increasing the recommendation weight of this section to ensure that users do not miss high-value cultural scenes.

[0074] The visual presentation quality score under current lighting and weather conditions addresses the pain point that the landscape experience of historical and cultural blocks is greatly affected by the environment. For example, the visual presentation quality score of outdoor ancient building sections decreases under rainy weather, and the model can prioritize recommending alternative scenes such as indoor intangible cultural heritage exhibition halls; at dusk, when the lighting is soft, the model can increase the recommendation priority of night view appreciation sections, making the service plan highly adapted to environmental conditions.

[0075] The service adaptation model comprehensively weighs real-time factors (human flow deviation, congestion changes), cultural experience factors (cultural experience intensity index, value of points of interest), and environmental comfort factors (visual presentation quality, noise, temperature), constructing a multi-objective optimization scoring system. This design avoids the drawbacks of traditional models that prioritize culture over experience or accessibility over value. For example, for families, the model can appropriately increase the weight of environmental comfort factors, prioritizing segments with low congestion and suitable lighting; for history enthusiasts, the model can increase the weight of cultural experience factors, including segments with slightly higher human flow in the recommendations.

[0076] The weights of the three-dimensional factors can be flexibly adjusted based on user profile tags to achieve personalized scoring. Combining the trend of visitor flow deviation with the cultural experience intensity index, the model can accurately recommend the optimal time to visit for each segment. For example, if the cultural experience intensity index of a certain intangible cultural heritage experience segment peaks at 10:00 AM, and the current visitor flow deviation is -30% (low visitor flow), the model will recommend that users visit during this time to avoid the afternoon peak and ensure that users receive a high-value, low-crowding experience.

[0077] More preferably, after determining the target service area in the digital twin model that matches the user abstract request, the method further includes: The edge computing node parses the abstract request to determine its dominant request pattern; based on the determined dominant request pattern, it invokes the service segmentation strategy associated with that pattern to divide the multiple sequentially associated service segments. When the dominant request mode is the first request mode, the optimal path from the starting point to the definite destination is calculated on the lightweight semantic map; topological key points on the path are identified, including road intersections, significant terrain change points, and necessary landmark side points; the path is divided into multiple service segments with the topological key points as boundaries, and the navigation complexity value is calculated for each segment. When the dominant request mode is the second request mode, the real-time interest attraction value of each region in the lightweight semantic map is calculated based on the interest preference vector in the user's abstract request; continuous regions with attraction values ​​higher than a threshold are clustered to form multiple core interest region blocks; a virtual access sequence flow is defined for the multiple core interest region blocks based on geographical proximity and interest correlation; each core interest region block and its surrounding buffer area are defined as a corresponding service segment, and the order of the segments is determined according to the virtual access sequence flow. When the dominant request mode is the third request mode, the multiple subtasks defined in the user abstract request are parsed into a corresponding task dependency graph. In the task dependency graph, nodes represent subtasks, and edges represent execution order dependencies or spatial proximity relationships. The task dependency graph is topologically sorted, and subtask clusters that can be executed in parallel or are logically closely related are divided into the same service segment. Based on the task dependency relationship, the execution order of each service segment is determined.

[0078] Edge computing nodes determine the dominant request mode by parsing abstract requests. For the three core needs of path navigation (first mode), interest exploration (second mode), and task execution (third mode), they configure exclusive segmentation strategies for each, which solves the pain points of traditional one-size-fits-all segmentation and disconnection from the core needs of users.

[0079] The first mode (path navigation) accurately matches the need for efficient travel from the starting point to the destination, segmenting the route to fit the topology and ensuring clarity and continuity of navigation; the second mode (interest exploration) adapts to the user's need for in-depth exploration and preference-oriented navigation, segmenting the route around the core interest areas to enhance the immersive experience of the cultural experience; the third mode (task execution) meets the need for the orderly completion of multiple sub-tasks, segmenting the route to fit the task dependency logic and ensuring the efficiency and logic of task execution.

[0080] By pre-associating corresponding service segmentation strategies with different dominant request patterns, edge computing nodes do not need to redesign segmentation logic and can directly call the matching strategy to complete the segmentation, thus adapting to the low-latency operation requirements of edge computing.

[0081] In the first mode, the system divides the route into segments using topological key points such as road intersections, terrain change points, and landmark side points as boundaries, and calculates the navigation complexity value for each segment. This solves the problem of vague navigation guidance and users easily getting lost on long routes. For example, in the complex alleyways of historical and cultural districts, the system uses the intersection of archways and landmarks as segment boundaries, and adjusts the level of detail in the guided tour based on the navigation complexity value (e.g., high complexity in narrow alleyways and low complexity in main roads). Segments with high complexity receive more voice guidance, while segments with low complexity push cultural content, balancing navigation accuracy and user experience richness.

[0082] In the second mode, regional attractiveness values ​​are calculated based on interest preference vectors. Highly attractive continuous areas are clustered into core interest zones, and the access sequence flow is defined by combining geographical proximity and interest relevance, ensuring that segmented planning perfectly aligns with users' interest orientation. For example, for users who prefer intangible cultural heritage handicrafts, the system will cluster paper-cutting, wood carving, and pottery experience areas within the block into core segments and plan a sequence flow of viewing, experiencing, and purchasing cultural and creative products, avoiding users having to backtrack between different interest areas and improving tour efficiency and experience continuity. At the same time, each core zone is accompanied by surrounding buffer areas (such as rest areas and cultural and creative shops), further enhancing the service completeness of the scene.

[0083] In the third mode, a task dependency graph is constructed by parsing subtasks. Based on topological sorting, parallel or strongly related subtasks are clustered into segments, ensuring that the segment order aligns with the task execution logic. This solves the problems of process confusion and task omissions in multi-task scenarios. For example, for a composite task involving visiting a cultural relics exhibition, experiencing intangible cultural heritage, and purchasing cultural and creative products, the system treats visiting the cultural relics exhibition as an independent segment, clusters the intangible cultural heritage experience and the purchase of cultural and creative products (spatially adjacent and logically related) into the same segment, and plans the segment execution process according to the dependency order of visiting and experiencing / purchasing. For subtasks that can be parallelized (such as booking an intangible cultural heritage experience and querying the location of a cultural and creative product store), they are integrated into the same segment for synchronous processing, significantly improving task completion efficiency.

[0084] The segmentation strategies corresponding to the three request modes are all modularly designed, allowing edge computing nodes to directly call the appropriate mode without needing to traverse and analyze the entire dataset. For example, when processing a request in the first mode, only the path topology data needs to be retrieved; when processing a request in the second mode, only the interest attraction value needs to be calculated, avoiding the processing of irrelevant data.

[0085] Key segmentation parameters in each mode (such as navigation complexity, interest attraction, and task-dependent topology) can be quickly calculated based on structured data from a lightweight semantic map, without relying on the full-domain high-precision data of a digital twin model. This design ensures that edge nodes can maintain a segmentation planning response speed of hundreds of milliseconds even when multiple users make concurrent requests in historical and cultural districts.

[0086] The core of dividing user requests into distinct modes is a precise categorization of user needs: users choosing mode one prioritize efficient access, mode two prioritize in-depth experience, and mode three prioritize task completion. This mode-based segmentation strategy can specifically address the core needs of different users, enabling personalized planning. The segmentation order in all three modes follows a clear logic (path topology logic, interest association logic, and task dependency logic), avoiding disordered segmentation. For example, the segmentation order in mode two is planned based on interest association, ensuring a smooth transition from basic cultural appreciation to in-depth interactive experiences; the segmentation order in mode three is planned based on task dependency relationships, ensuring a seamless transition from preceding tasks to subsequent tasks, significantly improving the smoothness and satisfaction of the service experience.

[0087] In addition to the above models, there is a fourth model, which employs a social field strength segmentation method. Specifically, the location or virtual imprint popularity of other users is modeled as social field strength in space. Users are guided to move along directions where social field strength increases. The path of social interaction is dynamically optimized, creating opportunities for chance encounters or efficient browsing.

[0088] Since the user's actual intentions may be mixed, dynamic mode fusion and switching can also be configured. A request can be parsed into a weighted combination of multiple modes (e.g., 60% experience + 40% social). The segmentation strategy is the weighted fusion result of the corresponding strategies. Based on the real-time context (e.g., a user suddenly lingers at a non-task point for an extended period), the system can dynamically adjust the weight of the dominant mode and smoothly switch segmentation strategies.

[0089] Specifically, the method supports request pattern fusion: parsing the abstract request into a probability distribution or weight combination of multiple basic request patterns; executing multiple specific service segmentation strategies in parallel or sequentially according to the weight combination; and using a strategy fusion algorithm to integrate the segmentation schemes generated by different strategies to generate a fusion final service segmentation scheme.

[0090] The method also includes dynamic mode switching based on real-time context: during service execution, the changes in the state of the first context are continuously monitored; when a behavior pattern that deviates significantly from the current dominant request pattern is detected, the weight or type of the dominant request pattern is re-evaluated and updated; based on the updated pattern, the subsequent service segments are dynamically re-divided, and users are smoothly guided to switch strategies.

[0091] The method in this embodiment of the invention upgrades location services from a tool to a cultural experience carrier through a technical chain of environmental perception, context analysis, edge computing, and precise response. It not only solves the pain points of complex signals and diverse service needs in historical and cultural blocks, but also realizes the digital revitalization of cultural resources through digital twin technology, ultimately achieving the dual goals of improving technical efficiency and deepening cultural experience; thus enhancing the user's tour experience.

[0092] Example 2 Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a location service system based on digital twin technology disclosed in an embodiment of the present invention. Figure 5 As shown, the location service system based on digital twin technology may include: Receiving module 21: Used to receive digital beacons broadcast by environmental sensing devices installed in the historical and cultural district, wherein the digital beacons include the environmental status of the area in which they are located; Generation module 22: used to determine the contextual state information and location information of the received digital beacon in the digital twin model; and to generate a corresponding user abstract request based on the contextual state information and the user's historical preferences; Edge computing module 23: used to send the user abstract request to the edge computing node at the edge of the corresponding historical and cultural block network, the edge computing node being used to maintain a lightweight semantic map derived from the digital twin model; Interactive service module 24: is used to generate a corresponding location service response package at the edge computing node based on the user abstract request and the current state of the lightweight semantic map; send the location service response package to the corresponding user terminal, and have the user terminal execute the response package to provide the user with location services deeply integrated with the historical and cultural blocks.

[0093] The method in this embodiment of the invention upgrades location services from a tool to a cultural experience carrier through a technical chain of environmental perception, context analysis, edge computing, and precise response. It not only solves the pain points of complex signals and diverse service needs in historical and cultural blocks, but also realizes the digital revitalization of cultural resources through digital twin technology, ultimately achieving the dual goals of improving technical efficiency and deepening cultural experience; thus enhancing the user's tour experience.

[0094] Example 3 Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device can be a computer, a server, etc. Of course, in certain cases, it can also be a mobile phone, tablet computer, monitoring terminal, or other smart device, as well as an image acquisition device with processing capabilities. Figure 6As shown, the electronic device may include: Memory 510 storing executable program code; Processor 520 coupled to memory 510; The processor 520 calls the executable program code stored in the memory 510 to execute some or all of the steps in the location service method based on digital twin technology in Embodiment 1.

[0095] This invention discloses a computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps in the location service method based on digital twin technology in Embodiment 1.

[0096] The above provides a detailed description of the location service method, system, electronic device, and storage medium based on digital twin technology disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A location service system based on digital twin technology, characterized in that, include: Receiving module: Used to receive digital beacons broadcast by environmental sensing devices installed in the historical and cultural district, wherein the digital beacons include the environmental status of the area in which they are located; Generation module: used to determine the contextual state information and location information of the received digital beacon in the digital twin model; And generate corresponding abstract user requests based on the contextual state information and the user's historical preferences; Edge computing module: used to send the user abstract request to the edge computing node at the edge of the corresponding historical and cultural block network, the edge computing node is used to maintain a lightweight semantic map derived from the digital twin model; Interactive service module: used to generate a corresponding location service response package at the edge computing node based on the user abstract request and the current state of the lightweight semantic map; send the location service response package to the corresponding user terminal, and have the user terminal execute the response package to provide the user with location services deeply integrated with the historical and cultural blocks; The step of generating a corresponding location service response package at the edge computing node based on the user abstract request and the current state of the lightweight semantic map includes: Determine the target service area in the digital twin model that matches the user's abstract request; According to the user terminal's expected movement direction or interest diffusion direction within the target service area, the target service area is divided into multiple sequentially associated service segments; For each service segment, a corresponding segment context dataset is extracted from the lightweight semantic map. The segment context dataset includes the segment's real-time congestion level, environmental atmosphere, set of accessible points of interest, and its current state. For the first-order service segment, the corresponding segment scenario dataset is input into the first service adaptation model to obtain the first service suitability score and recommended service content for the corresponding service segment. For each subsequent service segment that is not in the first order, its corresponding segment context dataset and the service suitability score of its immediate upstream service segment are input into the corresponding subsequent service adaptation model to obtain the service suitability score and recommended service content for that subsequent segment. The score of the upstream segment serves as the context constraint for the service recommendation of the downstream segment.

2. A location service method based on digital twin technology, characterized in that, include: Receives digital beacons broadcast by environmental sensing devices installed in historical and cultural districts, the digital beacons including the environmental status of their respective areas; The contextual state information and location information of the received digital beacon in the digital twin model are determined; and a corresponding user abstract request is generated based on the contextual state information and the user's historical preferences. The user abstract request is sent to the edge computing node at the edge of the corresponding historical and cultural block network. The edge computing node is used to maintain a lightweight semantic map derived from the digital twin model. The edge computing node generates a corresponding location service response package based on the user abstract request and the current state of the lightweight semantic map; the location service response package is sent to the corresponding user terminal, and the user terminal executes the response package to provide the user with location services deeply integrated with the historical and cultural blocks; The step of generating a corresponding location service response package at the edge computing node based on the user abstract request and the current state of the lightweight semantic map includes: Determine the target service area in the digital twin model that matches the user's abstract request; According to the user terminal's expected movement direction or interest diffusion direction within the target service area, the target service area is divided into multiple sequentially associated service segments; For each service segment, a corresponding segment context dataset is extracted from the lightweight semantic map. The segment context dataset includes the segment's real-time congestion level, environmental atmosphere, set of accessible points of interest, and its current state. For the first-order service segment, the corresponding segment scenario dataset is input into the first service adaptation model to obtain the first service suitability score and recommended service content for the corresponding service segment. For each subsequent service segment that is not in the first order, its corresponding segment context dataset and the service suitability score of its immediate upstream service segment are input into the corresponding subsequent service adaptation model to obtain the service suitability score and recommended service content for that subsequent segment. The score of the upstream segment serves as the context constraint for the service recommendation of the downstream segment.

3. The location service method based on digital twin technology as described in claim 2, characterized in that, Determining the contextual state information of the received digital beacon in the digital twin model includes: The digital beacon is parsed to obtain the environmental attributes encoded therein, including noise level, crowd density estimate, lighting conditions, and point of interest identifiers. Combine data from the inertial measurement unit built into the user terminal to determine the user's activity status; The environmental attributes are fused with the activity state to form a vector representation of the contextual state information in the digital twin model.

4. The location service method based on digital twin technology as described in claim 2, characterized in that, The step of generating a corresponding location service response package based on the user abstract request and the current state of the lightweight semantic map includes: The user's abstract request is parsed into corresponding constraints and optimization objectives; In the lightweight semantic map, find available service resources or paths that can satisfy the constraints; From all the solutions that meet the conditions, select the solution that meets the optimization objective and encapsulate it into the location service response package; wherein, the optimization objective includes one or more of the following: shortest time, lowest energy consumption, highest cultural experience value, or strongest congestion avoidance. The location service method further includes: During the execution of the location service response packet by the user terminal, the deviation between the user's actual interaction feedback and the expected effect of the edge computing node is monitored; The detected deviation data is fed back to the corresponding server; The bias data is used at the server to optimize and train a machine learning model for identifying user context patterns or generating service components.

5. The location service method based on digital twin technology as described in claim 2, characterized in that, The location service method further includes: Image frames containing the environment of the historical and cultural district are acquired through the image sensor of the user terminal; From the image frames, visual environment features at different spatial scales are extracted; the extraction of visual environment features at different spatial scales includes: Extract the first scale feature, which is the skyline contour feature vector calculated based on the boundary line between the roof of the building and the sky in the image; Extract second-scale features, which are highly discriminative local texture or structural feature points and their descriptors detected from the building facade area in the image. Extract third-scale features, which are paving material patterns, special markings, or landform features identified from ground areas in the image; The extracted visual environment features are matched with a reference feature library corresponding to different spatial scales, which is pre-rendered from the digital twin model. Based on the matching results, the first position is calculated.

6. The location service method based on digital twin technology as described in claim 2, characterized in that, The location service method further includes: After generating the location service response package, monitor the user's actual interaction feedback on each service segment during the execution process; When the actual feedback of a certain service segment is significantly lower than its predicted service suitability score, the segment is marked as a segment to be optimized. Analyze the service suitability scores of the adjacent upstream segments of the segment to be optimized and the actual behavior of users in the upstream segments, and calculate the influence contribution of the upstream segments to the segment to be optimized; If the impact contribution is greater than or equal to the preset contribution threshold, the upstream segment is updated and marked as a new segment to be optimized, and the analysis continues to trace upstream. If the impact contribution is less than the preset contribution threshold, then the current segment to be optimized is determined as the actual optimized service segment.

7. The location service method based on digital twin technology as described in claim 2, characterized in that, The location service method further includes: After determining the actual optimized service segments, a multi-dimensional feature vector is extracted for each actual optimized service segment. The feature vector includes the location in the service path, physical space features, real-time context status, and associated user profile tags. The multidimensional feature vector is input into the service strategy matching engine, and dynamic optimization strategies that match the segment features are retrieved from the pre-generated service optimization strategy library; wherein, the strategies in the service optimization strategy library include path detour strategies, service content replacement strategies, service timing adjustment strategies, and guidance enhancement strategies; For the retrieved dynamic optimization strategy, determine whether it contains an instant adjustment strategy that can be executed in real time by the edge computing node; If included, the edge computing node generates corresponding control parameters and updates the location service response packet pushed to the user terminal in real time, guiding the user to execute the optimized path or experience content in the current session; After users complete the optimized experience, new feedback on the actual optimized service segments is collected, and this new feedback is compared with the historical feedback before optimization to generate an optimization effect evaluation report for iterative updates to the service strategy matching engine or the service optimization strategy library.

8. The location service method based on digital twin technology as described in claim 2, characterized in that, The segmented context dataset also includes spatiotemporal attribute data, which is determined in the following manner: The deviation between the average pedestrian flow pattern in the same historical period and the current pedestrian flow in this segment; The cultural experience intensity index of each point of interest within the segment is calculated based on its historical events and real-time status; the visual presentation quality score of the segment under current lighting and weather conditions; wherein, when calculating the service suitability score, the service adaptation model comprehensively weighs the real-time factor, the cultural experience factor, and the environmental comfort factor.

9. The location service method based on digital twin technology as described in claim 2, characterized in that, After determining the target service area in the digital twin model that matches the user abstract request, the method further includes: The edge computing node parses the abstract request to determine its dominant request pattern; based on the determined dominant request pattern, it invokes the service segmentation strategy associated with that pattern to divide the multiple sequentially associated service segments. When the dominant request mode is the first request mode, the optimal path from the starting point to the definite destination is calculated on the lightweight semantic map; topological key points on the path are identified, including road intersections, significant terrain change points, and necessary landmark side points; the path is divided into multiple service segments with the topological key points as boundaries, and the navigation complexity value is calculated for each segment. When the dominant request mode is the second request mode, the real-time interest attraction value of each region in the lightweight semantic map is calculated based on the interest preference vector in the user's abstract request; continuous regions with attraction values ​​higher than a threshold are clustered to form multiple core interest region blocks; a virtual access sequence flow is defined for the multiple core interest region blocks based on geographical proximity and interest correlation; each core interest region block and its surrounding buffer area are defined as a corresponding service segment, and the order of the segments is determined according to the virtual access sequence flow. When the dominant request mode is the third request mode, the multiple subtasks defined in the user abstract request are parsed into corresponding task dependency graphs. In the task dependency graph, nodes represent subtasks, and edges represent execution order dependencies or spatial proximity relationships. The task dependency graph is topologically sorted, and subtask clusters that can be executed in parallel or are logically closely related are divided into the same service segment. Based on the task dependency relationship, the execution order of each service segment is determined.

Images