Artificial intelligence route guide system and method for smart tourism

Abstract

The invention discloses an artificial intelligence route guide system and method for smart tourism, and relates to the technical field of intelligent decision making, and the method comprises the steps: extracting a multi-dimensional behavior feature vector based on tourist trajectory data, carrying out vectorization coding through employing a deep embedding method, carrying out the dynamic clustering of a coding result, and generating a tourist interest preference semantic micro-cluster; fusing the tourist interest preference semantic micro-cluster and the environment state data, and constructing a digital twin tourism city virtual environment in combination with city space layout information; performing tourist trajectory and tourist interest preference simulation based on a digital twin tourism city virtual environment, and performing multi-dimensional simulation and dynamic optimization of a guide route to generate a personalized guide route scheme; and pushing the personalized guide route scheme to the tourist, performing interactive path adjustment in combination with feedback information of the tourist, and generating an optimized personalized guide route scheme according to an adjustment result. According to the invention, the scientificity, the practicability and the dynamic response capability of the navigation route are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent decision-making, and in particular to an artificial intelligence route guide system and method for smart tourism. BACKGROUND

[0002] With the deep integration of information technology and tourism industry, smart tourism has become an important direction for promoting the high-quality development of cultural tourism. In recent years, smart tourism applications based on advanced information technologies such as the Internet of Things, big data, cloud computing and artificial intelligence have gradually become popular, and user guide services have also shifted from traditional static recommendations to personalized and dynamic route planning. In the prior art, a relatively typical implementation path is to collect tourist positioning data and point of interest access behavior in real time through mobile terminal devices, and to recommend a guide path in combination with a knowledge graph or a rule engine. For example, some platforms sort paths based on tourist access history and popular attraction popularity rankings, supplemented by navigation guidance, to improve tour efficiency and user experience.

[0003] The existing smart tourism guide method generally faces the problems of insufficient characterization of individual behavior characteristics of tourists and weak fusion of spatial semantic structure and environmental state information. In a dynamic tourism scenario, the behavior trajectory of a tourist is often influenced by multiple factors such as interest preferences, environmental changes, and time and space limitations, and static rule matching and uniform path templates cannot meet the personalized needs. At the same time, the guide recommendation process usually lacks a high-fidelity spatial semantic mapping environment, making it difficult to jointly model and dynamically optimize path accessibility, interest relevance and environmental adaptability, especially in terms of multi-source data coupling and real-time feedback response. Therefore, the existing smart tourism route guide technology has certain limitations in achieving fine-grained modeling and intelligent path optimization. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides an artificial intelligence route guide method for smart tourism to solve the problem of insufficient modeling accuracy of tourist individual interest preferences.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides an artificial intelligence route guide method for smart tourism, which comprises collecting tourism scene data and preprocessing to obtain tourist trajectory data and environmental state data; extracting a multi-dimensional behavior feature vector based on the tourist trajectory data, using a deep embedding method for vectorization coding, and performing dynamic clustering on the coding result to generate a tourist interest preference semantic micro-cluster; fusing the tourist interest preference semantic micro-cluster and the environmental state data, combining city space layout information to construct a digital twin tourism city virtual environment; simulating tourist trajectory and tourist interest preference based on the digital twin tourism city virtual environment, and performing multi-dimensional simulation and dynamic optimization of the guide route to generate a personalized guide route scheme; pushing the personalized guide route scheme to the tourists, combining tourist feedback information for interactive path adjustment, and generating an optimized personalized guide path scheme according to the adjustment result.

[0007] As a preferred scheme of the artificial intelligence route guide method for smart tourism, the tourist trajectory data and the environmental state data are obtained by the following steps, Collecting traffic travel records, positioning trajectories, scenic spot passenger flow and weather monitoring data to form tourism scene data; Aligning the tourism scene data in time, unifying the space coordinates and filling in the missing values to generate standardized tourism scene data, and extracting fields related to tourist movement to identify continuous trajectory point sequences; Based on the continuous trajectory point sequence, the movement trajectory path, the stop point, the speed change and the access order of the tourist are calculated to generate the tourist trajectory data; Extracting scenic spot temperature, humidity, air quality and crowd density from the standardized tourism scene data, performing analysis and normalization modeling to generate environmental state data.

[0008] As a preferred scheme of the artificial intelligence route guide method for smart tourism, the generation of the tourist interest preference semantic micro-cluster comprises the following steps, Extracting time interval, stay duration, access frequency and path length from the tourist trajectory data to generate original behavior features; Normalizing and vectorizing the original behavior features to construct a multi-dimensional behavior feature vector; Using a deep embedding method to vectorize the multi-dimensional behavior feature vector, combining a time sliding window mechanism to construct a dynamic semantic behavior representation sequence, and performing density perception clustering analysis to generate a tourist interest preference semantic micro-cluster.

[0009] As a preferred scheme of the artificial intelligence route guide method for smart tourism, the construction of the digital twin tourism city virtual environment comprises the following steps, According to the semantic micro-cluster of tourist interest preference and the environmental state data, a fusion data unit is formed in combination with corresponding geographical position coordinates and time stamps; The fusion data unit is spatially mapped with city space layout information to generate a spatio-temporal fusion data set coupled with semantic features, environmental states and city geographical structures; A multi-dimensional dynamic expression framework is established based on the spatio-temporal fusion data set to construct a compound correlation model of tourist behavior and environmental state; Based on the compound correlation model, the correlation characteristics between the semantic micro-cluster of tourist interest preference and the city space layout information are analyzed to generate multi-dimensional relationship data; The multi-dimensional relationship data is mapped to a unified spatial coordinate system to construct a structured digital space representation, and corresponding scene modeling parameters and dynamic configuration information are loaded to construct a digital twin tourism city virtual environment.

[0010] As a preferred scheme of the intelligent tourism artificial intelligence route guide method, the multi-dimensional dynamic expression framework is established based on the spatio-temporal fusion data set to construct a compound correlation model of tourist behavior and environmental state, and the specific steps are as follows, Multi-dimensional indexes reflecting the interaction characteristics of behavior and environment are extracted from the spatio-temporal fusion data set to form an initial feature set, and normalization and semantic modeling are performed to construct a multi-dimensional dynamic expression framework of behavior and environment relationship; The multi-dimensional dynamic expression framework is fused and analyzed with the semantic micro-cluster of tourist interest preference to generate a compound correlation model of tourist behavior and environmental state.

[0011] As a preferred scheme of the intelligent tourism artificial intelligence route guide method, the multi-dimensional dynamic expression framework is established based on the spatio-temporal fusion data set to construct a compound correlation model of tourist behavior and environmental state, and the specific steps are as follows, The semantic micro-cluster of tourist interest preference and the original behavior feature are loaded in the digital twin tourism city virtual environment to generate a behavior parameter template; A multi-scene tourist trajectory is constructed according to the behavior parameter template to form a simulated trajectory data set; The compound correlation model is used to perform spatial-time-semantic joint evaluation on the simulated trajectory data set to generate a simulation path score result; Based on the digital twin tourism city virtual environment, the semantic micro-cluster of tourist interest preference is analyzed by the compound correlation model to generate a candidate guide route set; The candidate guide route set is dynamically optimized according to the simulation path score result to select an optimized path set; A candidate path is extracted from the optimized path set and matched and selected according to the tourist interest preference and environmental adaptation conditions to generate a personalized guide route scheme.

[0012] As a preferred scheme of the artificial intelligence route guide method of the smart tourism, the specific steps of the generation of the optimized personalized guide path scheme are as follows, The personalized guide route scheme is pushed to the tourists as a navigation basis, and feedback information in the use process of the tourists is obtained. The feedback information is analyzed to extract feedback features of preference changes, path adaptability and environmental response. The personalized guide route scheme is corrected according to the feedback features, an adjusted path scheme is generated, comprehensive evaluation and optimization are performed, and an optimized personalized guide path scheme is generated.

[0013] In a second aspect, the present application provides an artificial intelligence route guide system for smart tourism, comprising a data preprocessing module, a behavior clustering module, an environment construction module, a path simulation module and a path optimization module. The data preprocessing module is used for collecting tourism scene data and preprocessing to obtain tourist trajectory data and environmental state data. The behavior clustering module is used for extracting a multi-dimensional behavior feature vector based on the tourist trajectory data, using a deep embedding method for vectorization coding, and dynamically clustering the coding results to generate a tourist interest preference semantic micro-cluster. The environment construction module is used for fusing the tourist interest preference semantic micro-cluster and the environmental state data, combining city space layout information, and constructing a digital twin tourism city virtual environment. The path simulation module is used for simulating tourist trajectories and tourist interest preferences based on the digital twin tourism city virtual environment, and performing multi-dimensional simulation and dynamic optimization of the guide route to generate a personalized guide route scheme. The path optimization module is used for pushing the personalized guide route scheme to the tourists, combining the tourist feedback information for interactive path adjustment, and generating an optimized personalized guide path scheme according to the adjustment results.

[0014] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, any step of the artificial intelligence route guide method of the smart tourism according to the first aspect of the present application is implemented.

[0015] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, any step of the artificial intelligence route guide method of the smart tourism according to the first aspect of the present application is implemented.

[0016] The application has the beneficial effects that: by fusing the tourist interest preferences and the environment state data, and combining the multi-dimensional information of the city space layout, a highly realistic digital twin tourism city virtual environment is constructed, and deep coupling of the virtual environment and the real environment in behavior, space and semantics is realized. The simulation and the simulation can be carried out in a multi-dimensional and multi-level real context, and the influence of the environment dynamics and the space structure on the tourist behavior is fully considered. The use is to provide a comprehensive and accurate virtual platform for the simulation and dynamic optimization of the guide route, ensure that the personalized path planning not only fits the tourist interest, but also meets the actual environment constraints, so as to achieve the effect of improving the scientificity, practicality and dynamic response ability of the guide route. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creating labor.

[0018] Fig. 1 Flow chart of the artificial intelligence route guide method for smart tourism.

[0019] Fig. 2 Schematic diagram of the artificial intelligence route guide system for smart tourism.

[0020] Fig. 3 Flow chart of the digital twin tourism city virtual environment construction.

[0021] Fig. 4 Flow chart of the composite association model construction. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the specific embodiments of the application will be described in detail below with reference to the drawings of the specification.

[0023] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited by the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is the embodiment alone or selectively excluded from other embodiments.

[0025] Referring to Figs. 1-4 For an embodiment of the present application, the embodiment provides an artificial intelligence route guide method for smart tourism, comprising the following steps: S1, collect tourism scene data and preprocess to obtain tourist trajectory data and environmental state data.

[0026] S1.1, collect traffic travel records, positioning trajectories, scenic area passenger flow and weather monitoring data to form tourism scene data.

[0027] Specifically, collect traffic travel records, positioning trajectories, scenic area passenger flow and weather monitoring data, and perform data normalization processing according to time stamp and spatial coordinates respectively; perform path analysis on traffic travel records to extract start and end points and travel trajectories; perform trajectory point continuity detection on positioning trajectories, eliminate abnormal points, and generate continuous trajectory sequences; divide scenic area passenger flow data into regions and count the number of people to form people flow density information in different scenic area partitions; perform time alignment on weather monitoring data to extract real-time weather state and change trend; fuse each processed data according to time stamp and spatial coordinates to form tourism scene data.

[0028] S1.2, perform time alignment, spatial coordinate unification and missing value filling on the tourism scene data to generate standardized tourism scene data, and extract fields related to tourist movement to identify continuous trajectory point sequences.

[0029] Specifically, correct and align the time stamp of each data in the tourism scene data according to a unified time standard to ensure that the time series of different data sources are synchronized; perform unified projection conversion on spatial coordinate data to achieve consistency of spatial coordinates; for missing data in the time series, use interpolation method to complete the missing values to ensure data continuity and integrity; extract fields related to tourist movement from the standardized tourism scene data, including location information, time stamp, speed and direction, etc. Key parameters; according to the extracted location information and time stamp, identify and select trajectory point sequences that meet the continuity condition to form continuous trajectory point sequences.

[0030] S1.3, calculate the movement trajectory path, stop point, speed change and visit order of tourists based on the continuous trajectory point sequence to generate tourist trajectory data.

[0031] Specifically, based on the continuous trajectory point sequence, the spatial distance calculation method is used to calculate the distance between adjacent trajectory points in turn, and the moving trajectory path of the tourist is spliced; by using the time difference method, the time interval of the continuous positioning points is calculated, combined with the position change analysis, it is judged whether the positioning points remain stable within a certain range, so as to identify the position of the stay point and count the continuous stay time of the stay point position to obtain the stay time; by using the speed calculation formula, the speed change of the tourist is calculated according to the time difference and spatial distance of adjacent trajectory points; combined with the timestamp sorting of the trajectory points, the visiting order of the tourist is determined, and the tourist trajectory data is generated.

[0032] S1.4, extracting the temperature, humidity, air quality and crowd density of the scenic spot from the standardized tourism scene data, analyzing and modeling, and generating environment state data.

[0033] Specifically, the temperature, humidity, air quality and crowd density fields of the scenic spot are extracted from the standardized tourism scene data, and the extracted temperature, humidity, air quality and crowd density fields are subjected to format conversion and outlier processing; the values of temperature, humidity, air quality and crowd density are converted to a unified range by using the normalization method; based on the normalization result, the mean, variance, maximum and minimum of the normalized data are calculated, and the environmental state characteristic indexes are formed according to the calculation results of the statistical quantities.

[0034] S2, extracting multi-dimensional behavior feature vectors based on tourist trajectory data, using deep embedding method for vectorization coding, and dynamically clustering the coding results to generate tourist interest preference semantic micro-cluster.

[0035] S2.1, extracting time interval, stay time, visit frequency and path length from tourist trajectory data to generate original behavior features.

[0036] Specifically, the time difference between adjacent trajectory points is calculated from the tourist trajectory data in turn to obtain the time interval; the continuous stay time of each stay point is counted to form the stay time; the number of times the tourist visits the same scenic spot is accumulated to generate the visit frequency; the total length of the path is calculated using the path point coordinates to form the path length; the time interval, stay time, visit frequency and path length are summarized to generate the original behavior features.

[0037] S2.2, normalizing and vectorizing the original behavior features to construct multi-dimensional behavior feature vectors.

[0038] Specifically, the original behavior features are processed by normalization method respectively, and the numerical ranges of time interval, stay duration, access frequency and path length are mapped to a unified standard interval, for example, each feature is linearly scaled to the interval [0, 1]; each normalized behavior feature is combined into a tourist behavior feature vector in dimension order; all tourist behavior feature vectors are summarized according to the corresponding tourist individuals to form a tourist multi-dimensional behavior feature vector set.

[0039] S2.3, using a deep embedding method to vectorize the multi-dimensional behavior feature vector, combining a time sliding window mechanism to construct a dynamic semantic behavior representation sequence, and performing density-aware clustering analysis to generate a tourist interest preference semantic micro-cluster.

[0040] Specifically, the multi-dimensional behavior feature vector is input into a pre-trained deep embedding network, and each dimension feature value in the tourist multi-dimensional behavior feature vector is sequentially processed by multi-layer nonlinear transformation, and is mapped to a fixed-dimensional embedding representation vector at the output layer of the embedding network to obtain a vectorized encoding representation; based on the time sliding window mechanism, the vectorized encoding representation is sliced according to the window size and sliding step to construct a dynamic semantic behavior representation sequence, reflecting the change law of behavior features over time; each semantic vector in the dynamic semantic behavior representation sequence is used as clustering input data, and a density-aware clustering method is used to calculate the distance between each pair of semantic vectors, and a set of semantic vectors with high local density and close distance is identified as the initial clustering core, and the semantic vectors containing similar neighborhood points are further expanded to form multiple interest behavior clusters, and the semantic micro-cluster of tourist interest preference is generated according to the content composed of the semantic features contained in each interest behavior cluster.

[0041] It should also be noted that the specific steps of the pre-trained deep embedding network are as follows: collect historical multi-dimensional feature data related to tourist behavior as training samples; normalize and standardize the training samples to ensure stable distribution of the training samples; construct a neural network structure containing multiple hidden layers, usually using fully connected layers or convolutional layers, and activation functions such as ReLU; use supervised or self-supervised learning methods to iteratively train the deep embedding network based on a loss function (such as contrastive loss or reconstruction error), and continuously adjust the weight parameters of the deep embedding network through a backpropagation algorithm; during training, batch training and early stopping strategies are used to prevent overfitting; after training, save the deep embedding network model parameters for subsequent encoding conversion of tourist behavior feature vectors; After sorting the tourist behavior feature vector set according to the timestamp, the specific steps of sampling processing by using the time sliding window mechanism are as follows: determining the time length and sliding step of the sliding window, for example, setting a fixed window size and window moving interval in units of minutes or hours; sorting the time series data according to the timestamp, and taking the continuous time segment from the starting point of the sequence as the first window according to the sliding window length; according to the set sliding step, sliding the window to the next continuous time segment, and sequentially dividing the entire time series; the time series data in each sliding window is used as an input unit of a time period for subsequent dynamic behavior representation calculation and analysis; all time series data divided by the sliding window are arranged in time sequence to form a dynamic semantic behavior representation sequence, reflecting the dynamic characteristics of the tourist behavior changing over time.

[0042] S3, fuse the tourist interest preference semantic micro-cluster and the environment state data, combine the city space layout information, and construct a digital twin tourism city virtual environment.

[0043] S3.1, according to the tourist interest preference semantic micro-cluster and the environment state data, combining the corresponding geographical position coordinates and the timestamp, form a fusion data unit.

[0044] Specifically, the tourist interest preference semantic micro-cluster and the environment state data are matched according to the geographical position coordinates and the timestamp, and for each tourist interest preference semantic micro-cluster data record, the corresponding geographical position coordinates and the timestamp are extracted; in the environment state data, find the environment state data record closest to the geographical position coordinates and within the acceptable range of the timestamp; merge the field of the matched tourist interest preference semantic micro-cluster and the environment state data record, and combine to form a fusion data unit.

[0045] S3.2, spatially map the fusion data unit and the city space layout information to generate a spatio-temporal fusion data set coupled with the semantic features, the environment state and the city geographical structure.

[0046] Specifically, the geographic location coordinate field in the fusion data unit is spatially aligned with the geographic structure coordinates in the city space layout information, and a spatial index-based matching method is used to determine the geographic location attribution of the fusion data unit. Specifically, the latitude and longitude coordinate information in the fusion data unit is extracted, the geographic area unit corresponding to the coordinates is searched in the spatial index structure, and the fusion data unit is determined to belong to the geographic area name corresponding to the search result, which is taken as the geographic location attribution determination result. Identify the city space unit corresponding to each fusion data unit; the semantic feature field and the environmental state field in the fusion data unit are taken as attribute information and attached to the geographic structure of the corresponding city space unit to form a spatio-temporal fusion data record containing semantic features, environmental states and city geographic structures; according to the chronological order, the corresponding time information, spatial coordinate information and behavior observation data content of each fusion data unit are extracted, and the extraction results are uniformly formatted and reorganized; then all the fusion data units after format reorganization are arranged in turn to construct a spatio-temporal fusion data set with uniform field structure; It should also be noted that the source of city space layout information includes basic geographic spatial data in city geographic information database, city public facility distribution data and tourism scenic area partition layer data. The basic geographic spatial data includes road network, building outline, water system distribution and green land boundary, etc., which are mainly sourced from GIS platform data or city planning open data; the city public facility distribution data includes transportation hubs, catering places, medical points and information service station locations, which can be sourced from government data interface or open POI database; the tourism scenic area partition layer data includes scenic area boundary, functional partition, path accessibility and landscape node information, which are usually sourced from CAD layers or vector map files provided by scenic area management parties, and are uniformly stored in vector format and processed for coordinate consistency through spatial reference system standard. After fusion processing, the basic geographic spatial data, city public facility distribution data and tourism scenic area partition layer data constitute city space layout information with uniform coordinate system and layer structure.

[0047] S3.3、Based on the spatio-temporal fusion data set, a multi-dimensional dynamic expression framework is established to build a complex association model of tourist behavior and environmental state.

[0048] S3.3.1, multi-dimensional indexes reflecting the interaction characteristics of behavior and environment are extracted from the spatio-temporal fusion data set to form an initial feature set, and normalization and semantic modeling are performed to build a multi-dimensional dynamic expression framework of behavior and environment relationship.

[0049] Specifically, fields related to tourist behavior and environmental state are identified from the spatio-temporal fusion dataset, including path selection, stay time, and visit frequency corresponding to the semantic micro-cluster of tourist interest preferences, as well as environmental state data such as air temperature, humidity, air quality, and crowd density matching the spatio-temporal coordinates. Cross-indicators of tourist behavior and environmental state are constructed, such as average stay time under specific air quality conditions or path change probability under high crowd density conditions, and are uniformly converted into numerical feature vectors to form an initial feature set. Each type of initial feature in the initial feature set is processed using the maximum and minimum value normalization method to obtain a multi-dimensional feature vector with uniform scale. Semantic modeling is performed on feature fields with semantic attributes using the word embedding method, and symbolic behavior features containing text types or classification identifiers (such as vehicle types, scenic area types, and activity categories) are mapped to dense vectors and concatenated with normalized environmental state features to construct a multi-dimensional dynamic expression framework of behavior and environmental relationship.

[0050] S3.3.2, the multi-dimensional dynamic expression framework is fused and analyzed with the tourist interest preference semantic micro-cluster to generate a complex association model of tourist behavior and environmental state.

[0051] Specifically, each type of normalized and semantically modeled behavior and environmental interaction feature vector in the multi-dimensional dynamic expression framework is extracted, and the cluster center feature vector in the tourist interest preference semantic micro-cluster is resampled to unify the vector dimension and time scale. The cosine similarity method is used to calculate the matching degree between the tourist interest preference semantic micro-cluster and each behavior vector in the multi-dimensional dynamic expression framework, and a matching degree matrix is constructed. Based on the matching degree matrix, weighted fusion is performed, the interest intensity of the semantic micro-cluster is taken as a weight factor, the correlation coefficient between behavior features and environmental state is adjusted, and a fusion feature set with semantic preference characteristics is formed. The covariance matrix of the fusion feature set is calculated, and eigenvalue decomposition or singular value decomposition is performed, and the principal components or common factors with higher cumulative contribution rate are selected, and the behavior-environment interaction factors are extracted through linear combination transformation. According to the corresponding relationship between the tourist behavior feature vector and the environmental state feature vector, the correlation between the two is quantified using regression analysis or mutual information calculation method; select the behavior-environment feature pairs with high correlation to establish the mapping relationship between the features, and form a complex association model of tourist behavior and environmental state.

[0052] S3.4, based on the complex association model, the association characteristics between the tourist interest preference semantic micro-cluster and the city space layout information are analyzed, and multi-dimensional relationship data is generated.

[0053] Specifically, the normalized tourist behavior features and environment state features in the composite association model are extracted, and the corresponding interest labels and time series information in the tourist interest preference semantic micro-cluster are obtained. The geographic location coordinates of the tourist interest preference semantic micro-cluster are spatially matched with the spatial unit coordinates in the city space layout information, the distribution density of the interest points in the spatial unit is calculated, and the expression is: ; wherein, denotes the interest point distribution density in the th spatial unit, denotes the number of interest points contained in the th spatial unit, denotes the area of the th spatial unit, which can adopt square meters or square kilometers as the unit, denotes the index number of the spatial unit; The frequency of the appearance of the tourist interest label and the city space function type in the same geographic location is counted to form a co-occurrence frequency matrix of the tourist interest label and the city space function type. Then, the interest label frequency is taken as the dependent variable, the spatial coordinates of the space function type are taken as the independent variable, a regression analysis is performed by using a geographic weighted regression method, local regression coefficients in different spatial positions are obtained, significant spatial response factors of the tourist behavior are extracted, and a set of spatial response factors of the tourist behavior is formed. The set of spatial response factors of the tourist behavior is coded into multiple dimensions such as time, space, interest and function type to generate multi-dimensional relationship data reflecting the coupling characteristics of tourist preferences and spatial structure.

[0054] S3.5, map the multi-dimensional relationship data to a unified spatial coordinate system, construct a structured digital space representation, and load corresponding scene modeling parameters and dynamic configuration information to construct a digital twin tourism city virtual environment.

[0055] Specifically, the spatial coordinate information in the multi-dimensional relationship data is obtained, and a geographic coordinate conversion method is used to convert spatial data from different sources into a unified spatial coordinate system, such as using the WGS84 coordinate system for unification. The converted spatial coordinate data is indexed and constructed, and a quadtree or R-tree structure is used to manage the spatial data hierarchically. According to the requirements of the digital twin tourism city virtual environment, corresponding scene modeling parameters are loaded, including geographic entity attributes, environment parameters and tourist behavior parameters, and are updated in combination with dynamic configuration information such as time variation and environmental dynamic data. Based on the processing result, a structured digital space representation is constructed to form a digital twin tourism city virtual environment that integrates multi-dimensional information such as space, time and behavior.

[0056] S4. Simulate the tourist track and the tourist interest preference based on the digital twin tourism city virtual environment, and perform multi-dimensional simulation and dynamic optimization of the tour route to generate a personalized tour route scheme.

[0057] S4.1. Load the tourist interest preference semantic micro-cluster and the original behavior characteristics in the digital twin tourism city virtual environment to generate a behavior parameter template.

[0058] Specifically, the semantic micro-cluster data of the tourist interest preference is received, and the semantic representation vector and the cluster center coordinate information of each micro-cluster are extracted. The original behavior characteristic data of the tourist is obtained, including specific indexes such as time interval, stay duration, access frequency and path length. According to the unique tourist identifier, the tourist interest preference semantic micro-cluster is matched and associated with the corresponding original behavior characteristics to form a joint data set containing semantic and behavior indexes. According to the construction requirements of the digital twin tourism city virtual environment, the behavior parameters are selected from the joint data set, and the time characteristics, spatial characteristics and interest semantics are structured and coded respectively. The behavior parameters are uniformly arranged according to the preset format specification to generate a behavior parameter template.

[0059] S4.2. Construct a multi-scene tourist track according to the behavior parameter template to form a simulated track data set.

[0060] Specifically, the multi-dimensional behavior indexes and semantic characteristics contained in the behavior parameter template are obtained. According to the requirements of different tourism scenes, the time characteristics, spatial characteristics and interest preferences of the tourist behavior indexes are combined to set the track generation rules. According to the track generation rules, the moving track of the tourist in different scenes is simulated to determine the geographical position coordinates and time stamp sequence of the continuous track points. The simulated track is adjusted in terms of the stay points and the track sequence to ensure that the simulated track conforms to the behavior mode and the interest preference. The track data generated under all scenes are summarized to form a simulated track data set containing diversified paths and behavior characteristics.

[0061] It should also be noted that the specific steps of setting the track generation rules are as follows: according to the multi-dimensional behavior characteristics extracted from the tourist interest preference semantic micro-cluster and the behavior parameter template, the spatial range and time window of track generation are determined; the key stay points and priorities in the track are determined in combination with the city spatial layout information and environmental state data; the duration of the stay points and the access sequence in the track are set according to the indexes such as stay duration, access frequency and path length in the behavior parameter template; the moving path between the stay points is randomly generated by using the probability distribution method to ensure the diversity and authenticity of the track; finally, the track generation rules are dynamically adjusted by comprehensively considering the environmental changes and the crowd density to ensure the high matching of the track with the actual tourism scene.

[0062] S4.3, utilize the composite correlation model to perform space-time-semantic joint evaluation on the simulated trajectory data set, and generate simulation path scoring results.

[0063] Specifically, for each trajectory in the simulated trajectory data set, the spatial matching degree, time consistency and semantic similarity indicators are extracted according to the composite correlation model respectively. The spatial matching degree is obtained by calculating the distance error of the trajectory point and the corresponding position in the city space layout information through the Euclidean distance. The time consistency is obtained by comparing the trajectory timestamp with the time sequence of the environment state data. The semantic similarity is calculated by the vector similarity of the tourist interest preference semantic micro-cluster and the trajectory behavior feature, and the expression is: ; Among them, represents the semantic similarity, represents the vector representation of the tourist interest preference semantic micro-cluster, represents the vector representation of the trajectory behavior feature, represents the vector dot product. The spatial matching degree, time consistency and semantic similarity indicators are normalized and weighted according to the preset weight to form a space-time-semantic joint score. Finally, the joint scoring results of all trajectories are summarized to generate simulation path scoring results.

[0064] It should be noted that the preset weight can be determined based on the analysis results of historical tourist behavior data, for example, by statistically analyzing a plurality of real tourist trajectories and corresponding environment state and interest preference data, calculating the contribution proportion of spatial matching degree, time consistency and semantic similarity to the overall tourist satisfaction or behavior prediction accuracy, and obtaining example weight values such as spatial matching degree weight 0.5, time consistency weight 0.3, and semantic similarity weight 0.2 as the basis for weighted calculation.

[0065] S4.4, based on the digital twin tourism city virtual environment, the tourist interest preference semantic micro-cluster is analyzed by the composite correlation model to generate a candidate tour route set.

[0066] Specifically, in the digital twin tourism city virtual environment, the composite correlation model is called to input the tourist interest preference semantic micro-cluster data and the corresponding spatial coordinates and scene information in the digital twin tourism city virtual environment. The correlation strength between the tourist interest preference semantic micro-cluster and each scenic spot is calculated based on the composite correlation model, and the expression is: ; Among them, represents the correlation strength between the tourist interest preference semantic micro-cluster and the scenic spot , represents the tourist interest preference semantic micro-cluster with the scenic spots the matching degree at the semantic text level, representing the semantic micro-cluster of the tourist interest preference with the scenic spots the matching degree at the geographical spatial position, representing the semantic micro-cluster of the tourist interest preference with the scenic spots the matching degree at the environmental state dimension, the weighting coefficient representing the semantic matching degree, the weighting coefficient representing the geographical position matching degree, the weighting coefficient representing the environmental adaptation degree, representing the first the semantic micro-cluster of the tourist interest preference, representing the first the scenic spot; The correlation strength is quantified by using a similarity measurement method such as cosine similarity or Pearson correlation coefficient, and then a set of high-correlation scenic spots is screened out according to the correlation strength ranking. A set of candidate tour routes is generated by combining the tourist behavior path preference and the visit order constraint. Each route in the set of candidate tour routes includes a spatial sequence and a time sequence of multiple scenic spots. Finally, a set of candidate tour routes that meet the interest preference and satisfy the spatial continuity is output.

[0067] S4.5, dynamically optimizing the set of candidate tour routes according to the simulation path scoring results to screen out an optimized path set.

[0068] Specifically, according to the simulation path scoring results, the spatial-time-semantic joint scoring value corresponding to each route in the set of candidate tour routes is read. An initial screening threshold is set according to the scoring results, for example, the initial screening threshold is 0.7 in the example. The candidate tour routes with a score higher than the initial screening threshold are screened out. For the screened candidate tour routes, a heuristic optimization algorithm is used to locally adjust the routes according to the scoring results and the spatial continuity requirement. The adjustment steps include inserting, deleting or exchanging the scenic spot order. After adjustment, the simulation path score of each route is recalculated by using a weighted linear fusion method. The score changes before and after adjustment are compared. The routes with a significant score improvement are retained. The iteration is repeated to finally output an optimized path set that meets the scoring requirement and has good spatial continuity.

[0069] It should also be noted that the specific steps of setting the initial screening threshold according to the score results: count the score distribution of the simulation path of all lines in the candidate tour line set, calculate the maximum, minimum and mean of the score results; according to the characteristics of the score distribution, select a suitable statistic as the initial screening threshold, for example, the mean value can be selected as the initial screening threshold in the example, or the initial screening threshold is determined based on the interval range of the mean value plus or minus a certain proportion; The initial screening threshold is applied to the screening operation to distinguish between high-score and low-score candidate tour lines; The initial screening threshold can be adjusted according to actual application requirements to ensure that the screened lines meet the scoring standards and meet the expected quantity requirements.

[0070] S4.6, extract the candidate path from the optimized path set, and perform matching screening of the tourist interest preference and environmental adaptation condition to generate a personalized tour line scheme.

[0071] Specifically, each candidate path is extracted from the optimized path set in turn, and the corresponding path point sequence and related attribute information are obtained; according to the preference features in the tourist interest preference semantic micro-cluster, the interest matching degree and the environmental adaptation degree are calculated in combination with the environmental state data of the path point, and the expression is: ; Among them, represents the interest matching degree, represents the number of path points, represents the index of the path point, represents the feature vector of the tourist interest preference semantic micro-cluster, represents the behavior feature vector of the th path point; ; Among them, represents the environmental adaptation degree, represents the environmental state vector of the th path point, represents the tourist environmental preference vector, represents the Euclidean distance, represents the adjustment parameter; A weighted scoring method is used to comprehensively evaluate the interest matching degree and the environmental adaptation degree according to the preset weight, and generate a matching score; according to the preset initial screening threshold, the candidate paths with a matching score higher than the initial screening threshold are selected as the personalized tour line scheme; repeat the steps until all candidate paths are screened, and finally form a personalized tour line scheme that meets the tourist interest preference and environmental adaptation condition.

[0072] S5, push the personalized tour line scheme to the tourists, combine the tourist feedback information to perform interactive path adjustment, and generate an optimized personalized tour path scheme according to the adjustment result.

[0073] S5.1, push the personalized tour route scheme to the tourists as navigation basis, and obtain feedback information in the use process of the tourists.

[0074] Specifically, the personalized tour route scheme is pushed to the tourists through a terminal, and after the tourists receive the personalized tour route scheme, the navigation path and related information are displayed; in the use process of the tourists, the positioning trajectory, the stay duration, the path deviation and the operation feedback information of the tourist equipment are collected in real time; the collected feedback information is uploaded to the data storage at regular intervals or when an event is triggered, and the feedback information in the use process of the tourists is obtained.

[0075] S5.2, analyze the feedback information, and extract the feedback features of preference change, path adaptability and environmental response.

[0076] Specifically, the feedback information in the use process of the tourists is arranged in chronological order, the path deviation of the navigation path is calculated by using a dynamic time warping method according to the change of the positioning trajectory in the feedback information, and the path adaptability feature is extracted; according to the time change of the stay duration sequence and the access frequency sequence, the stay duration and the access frequency value corresponding to each interest label are arranged in chronological order, and then the values at adjacent time points are calculated by a difference value, to obtain the change amount of each interest label in a continuous time period, and finally the change amount is taken as the basis for measuring the interest preference change feature of the tourists; the environmental response feature is extracted by using a weighted average method in combination with the environmental state data and the environmental response index in the feedback; each type of feedback feature is standardized by using a minimum-maximum normalization method, to form the feedback features of preference change, path adaptability and environmental response.

[0077] S5.3, according to the feedback features, correct the personalized tour route scheme, generate an adjusted path scheme, and perform comprehensive evaluation and optimization to generate an optimized personalized tour path scheme.

[0078] Specifically, according to the feedback features, the path point order and the stay time in the personalized tour route scheme are corrected by using a weighted adjustment method to generate an adjusted path scheme; according to the adjusted path scheme, the spatial continuity, the interest matching degree and the environmental adaptation degree of the adjusted path are comprehensively evaluated by using a composite correlation model, and the comprehensive score of the adjusted path is calculated, and the expression is: wherein, represents the comprehensive score of the adjusted path, represents the interest matching degree, represents the environmental adaptation degree score, represents the weight coefficient of the interest matching degree score, represents the weight coefficient of the environmental adaptation degree score, and satisfies​​ According to the comprehensive score, a ranking algorithm is used to screen out a path with a higher score as an optimized personalized tour guide path scheme.

[0079] The embodiment also provides an artificial intelligence route tour guide system for smart tourism, which comprises a data preprocessing module, a behavior clustering module, an environment construction module, a path simulation module and a path optimization module; the data preprocessing module is used for collecting tourism scene data and preprocessing to obtain tourist trajectory data and environment state data; the behavior clustering module is used for extracting a multi-dimensional behavior feature vector based on the tourist trajectory data, performing vectorization coding by using a deep embedding method, and performing dynamic clustering on the coding result to generate a tourist interest preference semantic micro-cluster; the environment construction module is used for fusing the tourist interest preference semantic micro-cluster and the environment state data, combining city space layout information, and constructing a digital twin tourism city virtual environment; the path simulation module is used for simulating the tourist trajectory and the tourist interest preference based on the digital twin tourism city virtual environment, and performing multi-dimensional simulation and dynamic optimization of the tour guide route to generate a personalized tour guide route scheme; and the path optimization module is used for pushing the personalized tour guide route scheme to the tourists, performing interactive path adjustment in combination with the tourist feedback information, and generating an optimized personalized tour guide path scheme according to the adjustment result.

[0080] The embodiment also provides a computer device suitable for the artificial intelligence route tour guide method for smart tourism, which comprises a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the artificial intelligence route tour guide method for smart tourism as proposed in the above embodiment.

[0081] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device which are connected through a system bus. The processor of the computer device is used for providing computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless communication can be realized by WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, a trackball or a touchpad arranged on the shell of the computer device, or can be an external keyboard, a touchpad or a mouse, etc.

[0082] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the artificial intelligence route tour guide method for smart tourism proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0083] In summary, the application constructs a highly realistic digital twin tourism city virtual environment by fusing tourist interest preferences and environmental state data and combining multi-dimensional information of urban spatial layout, realizes deep coupling of virtual environment and real environment in behavior, space, and semantics, enables simulation and emulation to be performed in a multi-dimensional and multi-level real context, and fully considers the influence of environmental dynamics and spatial structure on tourist behavior. The application is used for providing a comprehensive and accurate virtual platform for simulation and dynamic optimization of a tour guide route, ensuring that personalized path planning not only fits tourist interests but also conforms to actual environmental constraints, thereby achieving the effect of improving the scientificity, practicality, and dynamic response capability of the tour guide route.

[0084] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application but not limit the application. Although the application is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the application, and all should be covered in the scope of the claims of the application.

Claims

1. An artificial intelligence route guide method for smart tourism, characterized by: The application relates to a method for generating an individualized tour route for a tourist in a digital twin tourist city, and a system thereof. Tourist scene data is collected and preprocessed to obtain tourist trajectory data and environment state data; Multi-dimensional behavior feature vectors are extracted from the tourist trajectory data, vectorization coding is performed by using a deep embedding method, and dynamic clustering is performed on the coding results to generate tourist interest preference semantic micro-clusters; The tourist interest preference semantic micro-clusters and the environment state data are fused, combined with city space layout information, and a digital twin tourist city virtual environment is constructed; Based on the digital twin tourist city virtual environment, tourist trajectory and tourist interest preference simulation are performed, multi-dimensional simulation and dynamic optimization of a tour route are performed, and an individualized tour route scheme is generated; The individualized tour route scheme is pushed to the tourist, interactive path adjustment is performed in combination with tourist feedback information, and an optimized individualized tour path scheme is generated according to the adjustment results. 2.The smart travel artificial intelligence route guide method of claim 1, wherein: The specific steps of obtaining the tourist trajectory data and the environment state data are as follows, Traffic travel records, positioning trajectories, scenic spot passenger flow and weather monitoring data are collected to form tourist scene data; The tourist scene data is time-aligned, space coordinates are unified, and missing values are filled to generate standardized tourist scene data, and fields related to tourist movement are extracted to identify continuous trajectory point sequences; Based on the continuous trajectory point sequences, the movement trajectory path, the stay point, the speed change and the visiting sequence of the tourist are calculated to generate the tourist trajectory data; From the standardized tourist scene data, scenic spot temperature, humidity, air quality and crowd density are extracted, analyzed and modeled to generate environment state data. 3.The smart travel artificial intelligence route guide method of claim 1, wherein: The specific steps of generating the tourist interest preference semantic micro-clusters are as follows, Time intervals, stay durations, visit frequencies and path lengths are extracted from the tourist trajectory data to generate original behavior features; The original behavior features are normalized and vectorized to construct multi-dimensional behavior feature vectors; The multi-dimensional behavior feature vectors are vectorization coded by using a deep embedding method, a dynamic semantic behavior representation sequence is constructed by combining a time sliding window mechanism, and density perception clustering analysis is performed to generate tourist interest preference semantic micro-clusters. 4.The smart travel artificial intelligence route guide method of claim 1, wherein: The specific steps of constructing the digital twin tourist city virtual environment are as follows, According to the tourist interest preference semantic micro-clusters and the environment state data, corresponding geographical position coordinates and time stamps are combined to form fusion data units; The fusion data units and city space layout information are spatially mapped to generate a spatio-temporal fusion data set coupled with semantic features, environment states and city geographical structures; A multi-dimensional dynamic expression framework is established based on the spatio-temporal fusion data set, and a composite association model of tourist behavior and environment state is constructed; Based on the composite association model, the association features between the tourist interest preference semantic micro-clusters and the city space layout information are analyzed to generate multi-dimensional relationship data; The multi-dimensional relationship data is mapped to a unified space coordinate system to construct a structured digital space representation, corresponding scene modeling parameters and dynamic configuration information are loaded, and a digital twin tourist city virtual environment is constructed. 5.The smart travel artificial intelligence route guide method of claim 4, wherein: The specific steps of establishing a multi-dimensional dynamic expression framework based on the spatio-temporal fusion data set and constructing a composite association model of tourist behavior and environment state are as follows, Extracting multi-dimensional indicators reflecting the interaction characteristics of behavior and environment from the spatio-temporal fusion data set, forming an initial feature set, and performing normalization and semantic modeling to construct a multi-dimensional dynamic expression framework of the relationship between behavior and environment; Fusion analysis of the multi-dimensional dynamic expression framework and the tourist interest preference semantic micro-cluster to generate a composite association model of tourist behavior and environmental state. 6.The smart travel artificial intelligence route guide method of claim 1, wherein: The specific steps of generating the personalized tour route scheme are as follows, Load the tourist interest preference semantic micro-cluster and the original behavior feature in the digital twin tourism city virtual environment to generate a behavior parameter template; According to the behavior parameter template, construct multi-scene tourist trajectories to form a simulated trajectory data set; Using the composite association model, the spatial-time-semantic joint evaluation of the simulated trajectory data set is performed to generate a simulation path scoring result; Based on the digital twin tourism city virtual environment, the tourist interest preference semantic micro-cluster is analyzed by the composite association model to generate a candidate tour route set; According to the simulation path scoring result, the candidate tour route set is dynamically optimized to filter an optimized path set; From the optimized path set, extract the candidate path and perform matching filtering of the tourist interest preference and environment adaptation conditions to generate the personalized tour route scheme. 7.The smart travel artificial intelligence route guide method of claim 1, wherein: The specific steps of generating the optimized personalized tour path scheme are as follows, Push the personalized tour route scheme to the tourists as a navigation basis, and obtain the feedback information in the process of use; Parse the feedback information to extract feedback features of preference changes, path adaptability and environmental response; According to the feedback features, the personalized tour route scheme is revised to generate an adjusted path scheme, and comprehensive evaluation and optimization are performed to generate the optimized personalized tour path scheme.

8. An artificial intelligence route guide system for smart tourism, based on the artificial intelligence route guide method for smart tourism according to any one of claims 1 to 7, characterized in that: It includes a data preprocessing module, a behavior clustering module, an environment construction module, a path simulation module and a path optimization module; The data preprocessing module is used for collecting tourism scene data and pre-processing to obtain tourist trajectory data and environmental state data; The behavior clustering module is used for extracting multi-dimensional behavior feature vectors based on tourist trajectory data, using deep embedding method for vectorization coding, and dynamically clustering the coding results to generate tourist interest preference semantic micro-clusters; The environment construction module is used for fusing the tourist interest preference semantic micro-cluster and the environmental state data, combining the city space layout information to construct the digital twin tourism city virtual environment; The path simulation module is used for simulating tourist trajectories and tourist interest preferences based on the digital twin tourism city virtual environment, and performing multi-dimensional simulation and dynamic optimization of the tour route to generate a personalized tour route scheme; The path optimization module is used for pushing the personalized tour route scheme to the tourists, combining the tourist feedback information for interactive path adjustment, and generating an optimized personalized tour path scheme according to the adjustment result. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the artificial intelligence route tour method of the smart tourism of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the artificial intelligence route tour method of the smart tourism of any one of claims 1-7.

Images