Scenic area passenger flow trend prediction analysis method based on AI big language model

Abstract

The invention belongs to the technical field of artificial intelligence and big data analysis, and particularly relates to a scenic spot passenger flow trend prediction analysis method based on an AI big language model. According to the method, big data and artificial intelligence technologies are utilized, LBS data, scenic spot data, holiday and festival information, third-party maps, weather data and other multi-source data are comprehensively applied, a comprehensive multi-dimensional knowledge base is constructed, and meanwhile, the knowledge search and generation capability is improved by utilizing an RAG technology; through model pre-training and fine tuning, the universal model capability is developed, and a multi-agent system is further constructed; based on the natural language processing capability of the AI large language model, the LLM can intelligently analyze the text input of the scenic region administrator, drive the cooperative work among multiple agents, and achieve the intelligent prediction and analysis of the scenic region passenger flow trend. The method can more accurately predict the passenger flow trend of the scenic area.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence and big data analysis, and specifically relates to a method for predicting and analyzing tourist flow trends in scenic spots based on an AI big language model. Background Art

[0002] In traditional scenic spot passenger flow management, data collection and analysis often rely on a single data source, such as scenic spot ticket sales data or tourists' online behavior data. Although this method can reflect passenger flow trends to a certain extent, it lacks comprehensiveness and depth. With the development of big data technology, more and more data sources can be used, such as social media data, meteorological data, traffic flow data, etc. The integration and analysis of these data can more accurately predict the passenger flow trends of scenic spots. Summary of the invention

[0003] In order to solve the above technical problem, that is, how to more accurately predict the passenger flow trend of a scenic spot, the present invention provides the following technical solutions:

[0004] A method for predicting and analyzing tourist flow trends in scenic spots based on an AI large language model comprises the following steps:

[0005] S1. Full data collection and processing: Collect data from multiple sources related to scenic spot passenger flow, including LBS data, scenic spot data, holiday information and weather information, and third-party map data, and clean, integrate and standardize these data;

[0006] S2. Build a knowledge base based on multi-dimensional data: Input the standardized data of the scenic spot itself and the operator's big data into the big data platform, and integrate the processed data into the knowledge base to build a comprehensive industry and scenario knowledge base to store and organize various types of information; use RAG technology to search the knowledge base and generate content;

[0007] S3. Capability to build general models: Use machine learning technology to pre-train general models based on multi-dimensional data in the knowledge base to identify patterns and associations in the data; general models include scenic spot passenger flow prediction models, travel chain analysis models, and tourist portrait models; at the same time, based on pre-training, fine-tuning technology is used to fine-tune the general model using multiple source data related to scenic spot passenger flow;

[0008] S4. Build a multi-agent system based on general model capabilities: Based on the general model obtained through pre-training and fine-tuning, build a multi-agent system including a travel group monitoring agent, a full-chain travel trajectory query agent, and a scenic spot passenger flow query agent;

[0009] S5. Drive multiple agents to perform predictive analysis based on LLM: By integrating the AI ​​large language model LLM, multiple agents are driven to work together to perform intelligent predictive analysis on the tourist flow trend of scenic spots.

[0010] Furthermore, LBS data includes the location information and movement trajectory of tourists. LBS data includes: user data, location data, timestamp, movement trajectory, geospatial data, point of interest data, user behavior data, and social network data.

[0011] Furthermore, in step S1, data cleaning includes removing irrelevant characters, removing duplicates, removing noise, filling missing values, data type and format verification, and logical error checking. Noise removal includes; identifying and cleaning non-numeric values ​​that appear in the ticket sales data in the scenic area data, logical error checking; including ensuring that the price in the ticket sales data in the scenic area data does not exceed the set ticket sales price cap.

[0012] Further, step S2 comprises,

[0013] S21, data integration: Integrate the processed data, including LBS data, scenic spot data, holiday information and weather information, and third-party map data;

[0014] S22. Feature extraction: Use model prediction analysis to extract key features and patterns from the integrated data. Key features include tourist behavior patterns, peak passenger flow periods, and the impact of weather on passenger flow. Use clustering algorithms to identify tourist behavior patterns, use time series analysis to predict peak passenger flow periods, and use regression models to assess the potential impact of weather changes on the number of tourists.

[0015] S23, Rule and Pattern Recognition: Identify rules and patterns in data through model analysis, including using hypothesis testing to determine the correlation between holidays and increased tourist flow, and applying time series analysis to identify the impact of weather changes on tourist flow;

[0016] S24. Use RAG technology to build industry and scenario knowledge bases: Integrate the extracted features, identified rules and patterns, and generated conclusions into a structured database to build a knowledge base.

[0017] Furthermore, step S21 specifically includes:

[0018] S211, reading LBS data, holiday information, weather information and third-party map data;

[0019] S212, executing data information and conversion operations, including removing invalid records in LBS data and converting timestamp format;

[0020] S213, using timestamp and location information as key fields, connecting multiple data sets of LBS data, holiday information, weather information, and third-party map data to form a unified data view;

[0021] S214, selecting columns required for analysis from the connected data set, the columns including timestamp, location information, holiday information, weather information, and third-party map data; renaming the selected columns to unify the format of the data view;

[0022] S215. Storing the data in a unified format in a knowledge base.

[0023] Furthermore, in step S22, using a clustering algorithm to identify tourist behavior patterns specifically includes: selecting location data, timestamp and weather information as features, applying a K-means algorithm to cluster tourist behaviors, and identifying different tourist behavior patterns.

[0024] Furthermore, in step S23, a hypothesis test is used to determine the correlation between holidays and the increase in tourist flow, including:

[0025] Add holiday labels to dates in the combined dataset;

[0026] Construct a linear regression model to evaluate the impact of holidays on the number of tourists;

[0027] Prints a statistical summary of the linear regression model.

[0028] Further, step S3 includes:

[0029] S31. Model pre-training: Use machine learning technology to pre-train the general model based on the multi-dimensional data in the knowledge base;

[0030] S32. Model fine-tuning: Use multiple source data related to scenic spot passenger flow to fine-tune the pre-trained model, including adjusting the model's output layer to adapt to the output format of the scenic spot passenger flow prediction model, travel chain analysis model, and tourist portrait model.

[0031] Further, step S4 includes:

[0032] S41. Design of the agent architecture: including the definition of the role and responsibility of each agent and the interaction mode between agents; the agents include the travel group monitoring agent, the full-chain travel trajectory query agent, and the scenic spot passenger flow query agent;

[0033] S42. General model integration: Based on the general model obtained through pre-training and fine-tuning, the general model is integrated into various agents, including the travel group monitoring agent integrating a deep learning model to analyze tourist behavior patterns, the full-chain travel trajectory query agent integrating a sequence prediction model to track the movement paths of tourists, and the scenic spot passenger flow query agent integrating a time series analysis model to provide passenger flow data;

[0034] S43, Functional Implementation and Optimization: The travel group monitoring agent includes an event detection system for identifying and responding to abnormal patterns in tourist behavior; the full-chain travel trajectory query agent includes a path tracking algorithm for processing and analyzing mobile data; the scenic spot passenger flow query agent includes a query optimization system for responding to passenger flow data query requests;

[0035] S44. Collaborative working mechanism: Intelligent agents communicate through well-defined communication protocols and data sharing mechanisms.

[0036] Further, step S5 includes,

[0037] S51, language input and intent analysis, including:

[0038] S511, the scenic area administrator inputs a query or instruction to the system through natural language, and the input includes a passenger flow prediction request for a specific date, time period or event. The LLM first receives the natural language input of the scenic area administrator, and the LLM automatically recognizes and analyzes the administrator's intention and determines the type of task to be performed;

[0039] S512, after receiving the input, LLM uses its natural language understanding (NLU) technology to analyze the administrator's query intent;

[0040] S513. Based on the analysis results, LLM determines the specific task types that need to be performed, including short-term passenger flow prediction, long-term trend analysis, and real-time data monitoring and abnormal flow warning;

[0041] S52, agent call and analysis execution: After determining the task type, according to the analysis results, LLM drives the multi-agent system to work together and call the corresponding sub-agent; the agents work together, share information and resources, select the best database path, access and analyze the historical data, real-time data and other relevant information stored in the knowledge base, and select the corresponding prediction model to perform scenic spot passenger flow prediction analysis;

[0042] S53, result feedback and text output: After completing the analysis, the sub-agent will feed back the prediction results to LLM, and LLM will provide the prediction analysis results to the scenic area administrator in the form of natural language.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] (1) The present invention achieves collaborative work and decision support of the multi-agent system by integrating the AI ​​large language model to drive the multi-agent system to predict and analyze the trend of scenic spot passenger flow, thereby realizing the collaborative work and decision support of the multi-agent system. This method not only improves the efficiency and effectiveness of the overall system, more accurately predicts the passenger flow trend of the scenic spot, but also returns the prediction results in the form of natural language, so that scenic spot managers can intuitively understand and use these prediction results. This solves the problem of complex and difficult to understand results in traditional systems, makes decision-making more convenient, and provides a more efficient, intelligent and user-friendly solution for scenic spot passenger flow management.

[0045] (2) Through the scenic spot passenger flow trend prediction and analysis method, scenic spot managers can more accurately grasp and predict passenger flow trends and make corresponding countermeasures in advance. This method not only improves the accuracy of prediction, but also improves the intelligent level of scenic spot management, providing technical support for the sustainable development of scenic spots.

[0046] (3) By integrating LBS location data, scenic spot data, holiday information, third-party maps and weather data, the present invention constructs a multi-dimensional knowledge base, and uses RAG technology to improve the ability to search and generate knowledge bases and external data. Through model pre-training and fine-tuning, general model capabilities including scenic spot passenger flow prediction, travel chain analysis and tourist portraits are developed. The application of these technologies enables the present invention to provide scenic spot managers with passenger flow trend prediction analysis based on natural language input, realize the collaborative work and decision support of multi-agent systems, and improve the intelligence level and response efficiency of scenic spot management. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flow chart of the scenic spot passenger flow trend prediction and analysis method of the present invention;

[0048] Figure 2 This is a data processing architecture diagram of the scenic spot passenger flow trend prediction and analysis method of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention, and all other embodiments obtained by ordinary technicians in the field without making creative work are within the protection scope of the present invention.

[0050] like Figure 1As shown, the present invention provides a method for predicting and analyzing the trend of scenic spot passenger flow based on AI large language model. The method utilizes big data and artificial intelligence technology, and comprehensively applies multi-source data such as LBS data, scenic spot data, holiday information, third-party maps and weather information, and constructs a comprehensive multi-dimensional knowledge base, while using RAG technology to enhance knowledge search and generation capabilities. On this basis, through model pre-training and fine-tuning, general model capabilities are developed, and a multi-agent system is further constructed. Then, based on the natural language processing capability of AI large language model (LLM), LLM can intelligently parse the text input of scenic spot administrators, drive the collaborative work between multiple agents, and realize the intelligent prediction and analysis of scenic spot passenger flow trends. This method aims to improve the scientificity and foresight of scenic spot management, optimize the tourist experience, and provide data support for the operational decision-making of scenic spots.

[0051] like Figure 1 As shown, the scenic spot passenger flow trend prediction and analysis method specifically includes the following steps:

[0052] S1. Full data collection and processing:

[0053] Collect data from multiple sources related to scenic spot passenger flow, including LBS data, scenic spot data, holiday and weather information, and third-party map data, and clean, integrate and standardize these data to prepare for subsequent analysis and model training, so as to better predict the changing trend and peak of scenic spot passenger flow in the future and help scenic spots formulate reasonable operation plans. Step S1 includes:

[0054] S11. Full data collection:

[0055] Full data collection is the basic step of the scenic spot passenger flow trend prediction and analysis method. For the scenic spot passenger flow trend prediction and analysis, in order to obtain more accurate prediction and analysis results, this method requires a large amount of original data, the collected data dimensions are more diverse, and the annotation method and evaluation criteria are more complicated.

[0056] Collect data from multiple sources related to scenic spot passenger flow, including LBS data, scenic spot data, holiday and weather information, and third-party map data. Data collection realizes automated data exchange through API interface, and manually uploads data files through FTP protocol. This process uses automated tools and API to grab data from different sources to ensure the comprehensiveness and diversity of data. At the same time, the obtained scenic spot crowd data is labeled to obtain crowd labels and user portraits. These data can provide rich background information and prediction basis for the scenic spot passenger flow prediction model.

[0057] 1. LBS (location-based service) data:

[0058] Real-time passenger flow analysis in scenic spots is crucial for effective management and planning. Accurate location information and real-time data updates are needed to predict and warn passenger flow trends. In this method, we need to collect LBS data, including the location information and movement trajectory of tourists. At the same time, we use third-party map data to enhance the accuracy and coverage of geographic location.

[0059] LBS (Location-Based Services) technology can obtain users' real-time geographic location information through mobile wireless networks and mobile positioning sensors. This technology combines positioning technology and mobile Internet, which can not only accurately track user locations, but also provide various location-based services and functions. LBS data collects a number of key information, including:

[0060] User data: basic information and preferences of users;

[0061] Location data: core data that records the user's specific location;

[0062] Timestamp: records the specific time when the location information is obtained;

[0063] Movement trajectory: the movement path of a user over a period of time;

[0064] Geospatial data: information about the geographic area of ​​the user's location;

[0065] POI (point of interest) data: information about points of interest visited by users;

[0066] User behavior data: user activities and behavior patterns in a specific location;

[0067] Social network data: User activities and interactions on social networks.

[0068] This data can be obtained through a variety of technologies, such as outdoor positioning technologies such as GPS and Beidou, and indoor positioning technologies such as WiFi and Bluetooth. LBS technology can collect location information of a large number of mobile phone users at a time point, which is particularly important in China where mobile Internet is highly popular.

[0069] In terms of tourist flow prediction in scenic spots, the advantages of LBS data are particularly obvious. Its high spatiotemporal accuracy, large-scale data capacity and time series characteristics provide a solid data foundation for tourist flow prediction and analysis. These data can not only help managers monitor the current tourist flow status in real time, but also predict the trend of tourist flow through analytical prediction models, thus providing decision support for the operation management and planning of scenic spots.

[0070] 2. Scenic spot data:

[0071] Collect and summarize scenic spot data, including scenic spot information, tourist capacity, facility distribution, ticket sales data, existing scenic spot passenger flow data, historical passenger flow data, tourist behavior data, etc.

[0072] (1) Ticket sales data:

[0073] Ticket sales data is a very important data source for tourist flow trend forecasting in scenic spots, reflecting the total number of tickets sold in a specific time period. It can be counted by day, week, month or specific activity period to help managers understand the scale of tourist flow. Data such as ticket booking, hotel reservation and tourism information consultation conducted through online platforms can be used as a reference for predicting future passenger flow.

[0074] (2) Real-time tourist flow data of scenic spots:

[0075] The scenic spot passenger flow analysis and statistics system collects multiple data sources, including entrance and exit gates, ticketing data, checkpoint passenger flow data, and core scenic spot Wi-Fi sniffing data to achieve real-time statistics of the number of tourists. Using sensors, cameras and other technical means, the system can accurately record the number, time and location of tourists entering the scenic area. By processing, analyzing and visualizing the collected data through machine learning technology, the scenic spot management department can understand the flow, density and distribution of tourists in real time.

[0076] Based on real-time passenger flow data, scenic spots can formulate reasonable tourist control strategies to optimize tourist experience and scenic spot management efficiency. When the passenger flow is too large, the system can take flow control measures or use LED screens to inform tourists of the number of tourists in the scenic area to avoid overcrowding. In addition, the system can also control the number of tourists in real time by extending the waiting time for tourists, limiting the number of people entering the park, adjusting the tour route, etc., to ensure the safety and order of the scenic area.

[0077] (3) Historical passenger flow data:

[0078] Historical passenger flow data can help identify cyclical patterns and trends in tourist traffic, such as seasonal fluctuations, holiday peaks, etc. Historical passenger flow data can be used to analyze the impact of different factors (such as weather, economic conditions, special events, etc.) on passenger flow. By analyzing past passenger flow data, forecasting models can be trained and validated to improve the accuracy of future forecasts.

[0079] Methods for obtaining historical passenger flow data include:

[0080] Internal data collection: Obtain data such as the number of tourists and ticket purchase time through the scenic spot's ticket sales system, as well as the scenic spot's admission records, and use the entrance counter or electronic ticketing system to record the number of daily visitors;

[0081] External data acquisition: On the one hand, obtain tourism statistics from the tourism management department over the years, and on the other hand, obtain passenger flow data of transportation tools (such as buses, subways, trains, etc.) to the scenic spot from the transportation management department;

[0082] Social media and online platforms: By analyzing discussions and comments on social media, tourists’ feedback and behavior patterns can be obtained. Tourists’ booking data and evaluation information can also be obtained from travel booking websites and travel evaluation platforms.

[0083] Historical event records: Obtain records of historical events in the scenic area (such as major events, natural disasters, etc.) and analyze the impact of these events on passenger flow.

[0084] (4) Visitor behavior data:

[0085] Tourist behavior patterns in scenic spots, such as tour routes, length of stay, and consumption habits, help understand tourist preferences and optimize scenic spot services. By analyzing tourists' tour routes and length of stay, it is possible to predict tourist flows in different time periods and scenic spots, helping scenic spot managers to arrange resources and personnel more reasonably.

[0086] 3. Holiday information:

[0087] Integrate holiday information to identify the impact of special dates on customer flow.

[0088] Legal holidays, school holidays, special events or festivals held by scenic spots, etc., usually lead to significant changes in tourist flow. Holiday and special event information can help. Holidays and special events are usually peak periods of tourist flow. By analyzing the passenger flow data during these periods, future peak periods can be predicted so that preparations can be made in advance. For example, legal holidays such as the Spring Festival, Labor Day, and National Day usually attract a large number of tourists, resulting in a surge in passenger flow. Thus, optimizing resource allocation in advance and understanding passenger flow trends during holidays and special events can help scenic area managers allocate resources more reasonably, such as adding staff, adjusting opening hours, optimizing transportation arrangements, etc., to cope with passenger flow peaks and enhance the tourist experience.

[0089] 4. Weather information:

[0090] Collect and analyze weather information to assess its potential impact on tourist behavior.

[0091] Weather conditions (such as temperature, precipitation, wind speed, humidity, etc.) have a significant impact on tourists' travel decisions. Sunny weather tends to attract more tourists, while bad weather may suppress tourist flow.

[0092] S12. Data cleaning and preprocessing:

[0093] like Figure 2As shown, the collected data is cleaned, and pre-processed such as deduplication and standardization are performed, including noise removal, missing value filling, and standardization, in preparation for subsequent analysis and model training.

[0094] 1. Data cleaning:

[0095] Data cleaning is an important step to ensure data quality and accuracy. As the first stop for data inflow, the interface server receives raw data from different sources. After receiving the raw data, the interface server performs basic cleaning operations on the received data, including removing irrelevant characters, removing duplicates, removing noise, filling missing values, data type and format verification, logical error checking, etc., to ensure that the data meets the preset standards for subsequent processing. The main tasks in this process are as follows:

[0096] Noise Removal: Identify and clean outliers or erroneous data from your data. This includes formatting errors, data type errors, or logical errors. For example, non-numeric values ​​may appear in your ticket sales data and need to be identified and cleaned.

[0097] Deduplication: Duplicate data can affect the accuracy of analytical results, especially when performing statistical analysis or training machine learning models. Deduplication identifies and removes duplicate records in a dataset. This includes not only identical records, but also those that are duplicated on key fields.

[0098] Filling missing values: Dealing with missing values ​​in the data. Common approaches include estimating missing values ​​using the mean, median, mode, or a predictive model. Choosing an appropriate imputation method depends on the nature of the data and the analytical requirements.

[0099] Data type and format validation: Ensure that data conforms to predetermined data type and format requirements, such as converting a string to a date format or converting a number to a floating point number.

[0100] Logical error checking. Check whether the data conforms to the business logic, such as ensuring that the price in the ticket sales data does not exceed a certain upper limit.

[0101] Through this series of cleaning steps, the collected raw data will be converted into a clean, tidy and consistent format, providing a data basis for subsequent data analysis, statistics and model training.

[0102] 2. Data standardization processing:

[0103] Data standardization ensures the consistency and availability of data. The following is a detailed description of the data standardization process:

[0104] (1) Format conversion. For date and time data, all data that do not conform to the standard format are uniformly converted to the yyyy-mm-dd format, ensuring the accuracy of time series analysis and the comparability of time data.

[0105] (2) Assign default values. Check each field in the data set. If there are null or missing values, assign an appropriate default value, such as 0, an empty string, or a statistical value (mean, median, etc.), based on business logic and data characteristics.

[0106] (3) Type conversion: unify the data types in different data sources. For example, convert the Number type in the database to the varchar2 type suitable for SQL operations to avoid type mismatch errors.

[0107] (4) Length conversion: For character data, adjust the length of the field as needed, such as expanding varchar2(10) to varchar2(20) to accommodate longer data content.

[0108] (5) Code conversion. For codes that have changed due to system upgrades, corresponding updates and mappings are performed to ensure data consistency, such as converting old code values ​​to new code values.

[0109] (6) Remove spaces. Clean character data and remove spaces before and after fields to ensure data neatness.

[0110] (7) Specific character conversion. For data containing special characters, replace or remove them according to business rules, such as replacing symbols such as +, -, *, and / in calculation fields with appropriate representations.

[0111] (8) Data type unification: Ensure that the data types in all data sets are consistent. For example, unify all date fields as date type and all numeric fields as numeric type.

[0112] (9) Data format unification. Text data is formatted, such as converting all text to lowercase and removing extra spaces and punctuation marks, to achieve consistency of text data.

[0113] (10) Data encoding. Encoding categorical data, such as converting weather conditions (sunny, cloudy, rainy) or holiday types into numerical labels, helps machine learning models better handle categorical features.

[0114] S2. Build a knowledge base based on multi-dimensional data:

[0115] The cleaned and standardized data, including the scenic spot's own data and the operator's big data, are imported into the big data platform. The data are stored in HBase and Hive respectively. HBase is used for real-time query services, and Hive is mainly used for data analysis and mining services. Hive is used as a data warehouse tool. It can organize data in HDFS and support SQL-like query languages, which facilitates data management and analysis. Finally, the processed and analyzed data is integrated into the knowledge base to provide insights and support for decision makers and help formulate data-based strategies.

[0116] Using the processed data, we build a comprehensive industry and scenario knowledge base to store and organize various types of information. At the same time, we use RAG technology to enhance the search and content generation capabilities of the knowledge base and external search engines, so that they can quickly respond to queries and provide relevant information.

[0117] Multimodal data processing is an advanced technology that enhances the predictive power and breadth of application scenarios of models by combining information from different sources and types of data. This approach has significant advantages in improving the accuracy and application scope of predictive models.

[0118] With the development of large neural network models, large multimodal models are able to simultaneously process and understand multiple data types including text, audio, images, and video, thereby obtaining a more comprehensive data perspective and demonstrating higher robustness in different scenarios, thereby enabling more accurate scenic spot passenger flow trend prediction and analysis applications.

[0119] In the context of passenger flow trend prediction, the application details of multimodal data processing can include the following steps:

[0120] The first is data fusion, which involves integrating data from different modalities, such as tourists’ comments, scenic spot visual content such as pictures and videos, and related content on social media. These data come from a variety of platforms, such as online travel review websites, social media platforms, and scenic spot monitoring systems.

[0121] In the feature extraction stage, each data modality may provide unique insights. For example, text analysis can reveal tourists' emotional tendencies and specific evaluations of scenic spots; image and video analysis can provide visual information about the real-time density of people flow and activity characteristics of scenic spots. Using deep learning technology, key features are extracted from these different modal data, such as emotional tendencies in text, crowd flow and activity recognition in images, etc.

[0122] By applying deep learning models, such as using convolutional neural networks (CNN) to process image data or recurrent neural networks (RNN) to process sequential data, these multimodal data can be deeply analyzed. These models can capture the complex relationships between different data modalities, thereby improving the accuracy and reliability of passenger flow prediction.

[0123] The advantage of multimodal data processing is not only to improve the accuracy of prediction, but also to greatly expand the scope of application of the model. In addition to passenger flow trend prediction, this method can also be applied to many fields such as scenic area security management, optimization of tourist experience, and formulation of marketing strategies. By comprehensively analyzing the image and video content on social media, we can more realistically understand the feelings and needs of tourists, and thus provide more customized and targeted services. Step S2 includes,

[0124] S21. Data Integration:

[0125] Data integration is the process of integrating processed data, including LBS data, scenic spot data, holiday and weather information, third-party map data, etc., to form a unified data view.

[0126] In the data integration phase of building a scenic spot knowledge base, the first thing to do is to collect and process multi-source data. This process involves obtaining data from different data sources, such as obtaining real-time location data of tourists from LBS services through APIs, extracting historical scenic spot visit records from databases, obtaining weather information from meteorological service interfaces, and integrating geospatial data from third-party map services. The integration of these data can be achieved through SQL queries, ETL (Extract, Transform, Load) jobs, or using data processing frameworks such as Apache Spark.

[0127] S22. Feature extraction:

[0128] Using model predictive analysis, key features and patterns are extracted from the integrated data. These features include tourist behavior patterns, peak passenger flow periods, the impact of weather on passenger flow, etc.

[0129] In the feature extraction stage of the scenic spot knowledge base construction process, advanced data analysis and machine learning techniques are used to identify and extract key features in the data. This step is achieved by analyzing multi-dimensional information such as tourists' location data in the scenic area, visit time, weather conditions, etc. For example, clustering algorithms such as K-means or DBSCAN are used to identify tourist behavior patterns, or time series analysis is used to predict peak passenger flow periods. In addition, regression models are used to evaluate the potential impact of weather changes on the number of tourists.

[0130] By selecting location, timestamp and weather conditions as features, the K-means algorithm is applied to cluster tourist behaviors and identify different tourist behavior patterns. These patterns can provide scenic area managers with insights into tourist preferences and behavior habits, which helps optimize scenic area services and tourist experience.

[0131] The extracted features and patterns are analyzed in depth to form rules and conclusions that have practical guiding significance for scenic spot operations. For example, through analysis, it is found that the number of tourists decreases significantly under certain weather conditions, or the number of tourists increases sharply during a specific time period. These analysis results will be integrated into the knowledge base to provide data support for the scenic spot's passenger flow management and resource allocation.

[0132] S23. Rules and pattern recognition:

[0133] Through model analysis, we can identify rules and patterns in the data, such as the positive impact of holidays on tourist flow, or the changing trend of tourist flow under specific weather conditions.

[0134] In the rule and pattern recognition stage of building the scenic spot knowledge base, statistical analysis and machine learning models are used to discover potential patterns in the data. Hypothesis testing is used to determine the correlation between holidays and increased tourist flow, or time series analysis is applied to identify the impact of weather changes on tourist flow.

[0135] First, a holiday label is added to the dates in the dataset, and then a linear regression model is built to evaluate the impact of holidays on the number of tourists. The statistical summary of the linear regression model provides the significance level and effect size of the holiday effect.

[0136] Furthermore, sequence analysis is used to identify the patterns of weather conditions on tourist flow. An ARIMA model is constructed to predict and analyze the impact of weather changes on tourist flow.

[0137] S24. Use RAG technology to build industry and scenario knowledge base:

[0138] The extracted features, identified rules and patterns, and generated conclusions are integrated into a structured database to build a knowledge base. The knowledge base should be designed to facilitate information retrieval and updating. In this process, RAG technology is used to build a knowledge base for industries and scenarios.

[0139] RAG (Retrieval-Augmented Generation) technology is a deep learning model that combines retrieval and generation to improve the performance of natural language processing tasks. The RAG model enhances the traditional generation model by retrieving relevant information, enabling it to use the knowledge base to generate more accurate and rich responses.

[0140] 1. RAG model design:

[0141] Design a RAG model, including a retrieval component and a generation component; the retrieval component is responsible for retrieving information related to the current task from the knowledge base; the generation component generates prediction results based on the retrieved information and the input context.

[0142] 2. Model training:

[0143] Use historical passenger flow data and related influencing factors to train the RAG model;

[0144] During the training process, the model learns how to retrieve relevant information from the knowledge base and combine this information to predict passenger flow trends.

[0145] S3. Ability to build general models:

[0146] General model capabilities refer to the cross-domain and cross-task processing capabilities demonstrated by general models. This capability enables general models to perform well in different application scenarios without the need for a large amount of customized development for each specific task. The level of general model capabilities directly determines the effectiveness and reliability of the model in practical applications.

[0147] The general model includes scenic spot passenger flow prediction model, travel chain analysis model, tourist portrait model, etc. Using machine learning technology, the general model is pre-trained based on the multidimensional data in the knowledge base to identify patterns and associations in the data. At the same time, based on the model pre-training, the model is fine-tuned using scenic spot-related data through fine-tuning technology to optimize the model performance for specific tasks (such as scenic spot passenger flow prediction, travel chain analysis, and tourist portrait construction), enhance its intelligent prediction and analysis capabilities for scenic spot passenger flow trends, and the trained model can better adapt to passenger flow prediction tasks.

[0148] S31. Model pre-training:

[0149] Using machine learning techniques, a general model is pre-trained based on multi-dimensional data in a knowledge base to identify patterns and associations in the data.

[0150] The pre-training process first requires in-depth analysis and understanding of the multi-dimensional data in the knowledge base. This step involves identifying different types and structures in the data, such as text, images, time series, etc. After the data preparation is completed, the next step is to choose the appropriate machine learning model architecture for pre-training. This is a complex decision-making process that needs to consider the characteristics of the data, the requirements of the task, and the limitations of computing resources, as well as the need to develop a training strategy, including choosing an appropriate loss function, optimization algorithm, and learning rate scheduling to ensure that the model can effectively learn from the data.

[0151] The final stage of pre-training is model training. In this stage, the model is trained on a large amount of data to learn patterns and associations in the data. This process usually involves a lot of computing resources and time, as the model needs to go through multiple iterations on a large dataset to converge. During the training process, the model gradually adjusts its parameters to minimize the prediction error. As training progresses, the model begins to identify high-level features in the data that can capture complex data structures and potential semantic information. After pre-training is completed, the model is able to extract rich feature representations that can serve as a starting point for downstream tasks and provide a solid foundation for fine-tuning for specific tasks.

[0152] S32. Model fine-tuning:

[0153] Fine-tuning is a key step in the model training process. In this step, the pre-trained model is fine-tuned using scenic spot-specific data, including adjusting the model's output layer to adapt to the output format of a specific task, or further training the model to learn task-specific features. The fine-tuned model is trained using both the training set and the validation set, and the performance of the model on the validation set is monitored so that it can better adapt to the passenger flow prediction task.

[0154] Fine-tuning is the process of adapting a pre-trained model to a specific task. Fine-tuning is a common strategy in machine learning and deep learning that allows a model pre-trained on a large-scale dataset to be adjusted and optimized to fit a specific task. The key to this approach is that the pre-trained model has learned a wide range of languages ​​and patterns, and the fine-tuning process further adjusts this learned knowledge to make it more suitable for specific application scenarios. In the scenario of scenic spot passenger flow prediction, fine-tuning enables the model to understand and predict tourist behavior and traffic patterns more accurately.

[0155] The method used is parameter-efficient fine-tuning (PEFT). Language models and fine-tuning are powerful tools in the field of natural language processing. PEFT technology combined with parameter efficiency strategies such as LoRA and quantization can effectively make full use of these models.

[0156] Parameter-EfficientFine-tuning (PEFT) is an improved form of fine-tuning techniques that aims to reduce the number of parameters that need to be adjusted during the fine-tuning process. This approach is particularly important for large models, which often contain hundreds of millions of parameters. Fully fine-tuning these parameters is not only computationally expensive, but may also lead to overfitting. PEFT enables the model to quickly adapt to new tasks while maintaining the extensive knowledge learned during the pre-training phase by adjusting only a small number of parameters in the model.

[0157] Low-Rank Adaptation (LoRA) is a typical PEFT method. Methods such as P-Tuning, P-Tuningv2, AdapterTuning and its variants, LoRA, AdaLoRA, QLoRA, MAMAdapter and UniPELT are also important components of efficient parameter fine-tuning techniques. These methods have their own characteristics and can be selected and adjusted according to specific tasks and data sets. Among them, the core idea of ​​LoRA is to use low-rank matrices to approximate the update of model weights. It reduces the number of parameters by introducing low-rank matrices to approximate the update of model weights. This method is particularly suitable for large models because these models contain a huge number of parameters. If all parameters are fully fine-tuned, not only is the computational cost high, but there is also a risk of overfitting. LoRA enables the model to quickly adapt to new tasks while retaining the rich knowledge learned in the pre-training phase by adjusting only a small number of parameters in the model.

[0158] Efficient parameter fine-tuning technology is one of the important research directions in the field of deep learning. By selecting appropriate fine-tuning methods and optimizing them in combination with specific tasks and datasets, the computational and storage costs of the model can be greatly reduced while maintaining the performance of the model.

[0159] S4. Building multi-agent systems based on general model capabilities:

[0160] Based on the general model obtained through pre-training and fine-tuning, a multi-agent system including travel group monitoring agents, full-chain travel trajectory query agents, scenic spot passenger flow query agents, etc. is further constructed based on these capabilities. These agents work together to share information and resources to optimize the prediction results. A multi-agent system is a computing system composed of multiple interacting agents, which aims to solve problems that are difficult for a single agent to handle through the collaborative work between agents. Multi-agent systems are an important branch of distributed artificial intelligence and are a frontier discipline of artificial intelligence in the world from the late 20th century to the early 21st century.

[0161] Step S4 includes:

[0162] S41. Agent Architecture Design:

[0163] Before building a multi-agent system, it is necessary to first design the agent architecture. This includes defining the roles and responsibilities of each agent and how they interact with each other. The agent architecture adopts a modular design for easy management and expansion. Each agent is designed to be autonomous, social, reactive and proactive, able to perform tasks independently and collaborate with other agents to provide more comprehensive services. Agents include travel group monitoring agents, full-chain travel trajectory query agents, and scenic spot passenger flow query agents.

[0164] Each agent is responsible for different functions:

[0165] Travel group monitoring agent: monitors tourists’ travel patterns and behaviors.

[0166] Full-chain travel trajectory query agent: Tracking tourists’ complete travel paths.

[0167] Scenic spot passenger flow query agent: provides real-time or historical passenger flow data query services.

[0168] S42. General Model Integration:

[0169] Based on the general models obtained through pre-training and fine-tuning, these models are integrated into various agents as their decision-making and reasoning engines. For example, the travel group monitoring agent will integrate a deep learning model to analyze tourist behavior patterns; the full-chain travel trajectory query agent will use a sequence prediction model to track the movement paths of tourists; and the scenic spot passenger flow query agent may integrate a time series analysis model to provide passenger flow data.

[0170] S43, Function Implementation and Optimization:

[0171] The functional realization of each agent requires meticulous algorithm development and optimization. For example, a travel group monitoring agent may need to implement a complex event detection system to identify and respond to abnormal patterns in tourist behavior. The full-chain travel trajectory query agent needs to develop an efficient path tracking algorithm to process and analyze large amounts of mobile data. The scenic spot passenger flow query agent needs to implement a query optimization system to quickly respond to query requests for passenger flow data.

[0172] S44. Collaborative working mechanism:

[0173] The collaborative work between agents is achieved through well-defined communication protocols and data sharing mechanisms. This involves the use of message queues, RESTful APIs, or other middleware technologies to enable information exchange between agents. In addition, a coordinator or scheduler needs to be implemented to manage the task allocation and resource scheduling of agents to ensure the overall efficiency and responsiveness of the system.

[0174] S5. LLM-based multi-agent driven predictive analysis:

[0175] The AI ​​Large Language Model (LLM) has powerful natural language processing capabilities and can understand and process the complexity of human language. LLM plays the role of the brain in this system, responsible for receiving input, parsing intent, guiding decision-making and text integration output. First, a large amount of historical passenger flow data is collected, including factors such as the number of tourists, time, weather conditions, holidays, etc. Then, through natural language processing technology, these data are converted into a machine-understandable form and input into the deep learning model. Through the deep learning algorithm, the historical behavior data of tourists in the scenic area are learned, and the possible passenger flow at different times and different periods in the future is predicted, so as to realize the analysis and prediction of the passenger flow trend of the scenic area.

[0176] By integrating the AI ​​Large Language Model (LLM), the LLM is able to understand and process natural language input, such as queries or instructions from scenic area administrators, and drive multiple agents to realize intelligent predictive analysis of scenic area passenger flow trends.

[0177] Through the natural language processing technology of the Large Language Model (LLM), LLM can parse the natural language input of scenic area managers. Based on the analysis results, LLM will drive multiple agents to work together to achieve forecast analysis of scenic area passenger flow trends. LLM will output the results of the forecast analysis to scenic area managers in an easy-to-understand text form to help them make more accurate decisions.

[0178] Natural language processing (NLP) is a key technology in the field of artificial intelligence, which enables machines to understand and process human language. In the present invention, NLP technology plays a vital role. Through deep learning algorithms, these models can parse and understand large amounts of text data, such as tourists' comments, posts on social media, news reports, etc. These text data contain rich information, such as tourists' satisfaction, expectations for scenic spots, and their behavior patterns. By analyzing this information, AI models can identify key factors that affect passenger flow, such as the attractiveness of scenic spots, service quality, and transportation convenience.

[0179] Here is how the method works and how it works:

[0180] S51, Language input and intention analysis:

[0181] The scenic area administrator inputs queries or instructions to the system through natural language (such as text or voice). These inputs include passenger flow forecast requests for specific dates, time periods, or events. LLM first receives the natural language input from the scenic area administrator, and LLM automatically recognizes and parses the administrator's intent to determine the type of task that needs to be performed.

[0182] Once LLM receives the input, it uses its advanced natural language understanding (NLU) technology to analyze the administrator's query intent. The system identifies key words, phrases, and contextual information to determine the administrator's specific needs, such as whether short-term forecasts, long-term trend analysis, or other specific types of analysis are needed.

[0183] Based on the analysis results, LLM determines the specific types of tasks that need to be performed, including short-term passenger flow forecasting (such as during holidays or special events), long-term trend analysis (such as seasonal changes or annual growth forecasts), and real-time data monitoring and abnormal traffic warnings.

[0184] S52, Agent call and analysis execution:

[0185] After the task type is determined, LLM will drive the multi-agent system to work together and call the corresponding sub-agents according to the analysis results. Each sub-agent is specially designed to handle a specific type of analysis task.

[0186] Intelligent agents work together, share information and resources, select the best database path, access and analyze historical data, real-time data and other relevant information stored in the knowledge base, and select corresponding prediction models to conduct scenic spot passenger flow prediction analysis.

[0187] S53, Result feedback and text output:

[0188] After completing the analysis, the sub-agent will feed back the prediction results to LLM. LLM will provide the prediction analysis results to scenic area managers in the form of natural language, allowing them to quickly grasp the passenger flow trend and make corresponding plans and adjustments.

[0189] Through this LLM-based multi-agent approach, scenic area managers can more effectively respond to changes in passenger flow, optimize resource allocation, make plans and adjustments in advance, improve visitor experience, and ultimately achieve efficient and sustainable development of scenic area operations.

[0190] Sub-agents usually refer to agents that undertake specific tasks or sub-tasks in a multi-agent system. They are the basic units that constitute a multi-agent system and have similar characteristics as multi-agent systems, such as autonomy, responsiveness, and sociality. However, compared with multi-agent systems, the functions and task scope of sub-agents are more specific and limited. In a multi-agent system, sub-agents work together to complete complex tasks through mutual cooperation and coordination.

[0191] Through the above steps, scenic area managers can more accurately predict passenger flow trends and make corresponding countermeasures in advance. This method not only improves the accuracy of prediction, but also improves the intelligent level of scenic area management, providing technical support for the sustainable development of scenic areas.

[0192] The scenic spot passenger flow trend prediction and analysis method based on the AI ​​large language model can effectively improve the intelligent level of scenic spot management and provide a scientific basis for scenic spot operations by comprehensively applying a variety of advanced data analysis technologies and machine learning algorithms. In practical applications, this method can be effectively used to monitor the passenger flow of scenic spots in real time and predict future passenger flow trends through prediction models. The following benefits can be achieved:

[0193] 1. Optimization of passenger flow management. Using AI big language models, scenic area managers can monitor and predict passenger flow trends in real time, so as to manage passenger flow in advance during holidays or special events, optimize resource allocation, effectively reduce congestion and improve tourist experience.

[0194] 2. Accurate resource allocation. Through an in-depth understanding of tourists' behavior patterns and preferences, scenic area managers can more accurately allocate security, service, cleaning and other resources to ensure sufficient human and material support at critical moments and improve the operational efficiency of the scenic area.

[0195] 3. Personalized marketing strategies. Based on the analysis of tourist data, scenic area managers can formulate more targeted marketing strategies to attract more target tourists, enhance the popularity and attractiveness of the scenic area, and strengthen market competitiveness.

[0196] 4. Timely response to risks: When faced with unexpected events or emergencies, help scenic area managers to adjust management strategies in a timely manner, reduce risks, and ensure the safety of tourists and order in scenic areas.

[0197] 5. Decision support dataization. The constructed multi-dimensional knowledge base and intelligent tool set provide rich data support for scenic area managers, making the decision-making process more scientific and accurate, promoting the development of scenic area management in the direction of data-driven, and improving the overall management level.

[0198] By using AI big language models to predict passenger flow trends, scenic area managers can predict passenger flow during holidays or special events in advance, optimize resource allocation, and reduce congestion and resource waste. For example, during holidays or special events, scenic area managers can manage passenger flow in advance based on the prediction results, optimize resource allocation, and improve the experience of tourists. At the same time, this method can also help scenic area managers better understand tourists' behavior patterns and preferences, so as to formulate more targeted marketing strategies and improvement measures to enhance the overall attractiveness and competitiveness of the scenic area.

[0199] The above technical features constitute the best embodiment of the present invention, which has strong adaptability and best implementation effect. Non-essential technical features can be added or reduced according to actual needs to meet the needs of different situations.

[0200] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A method for predicting and analyzing tourist flow trends in scenic spots based on an AI large language model, characterized in that: The steps include: S1. Full data collection and processing: Collect data from multiple sources related to scenic spot passenger flow, including LBS data, scenic spot data, holiday information and weather information, and third-party map data, and clean, integrate and standardize these data; S2. Build a knowledge base based on multi-dimensional data: Input the standardized data of the scenic spot itself and the operator's big data into the big data platform, and integrate the processed data into the knowledge base to build a comprehensive industry and scenario knowledge base to store and organize various types of information; Use RAG technology to search and generate content for knowledge bases; S3. Capability to build general models: Use machine learning technology to pre-train general models based on multi-dimensional data in the knowledge base to identify patterns and associations in the data; general models include scenic spot passenger flow prediction models, travel chain analysis models, and tourist portrait models; at the same time, based on pre-training, fine-tuning technology is used to fine-tune the general model using multiple source data related to scenic spot passenger flow; S4. Build a multi-agent system based on general model capabilities: Based on the general model obtained through pre-training and fine-tuning, build a multi-agent system including a travel group monitoring agent, a full-chain travel trajectory query agent, and a scenic spot passenger flow query agent; S5. Drive multiple agents to perform predictive analysis based on LLM: By integrating the AI ​​large language model LLM, multiple agents are driven to work together to perform intelligent predictive analysis on the tourist flow trend of scenic spots.

2. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 1, characterized in that: LBS data includes: user data, location data, timestamp, movement trajectory, geospatial data, point of interest data, user behavior data, and social network data.

3. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 1, characterized in that: In step S1, data cleaning includes removing irrelevant characters, removing duplicates, removing noise, filling missing values, data type and format verification, and logical error checking; Noise removal includes: identifying and cleaning non-numeric values ​​in the ticket sales data in the scenic spot data; logical error checking: including ensuring that the price in the ticket sales data in the scenic spot data does not exceed the set ticket sales price cap.

4. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 2, characterized in that: Step S2 comprises, S21, data integration: Integrate the processed data, including LBS data, scenic spot data, holiday information and weather information, and third-party map data; S22. Feature extraction: Extract key features and patterns from the integrated data. Key features include tourist behavior patterns, peak passenger flow periods, and the impact of weather on passenger flow. Use clustering algorithms to identify tourist behavior patterns, use time series analysis to predict peak passenger flow periods, and use regression models to assess the potential impact of weather changes on the number of tourists. S23, Rule and Pattern Recognition: Identify rules and patterns in data, including using hypothesis testing to determine the correlation between holidays and increased tourist flow, and applying time series analysis to identify the impact of weather changes on tourist flow; S24. Use RAG technology to build industry and scenario knowledge bases: Integrate the extracted features, identified rules and patterns, and generated conclusions into a structured database to build a knowledge base.

5. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 4 is characterized in that: Step S21 specifically includes: S211, reading LBS data, holiday information, weather information and third-party map data; S212, executing data information and conversion operations, including removing invalid records in LBS data and converting timestamp format; S213, using timestamp and location information as key fields, connecting multiple data sets of LBS data, holiday information, weather information, and third-party map data to form a unified data view; S214, selecting columns required for analysis from the connected data set, the columns including timestamp, location information, holiday information, weather information, and third-party map data; renaming the selected columns to unify the format of the data view; S215. Storing the data in a unified format in a knowledge base.

6. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 5, characterized in that: In step S22, using a clustering algorithm to identify tourist behavior patterns specifically includes: selecting location data, timestamp and weather information as features, applying a K-means algorithm to cluster tourist behaviors, and identifying different tourist behavior patterns.

7. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 6, characterized in that: In step S23, hypothesis testing is used to determine the correlation between holidays and the increase in tourist flow, including: Add holiday labels to dates in the combined dataset; Construct a linear regression model to evaluate the impact of holidays on the number of tourists; Prints a statistical summary of the linear regression model.

8. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 1, characterized in that: Step S3 includes: S31. Model pre-training: Use machine learning technology to pre-train the general model based on the multi-dimensional data in the knowledge base; S32. Model fine-tuning: Use multiple source data related to scenic spot passenger flow to fine-tune the pre-trained model, including adjusting the model's output layer to adapt to the output format of the scenic spot passenger flow prediction model, travel chain analysis model, and tourist portrait model.

9. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 8, characterized in that: Step S4 includes: S41. Design of the agent architecture: including the definition of the role and responsibility of each agent and the interaction mode between agents; the agents include the travel group monitoring agent, the full-chain travel trajectory query agent, and the scenic spot passenger flow query agent; S42. General model integration: Based on the general model obtained through pre-training and fine-tuning, the general model is integrated into each agent, including the travel group monitoring agent integrating a deep learning model to analyze tourist behavior patterns; the full-chain travel trajectory query agent integrating a sequence prediction model to track the movement paths of tourists; and the scenic spot passenger flow query agent integrating a time series analysis model to provide passenger flow data. S43, Functional Implementation and Optimization: The travel group monitoring agent includes an event detection system for identifying and responding to abnormal patterns in tourist behavior; the full-chain travel trajectory query agent includes a path tracking algorithm for processing and analyzing mobile data; the scenic spot passenger flow query agent includes a query optimization system for responding to passenger flow data query requests; S44. Collaborative working mechanism: Intelligent agents communicate through well-defined communication protocols and data sharing mechanisms.

10. The method for predicting and analyzing tourist flow trends in scenic spots according to claim 9, characterized in that: Step S5 comprises, S51, language input and intent analysis, including: S511, the scenic area administrator inputs a query or instruction to the system through natural language, and the input includes a passenger flow prediction request for a specific date, time period or event. The LLM first receives the natural language input of the scenic area administrator, and the LLM automatically recognizes and analyzes the administrator's intention and determines the type of task to be performed; S512, after receiving the input, LLM uses its natural language understanding (NLU) technology to analyze the administrator's query intent; S513. Based on the analysis results, LLM determines the specific task types that need to be performed, including short-term passenger flow prediction, long-term trend analysis, and real-time data monitoring and abnormal flow warning; S52, agent call and analysis execution: After determining the task type, according to the analysis results, LLM drives the multi-agent system to work together and call the corresponding sub-agent; the agents work together, share information and resources, select the best database path, access and analyze the historical data, real-time data and other relevant information stored in the knowledge base, and select the corresponding prediction model to perform scenic spot passenger flow prediction analysis; S53, result feedback and text output: After completing the analysis, the sub-agent will feed back the prediction results to LLM, and LLM will provide the prediction analysis results to the scenic area administrator in the form of natural language.

Images