Intelligent guide explanation system and method for museum
By using real-time location monitoring and traffic awareness, combined with visitor preferences to optimize routes, and providing personalized voice interaction services, the problems of overcrowding and limited content in museum tour guide systems during peak hours have been solved, improving visitor experience and resource utilization efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-12
- Publication Date
- 2026-03-27
AI Technical Summary
Existing museum tour guide systems are prone to overcrowding and waiting during peak hours. They lack real-time location monitoring and traffic awareness, the content of the explanations is fixed and cannot be personalized, and the audio function is missing, resulting in low visitor satisfaction.
The system uses a device distribution and location monitoring module to monitor tourist locations in real time, generate traffic heat maps, dynamically optimize routes based on tourist preferences, and provide personalized services by integrating TTS and NLP modules through voice interaction to identify needs, forming a closed-loop feedback mechanism.
Effectively avoid peak-hour congestion, improve resource utilization and tourist satisfaction, provide personalized routes and intelligent dialogue services, enhance immersion, reduce operating costs, and improve multilingual support and data privacy protection.
Smart Images
Figure CN121745429A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of path planning, more particularly, the present application relates to a museum intelligent tour guide system and method. BACKGROUND
[0002] At present, museum tour usually relies on artificial tour guides or fixed audio equipment, which can easily lead to congestion and waiting problems during peak tourist season. Although the existing system has an APP tour, it lacks real-time location monitoring and traffic awareness, and cannot dynamically avoid congested areas. At the same time, the explanation content is fixed and cannot be interacted according to the needs of tourists, resulting in a single experience. The lack of voice function cannot understand the preferences of tourists through conversation and provide personalized services. These problems cause low tourist satisfaction, uneven use of museum resources, and affect overall operational efficiency.
[0003] Based on the above problems, a museum intelligent tour guide system and method are proposed. SUMMARY
[0004] In order to overcome the above-mentioned defects of the prior art, in order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a museum intelligent tour guide system, comprising:
[0005] Device distribution and position monitoring module: for distributing receiving devices and monitoring the position of each receiving device in real time to generate position data;
[0006] Visitor flow perception module: for obtaining the number of visitors in each area, and generating a traffic heat map in combination with the position data;
[0007] Dynamic route optimization module: generating the best tour route based on the traffic heat map and visitor preferences, and dynamically adjusting;
[0008] Voice interaction and explanation module: for providing artifact explanations through receiving devices and supporting voice conversations, identifying visitor needs and providing services.
[0009] Preferably, the specific implementation method of the device distribution and position monitoring module comprises:
[0010] When the visitor enters the museum, the receiving device is distributed;
[0011] The position data is collected in real time by the positioning sensor built-in the receiving device and uploaded to the central server;
[0012] The central server aggregates all the position data to generate a global position map.
[0013] Preferably, the specific implementation method of the visitor flow perception module comprises:
[0014] The system obtains real-time location data from a central server, divides the museum map into multiple areas using a grid partitioning method, counts the receiving devices in each area, calculates visitor density, and applies a sliding window to smooth the data.
[0015] Using GIS tools, the number and density of visitors are mapped onto the museum map to create a traffic heat map. The heat map uses color gradients and overlays area boundaries and artifact markers.
[0016] The heatmap includes timestamps and congestion scores, is updated regularly, and is pushed to receiving devices via API.
[0017] Preferably, the specific implementation method of the dynamic route optimization module includes:
[0018] A graph model is constructed with nodes as exhibition areas and edges as path distances. Priorities are added to each node according to the order of cultural relic visits, and visitor preference matching degree is added to each node, with physical distance as the initial weight.
[0019] The edge weights are adjusted using heatmap data. The adjusted edge weights = initial weights + (heatmap congestion score × penalty factor). The attraction score is calculated as the reward for path optimization, combining tourist preference matching degree and cultural relic priority. The attraction score = tourist preference matching degree × node priority × reward factor. The tourist preference matching degree is the cosine similarity between the tourist preference vector and the node type vector. The attraction score is subtracted from the adjusted edge weights to obtain the final edge weights.
[0020] Run the A* algorithm on the adjusted graph model to search for the optimal path from the starting point to the ending point. Combine the actual cost g(n) and the heuristic cost h(n) as the cost function, i.e., total cost = g(n) + h(n), where g(n) is the cumulative edge weight to the current node, and the heuristic cost h(n) = initial weight + (heatmap congestion score × 1.5) - attraction score; set the constraint that the total path duration < a specified time, and output a list of path sequences.
[0021] If congestion occurs, the route is dynamically adjusted by recalculating using the A* algorithm, prioritizing low-density alternative paths, and updating the receiving device.
[0022] Preferably, the specific implementation method of the voice interaction and explanation module includes:
[0023] The explanation of cultural relics is triggered based on location data, and the audio is played using TTS technology.
[0024] We analyze tourist queries using speech recognition; we identify tourist needs using NLP models and provide services based on those needs.
[0025] Preferably, the receiving device is also used to collect tourist satisfaction feedback; analyze the feedback using a machine learning model, update tourist preferences and parameters, and form a closed loop.
[0026] Preferably, the voice dialogue method includes:
[0027] Receive voice input from tourists; convert it to text using ASR; generate a response through a dialogue model and output it as voice.
[0028] Preferably, the heat map congestion score is obtained from heat map data, which is a comprehensive output generated by the traffic flow sensing module, including region ID, number of tourists, density, timestamp, congestion score and other auxiliary data;
[0029] The maximum capacity, time factor, and balance weight of each area are obtained from other auxiliary data. The congestion score is calculated as (current density / maximum capacity) × 100 + time factor × weight. The calculated congestion score is added to the heat map data and visualized using GIS tools.
[0030] Preferably, in the tourist preference vector and the node type vector, the tourist preference vector is a numerical vector that represents the intensity of tourists' interest in different types of cultural relics. The vector dimension is equal to the number of cultural relic categories, and the value range of each element is [0,1]. The higher the value, the greater the interest.
[0031] The node type vector represents the type representation vector for each exhibition area, and its dimension is the same as that of the visitor preference vector.
[0032] A smart guided tour system for museums includes:
[0033] S1: When visitors enter the museum, receiving devices are distributed to collect location data in real time. The central server aggregates all location data and generates a global location map.
[0034] S2: Divide the museum map into multiple areas using a grid partitioning method, calculate the visitor density in each area, generate a traffic heat map using GIS tools, and push it to the receiving device via API;
[0035] S3: Construct a graph model, add priority and tourist preference matching degree to each node, adjust the edge weights using heatmap data, run the A algorithm on the adjusted graph model, and output a list of path sequences; if congestion occurs, dynamically adjust the route and update it to the receiving device.
[0036] S4: Triggering explanations of cultural relics based on location data, and playing audio using TTS technology;
[0037] S5: Collect tourist satisfaction feedback through receiving devices, update tourist preferences and parameters, and form a closed loop.
[0038] The technical effects and advantages of the intelligent guided tour method for museums proposed in this invention are as follows:
[0039] By dynamically optimizing tour routes through real-time location monitoring and traffic heatmap generation, congestion during peak tourist seasons can be effectively avoided. Compared to traditional fixed guided tours, this method combines heatmap congestion scoring and A* algorithm to adjust path weights, reducing waiting time and improving museum resource utilization and visitor satisfaction.
[0040] By integrating visitor preference matching and cultural relic priority, personalized routes can be generated. Cosine similarity is used to calculate the matching degree and serve as a path reward factor to ensure that the routes are not only efficient but also in line with visitor interests. This addresses the pain point of existing systems having fixed explanation content and improves interactivity and repeat visit rates.
[0041] It supports voice interaction and narration modules, responding to visitor needs in real time through TTS and NLP models, providing details about artifacts and route information, forming an intelligent dialogue service. This enhances the immersiveness and convenience of guided tours, is suitable for multilingual environments, significantly reduces reliance on human guides, and saves museum operating costs.
[0042] The feedback loop mechanism uses machine learning to analyze satisfaction data, continuously updating preferences and system parameters to achieve adaptive optimization. This method forms a perception-optimization-feedback cycle, improving system accuracy, exhibiting good scalability, and applicability to multiple museum scenarios, bringing economic benefits such as increased visitor dwell time and ticket revenue.
[0043] The overall approach is computationally efficient and protects data privacy (such as anonymous IDs), overcoming the static design flaws of existing technologies, providing a secure and intelligent guided tour experience, and promoting the digital transformation of museums. Attached Figure Description
[0044] Figure 1 This is a schematic diagram of the structure of an intelligent guided tour explanation method for museums according to the present invention;
[0045] Figure 2 This is a schematic diagram illustrating the steps of an intelligent guided tour explanation method for museums according to the present invention. Detailed Implementation
[0046] 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.
[0047] Example 1
[0048] Please see Figure 1 and Figure 2 This embodiment provides an intelligent guided tour system and method for museums, which may specifically include the following steps:
[0049] The core problems with guided museum tours are overcrowding and a lack of personalization. Existing systems, such as audio guides, only play fixed content and cannot adjust routes or respond to queries in real time, leading to congestion during peak periods (such as queues in popular exhibits) and low satisfaction (such as ignoring visitor interests). The lack of location monitoring means that visitor flow cannot be detected, resulting in wasted resources.
[0050] For example, in a 1000-square-meter exhibition hall, if visitors concentrate on popular artifact areas while other areas remain empty, overall efficiency is low. The lack of voice interaction prevents dialogue to understand needs and provide services, such as handling visitors' immediate questions or adjusting preferences. These problems stem from static design, which cannot adapt to dynamic environments. Furthermore, insufficient data privacy and multilingual support further limit application.
[0051] To address the aforementioned issues, a smart guided tour system and methodology for museums was designed, including equipment distribution and location monitoring, visitor flow perception, dynamic route optimization, voice interaction and explanation, and feedback and optimization. These features enable real-time location monitoring, visitor flow perception, dynamic route generation, voice interaction, and feedback optimization, forming a complete smart guided tour loop. The system is deployed on a central server and supports mobile receiving devices (such as smart bracelets or apps), making it suitable for large museum environments.
[0052] Device distribution and location monitoring module: used to distribute receiving devices and monitor the location of each receiving device in real time, generating location data;
[0053] Receiving devices are distributed to visitors as they enter the museum;
[0054] The device collects location data in real time through its built-in positioning sensor and uploads it to a central server.
[0055] The central server aggregates all location data and generates a global location map.
[0056] Specifically, automated dispensing machines or manual counters are set up at the entrance. Visitors can obtain a device (such as a handheld guide device integrated with RFID and positioning chips, or linked to a mobile app) by scanning a QR code or swiping their ID card. Each device has a built-in unique ID (UUID format, such as "UUID-12345") for anonymous tracking. After issuance, the device is linked to the visitor via API (e.g., age, language preference, obtained from the registration form).
[0057] The device utilizes multimodal positioning technologies, including GPS (outdoor accuracy <5m), iBeacon Bluetooth beacon (indoor accuracy <1m), and UWB (high accuracy <0.5m, suitable for densely populated areas). Data is collected once per second, in JSON object format (e.g., {Device ID: UUID-12345, Longitude: 116.397, Latitude: 39.908, Timestamp: 2023-10-01T12:00:00}). Uploads are made using HTTPS protocol, transmitted to a central server via 5G or Wi-Fi, ensuring latency <100ms. The server uses load balancing (e.g., Nginx) to handle concurrent uploads, supporting over 1000 devices simultaneously during peak periods.
[0058] The server uses a NoSQL database (such as MongoDB) to store data. The aggregation process includes data cleaning (removing outliers, such as filtering coordinates using Z-scores) and map generation (generating a vector map based on OpenStreetMap or a custom museum floor plan, marking exhibition area boundaries). The map supports a real-time query API for easy access by other modules. Example: A museum has 10 exhibition areas; the server aggregates data every minute to generate a map containing the locations of 500 devices.
[0059] Visitor flow sensing module: used to obtain the number of visitors in each area and generate a flow heat map by combining location data;
[0060] The system obtains real-time location data from a central server, divides the museum map into multiple areas using a grid partitioning method, counts the receiving devices in each area, calculates visitor density, and applies a sliding window to smooth the data.
[0061] Using GIS tools, the number and density of visitors are mapped onto the museum map to create a traffic heat map. The heat map uses color gradients and overlays area boundaries and artifact markers.
[0062] The heatmap includes timestamps and congestion scores, is updated regularly, and is pushed to receiving devices via API. The heatmap congestion score is obtained from the heatmap data, which is a comprehensive output generated by the traffic awareness module, including area ID, number of visitors, density, timestamps, congestion scores, and other auxiliary data.
[0063] The maximum capacity, time factor, and balance weight of each area are obtained from other auxiliary data. The congestion score is calculated as (current density / maximum capacity) × 100 + time factor × weight. The calculated congestion score is added to the heat map data and visualized using GIS tools.
[0064] Specifically, this includes: obtaining real-time location data from a central server; dividing the museum map into multiple areas using a grid partitioning method (e.g., each area is 10m × 10m, with the grid size customized based on the building layout, and the total number of grids > 100); counting the number of receiving devices _t at the current time t within each area; and calculating the visitor density at the current time t (density _t = number of devices _t / area, unit: people / m²). 2 A sliding window algorithm (window size of 5 minutes, formula: smooth density = ∑(density_t × weight_t) / ∑weight_t, where weight_t is a time decay factor, weight_t = e^(-Δt / τ), Δt = 5 minutes, τ = 60s) is applied to smooth the data to reduce noise; for example, a sudden influx of tourists will not immediately trigger a high-density alarm. The counting process uses spatial indexes (such as R-trees) to accelerate queries, with a processing time of <1 second.
[0065] Import the museum vector map, overlay density data, and use a color gradient (green indicates density <0.5 people / m²). 2 Yellow indicates low density (0.5-1.0 for medium density), and red indicates high density (>1.0 for high density). The heatmap supports interaction (e.g., clicking on an area to display details) and can overlay area boundaries, artifact markers, and path lines. Example: An exhibition area density = 1.2 people / m² 2 The image is mapped to red and marked with "Congestion Warning".
[0066] Congestion score = (Current density / Maximum capacity) × 100 + Time factor × Weight, where maximum capacity is a preset value for the area, such as 20 people / area, the time factor is a peak period weighting, such as (10:00 AM - 12:00 PM) = 1.2, the weight is 0.2, the score range is 0-100, and it is updated regularly (once per minute, implemented through a background scheduled task such as a Cron Job) and pushed to receiving devices via API to ensure real-time availability. Example: Timestamp "2023-10-01T12:00:00", congestion score = 85, indicating high congestion, push notification "Recommended detour".
[0067] Dynamic route optimization module: Generates the best tour route based on traffic heatmaps and tourist preferences, and makes dynamic adjustments accordingly;
[0068] A graph model is constructed with nodes as exhibition areas and edges as path distances. Priorities are added to each node according to the order of cultural relic visits, and visitor preference matching degree is added to each node, with physical distance as the initial weight.
[0069] The edge weights are adjusted using heatmap data. The adjusted edge weights = initial weights + (heatmap congestion score × penalty factor). The attraction score is calculated as the reward for path optimization, combining tourist preference matching degree and cultural relic priority. The attraction score = tourist preference matching degree × node priority × reward factor. The tourist preference matching degree is the cosine similarity between the tourist preference vector and the node type vector. The attraction score is subtracted from the adjusted edge weights to obtain the final edge weights.
[0070] In the tourist preference vector and node type vector, the tourist preference vector is a numerical vector that represents the intensity of tourists' interest in different types of cultural relics. The vector dimension is equal to the number of cultural relic categories, and the value of each element ranges from [0,1]. The higher the value, the greater the interest.
[0071] The node type vector represents the type representation vector for each exhibition area, and its dimension is the same as that of the visitor preference vector.
[0072] Run the A* algorithm on the adjusted graph model to search for the optimal path from the starting point to the ending point. Combine the actual cost g(n) and the heuristic cost h(n) as the cost function, i.e., total cost = g(n) + h(n), where g(n) is the cumulative edge weight to the current node, and the heuristic cost h(n) = initial weight + (heatmap congestion score × 1.5) - attraction score; set the constraint that the total path duration < a specified time, and output a list of path sequences.
[0073] If congestion occurs, the route is dynamically adjusted by recalculating using the A* algorithm, prioritizing low-density alternative paths, and updating the receiving device.
[0074] Specifically, visitor preferences are obtained from user registration or historical data. For example, preferred cultural relic types are selected through questionnaires or apps. Cultural relic priorities are preset based on exhibit popularity, such as a priority of 0.8 for popular cultural relics. The data is loaded from the database, a graph model is constructed, and heat map data is input as edge weights for adjustment to generate an initial path sequence (e.g., starting point → exhibition area A → exhibition area C).
[0075] The total duration should not exceed 2 hours, the total distance should be less than 2km, and a Dijkstra variant should be used for checking. The algorithm runs on the server side with a time complexity of O(ElogV), where E is the number of edges and V is the number of nodes. Example: Input heatmap rating = 70, preference = historical artifacts, output total path cost = 1500 (in meters + penalty).
[0076] By recalculating path A*, low-density alternative paths are prioritized (3 alternative paths are pre-calculated using the branch and bound algorithm). The adjustment process takes less than 5 seconds, and the receiving device is notified of the update (e.g., a push notification saying "Route change: Detour to exhibition area B"). Example: If congestion is detected in exhibition area A, the route is adjusted to origin → exhibition area D → exhibition area A, reducing waiting time by 20%.
[0077] Voice interaction and explanation module: used to provide explanations of cultural relics through receiving devices, support voice dialogue, identify visitor needs and provide services.
[0078] The explanation of cultural relics is triggered based on location data, and the audio is played using TTS technology.
[0079] We analyze tourist queries using speech recognition; we identify tourist needs using NLP models and provide services based on those needs.
[0080] The receiving equipment is also used to collect tourist satisfaction feedback; the feedback is analyzed using machine learning models to update tourist preferences and parameters, forming a closed loop.
[0081] Receive voice input from tourists; convert it to text using ASR; generate a response through a dialogue model and output it as voice.
[0082] When the device enters within 5 meters of the exhibit, it queries the database and loads explanatory content (multilingual versions, such as Chinese / English). TTS parameters: speech rate 120 words / minute, natural tone. Example: Approaching a porcelain exhibit, play "This porcelain originated in the Ming Dynasty, with exquisite craftsmanship..."
[0083] It supports Automatic Speech Recognition (ASR, such as Google Speech-to-Text or iFlytek) to parse visitor queries. The device's microphone has a sampling rate of 16kHz, and the text is converted after noise suppression, supporting keyword extraction (such as "history").
[0084] The model takes text as input, outputs an intent classification (e.g., "Route Inquiry" with a probability of 0.9), and generates a response (e.g., "Recommended Detour Route"). Services include route inquiries, artifact details, or booking guided tours. Example: A visitor asks, "What is the historical background of this exhibit?", and the model responds with "Ming Dynasty porcelain, symbolizing the imperial family..." and outputs it via Text-to-Speech (TTS).
[0085] Visitor satisfaction feedback is collected through receiving devices, and a pop-up interface (such as a star rating of 1-5 or a voice comment "How was the experience?") appears at the end of the exhibition, and the data is uploaded to the server.
[0086] Machine learning models (such as random forests or LSTMs) are used to analyze feedback and update tourist preferences and system parameters. Specifically: Input feedback vectors (e.g., [ratings, review keywords]), use random forests to classify satisfaction, use LSTM sequences to analyze trends, update preferences (preference vector = [class weights], optimized via gradient descent with a learning rate of 0.01) and parameters (e.g., adjusting the A* factor from 1.5 to 1.6), train the model with feedback data, iterate weekly, and achieve an accuracy improvement of >10%. Example: Low ratings indicate "congestion," so optimizing the heatmap threshold reduces future congestion, forming a closed loop to ensure future service optimization.
[0087] Example 2
[0088] Please see Figure 2 This invention also provides an intelligent guided tour system for museums, comprising:
[0089] S1: When visitors enter the museum, receiving devices are distributed to collect location data in real time. The central server aggregates all location data and generates a global location map.
[0090] S2: Divide the museum map into multiple areas using a grid partitioning method, calculate the visitor density in each area, generate a traffic heat map using GIS tools, and push it to the receiving device via API;
[0091] S3: Construct a graph model, add priority and tourist preference matching degree to each node, adjust the edge weights using heatmap data, run the A algorithm on the adjusted graph model, and output a list of path sequences; if congestion occurs, dynamically adjust the route and update it to the receiving device.
[0092] S4: Triggering explanations of cultural relics based on location data, and playing audio using TTS technology;
[0093] S5: Collect tourist satisfaction feedback through receiving devices, update tourist preferences and parameters, and form a closed loop.
[0094] Furthermore, the present invention also provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned intelligent guided tour method for museums. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.
[0095] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.
Claims
1. A smart guided tour system for museums, characterized in that, include: Device distribution and location monitoring module: used to distribute receiving devices and monitor the location of each receiving device in real time, generating location data; Visitor flow sensing module: used to obtain the number of visitors in each area and generate a flow heat map by combining location data; Dynamic route optimization module: Generates the best tour route based on traffic heatmaps and tourist preferences, and makes dynamic adjustments accordingly; Voice interaction and explanation module: used to provide explanations of cultural relics through receiving devices, support voice dialogue, identify visitor needs and provide services.
2. The intelligent guided tour system for museums according to claim 1, characterized in that, The specific implementation method of the device distribution and location monitoring module includes: Receiving devices are distributed to visitors as they enter the museum; The device collects location data in real time through its built-in positioning sensor and uploads it to a central server. The central server aggregates all location data and generates a global location map.
3. The intelligent guided tour system for museums according to claim 2, characterized in that, The specific implementation method of the tourist flow sensing module includes: The system obtains real-time location data from a central server, divides the museum map into multiple areas using a grid partitioning method, counts the receiving devices in each area, calculates visitor density, and applies a sliding window to smooth the data. Using GIS tools, the number and density of visitors are mapped onto the museum map to create a traffic heat map. The heat map uses color gradients and overlays area boundaries and artifact markers. The heatmap includes timestamps and congestion scores, is updated regularly, and is pushed to receiving devices via API.
4. The intelligent guided tour system for museums according to claim 3, characterized in that, The specific implementation method of the dynamic route optimization module includes: A graph model is constructed with nodes as exhibition areas and edges as path distances. Priorities are added to each node according to the order of cultural relic visits, and visitor preference matching degree is added to each node, with physical distance as the initial weight. The edge weights are adjusted using heatmap data. The adjusted edge weights = initial weights + (heatmap congestion score × penalty factor). The attraction score is calculated as the reward for path optimization, combining tourist preference matching degree and cultural relic priority. The attraction score = tourist preference matching degree × node priority × reward factor. The tourist preference matching degree is the cosine similarity between the tourist preference vector and the node type vector. The attraction score is subtracted from the adjusted edge weights to obtain the final edge weights. Run the A* algorithm on the adjusted graph model to search for the optimal path from the starting point to the ending point. Combine the actual cost g(n) and the heuristic cost h(n) as the cost function, i.e., total cost = g(n) + h(n), where g(n) is the cumulative edge weight to the current node, and the heuristic cost h(n) = initial weight + (heatmap congestion score × 1.5) - attraction score; set the constraint that the total path duration < a specified time, and output a list of path sequences. If congestion occurs, the route is dynamically adjusted by recalculating using the A* algorithm, prioritizing low-density alternative paths, and updating the receiving device.
5. The intelligent guided tour system for museums according to claim 4, characterized in that, The specific implementation method of the voice interaction and explanation module includes: The explanation of cultural relics is triggered based on location data, and the audio is played using TTS technology. We analyze tourist queries using speech recognition; we identify tourist needs using NLP models and provide services based on those needs.
6. The intelligent guided tour system for museums according to claim 5, characterized in that, The receiving device is also used to collect tourist satisfaction feedback; the feedback is analyzed using a machine learning model to update tourist preferences and parameters, forming a closed loop.
7. The intelligent guided tour system for museums according to claim 6, characterized in that, The method for voice dialogue includes: Receive voice input from tourists; convert it to text using ASR; generate a response through a dialogue model and output it as voice.
8. The intelligent guided tour system for museums according to claim 7, characterized in that, The heatmap congestion score is obtained from heatmap data, which is a comprehensive output generated by the traffic perception module, including region ID, number of tourists, density, timestamp, congestion score and other auxiliary data; The maximum capacity, time factor, and balance weight of each area are obtained from other auxiliary data. The congestion score is calculated as (current density / maximum capacity) × 100 + time factor × weight. The calculated congestion score is added to the heat map data and visualized using GIS tools.
9. A smart guided tour system for museums according to claim 8, characterized in that, In the tourist preference vector and node type vector, the tourist preference vector is a numerical vector that represents the intensity of tourists' interest in different types of cultural relics. The vector dimension is equal to the number of cultural relic categories, and the value range of each element is [0,1]. The higher the value, the greater the interest. The node type vector represents the type representation vector for each exhibition area, and its dimension is the same as that of the visitor preference vector.
10. A smart guided tour system for museums, used to implement the smart guided tour method for museums as described in claims 1-9, characterized in that, include: S1: When visitors enter the museum, receiving devices are distributed to collect location data in real time. The central server aggregates all location data and generates a global location map. S2: Divide the museum map into multiple areas using a grid partitioning method, calculate the visitor density in each area, generate a traffic heat map using GIS tools, and push it to the receiving device via API; S3: Construct a graph model, add priority and tourist preference matching degree to each node, adjust the edge weights using heatmap data, run Algorithm A on the adjusted graph model, and output a list of path sequence. If congestion occurs, the route will be dynamically adjusted and updated to the receiving device; S4: Triggering explanations of cultural relics based on location data, and playing audio using TTS technology; S5: Collect tourist satisfaction feedback through receiving devices, update tourist preferences and parameters, and form a closed loop.