Intelligent optimized navigation method based on dynamic people flow and user interest

Abstract

The invention discloses an intelligent optimized navigation method based on dynamic people flow and user interest, which comprises the following steps: acquiring people flow density, interest data and time sequence data by a system through a sensor, predicting future people flow density by utilizing an ARIMA model, and marking a congestion risk area. And a personalized optimal visiting path is generated through a Dijkstra algorithm in combination with user interest portraits and venue item characteristics. In the path generation process, the system comprehensively evaluates the project coverage rate, the time limit and the congestion avoidance condition of the path to ensure that the recommended path can meet the user demand to the maximum extent, and dynamically adjusts the path planning weight, shunts crowds and avoids congestion. And if the actual visiting path of the user deviates or the real-time density in the venue changes, the system can quickly update the path and recommend the path again, so that the efficient and safe visiting experience is ensured. Through the method, the stadium can effectively optimize people flow management, improve the experience of visitors, and provide scientific support and decision basis for operation management.

Description

Technical Field

[0001] This patent involves the field of intelligent navigation and optimization, specifically a method for generating and optimizing study routes based on dynamic crowd flow prediction and user interest modeling. Background Art

[0002] As life experience halls and project learning centers gradually become important places for the public to participate in popular science education and study tours, how to balance the personalized needs of visitors with the carrying capacity of the venues and improve the visiting experience has become an urgent problem to be solved in the industry. The traditional fixed route planning method is difficult to meet the diverse interests and preferences of visitors. At the same time, visiting activities during peak periods such as holidays are often affected by crowds and affect efficiency and safety. In addition, visitors often deviate from the preset route during the actual tour, and these behavioral data are not fully utilized, making it difficult to continuously improve route optimization.

[0003] Existing patent documents such as patent number CN201811624965.2 disclose a real-time tourist flow diversion and flow limit information prompt method, device and storage medium, including the following steps: execute the application installed on the user's mobile terminal, obtain the scenic area navigation map through the management server; determine at least one display location on the scenic area navigation map based on a preset method; obtain the road condition information corresponding to each display location; display the corresponding scenic area road condition information at each display location. The present invention can enable tourists to obtain the corresponding scenic area road condition information displayed at each display location by executing the application installed on the user's mobile terminal, and obtain whether a certain place in the scenic area is severely congested, congested, crowded, sparsely populated or tourist density.

[0004] However, the above method is managed according to the user's reservation method, which has a certain delay efficiency and cannot manage the flow of people in real time according to the actual movement status of the flow of people. It has certain limitations and lags, and still has the risk of congestion.

[0005] In order to help venues manage personnel flow more effectively and reduce the accident rate, this paper designs an intelligent optimization navigation method based on dynamic human flow and user interests based on the fusion of human flow data, interest data and behavior data, and generates efficient dynamic navigation and diversion strategies through the collaborative work of prediction models and behavior learning algorithms to avoid congestion in venues. Summary of the invention

[0006] The purpose of the present invention is to provide an intelligent optimization navigation method based on dynamic human flow and user interests, aiming to improve the problem of human flow navigation, which is cumbersome to change and affects paper cutting.

[0007] The present invention is implemented as follows: an intelligent optimization navigation method based on dynamic human flow and user interests, including the following methods: Step 1: Deploy sensors at each main channel and project entrance. The system collects static area information of the venue and first initializes the static information; Step 2: The system collects the reservation information of each user, generates the interest profile of each user, makes the first crowd flow prediction, and uses the initial crowd flow and density information to build a weighted graph model; Step 3: The system calculates the real-time crowd density in each area based on the crowd flow data collected by sensors in each area in real time, records it in real time, generates a time series collection, and inputs it into the ARIMA model for future density prediction; Step 4: The system calculates the interest matching degree of each user for the project, combines the real-time crowd density and the distance between areas, and uses the Dijkstra path planning algorithm to generate the initial optimal recommended path from the entrance area to the exit area according to the weight; Step 5: After generating the preliminary optimal recommended path, the system further evaluates the path in real time, sorts the candidate paths according to the three conditions of coverage, time limit and congestion avoidance, and finally selects the optimal path as the user's recommended route; Step 6: During the navigation process, the system dynamically adjusts the weights of the path edges in the weighted graph according to the real-time crowd density, and adjusts the priority of the navigation path in real time; Step 7: During the navigation process, when the user deviates from the path or congestion occurs, the system replans the path based on the adjusted weights using the Dijkstra algorithm, regenerates the optimal path in real time, and avoids the congested area.

[0008] As an embodiment of the present invention, the following steps are included: In step 1: at the initial stage of system operation: loading the basic static information of the venue; the sensor is an image collection device and an infrared counter; Extract the structured information of the venue from the database, including the collection of all areas: , where each region Including specific location, area , Maximum capacity and topic tags ; Constructing region information as a weighted graph , where the node set Represents all regions, edge sets Represents the connection relationship between regions. The initial weight of the edge is the physical distance between regions. ; Then, the system loads all the experience items in the venue , each project Include its hashtag , average experience duration , Maximum number of people that can be received at one time , and the region where the project is located .

[0009] As an embodiment of the present invention, in step 2: the system obtains the reservation information of the user Extracting interest preferences , Estimated time of arrival and planned visit duration , generate a preliminary interest profile for each user .

[0010] As an embodiment of the present invention, in step 3, the system collects the graphic information of each venue through the image collection device and the infrared counter to generate a collection , transmitted to the system's data center, the system collects the real-time number of people entering and number of people leaving , combined with the area , and according to the formula: ; Calculate the current crowd density , updated every minute , and store it as a time series , and then predict the future density through the ARIMA model , and mark the predicted density exceeds the regional capacity congestion risk areas.

[0011] As an embodiment of the present invention, the prediction formula of the ARIMA model is: ; The model parameter p represents the autoregressive order, d is the number of differences, and q is the moving average order. The model fits the parameters by minimizing the mean square error between the predicted error and the true value; Predict the density value of the next k time steps, and the model generates prediction sequences in sequence ; Each predicted value According to the historical sequence values ​​and the prediction error calculation, the prediction process is iterated step by step, using the prediction value of the previous step as the input of the next step; the shape of the output prediction sequence is (k, 1).

[0012] As an implementation mode of the present invention, after the prediction is completed, the system performs congestion detection on the density prediction result; If there is any parameter j greater than 0 and less than or equal to k in the forecast time series, such that , then the area Marked as a congestion risk area; this mark is directly input into the path planning module as the basis for dynamically adjusting the weight. The system optimizes the user path in real time and guides users to avoid high-density areas.

[0013] As an implementation mode of the present invention, in step 4: based on the user in the user set U Interest profile and the topic labels of the items in the item set P and weight factor , the system calculates each user's interest matching degree for the project ; Combined with real-time crowd density and distance between regions The system uses the Dijkstra path planning algorithm according to the weights Generate the optimal recommended path from the user's entry area to the exit area .

[0014] As an embodiment of the present invention, in step 5: after generating the preliminary optimal recommended path, the system further evaluates the path and selects the path that meets the following conditions as the final recommended path : First, the system calculates the path based on the set of items covered by the path and the set of all items that the user is interested in. Project coverage in ,give priority to the path that covers the highest proportion of high-interest projects; Secondly, based on the average experience time of the project and the travel time of the path edges Calculate the total time of the path to ensure the total duration of the path Not exceed the user's planned visit duration ; Finally, combined with the prediction results , eliminating the future predicted density congested area paths; The system sorts candidate routes according to three conditions: coverage, time limit and congestion avoidance, and finally selects the best route as the user's recommended route.

[0015] As an embodiment of the present invention, in step 6: the system continuously collects data from each area through sensors. Real-time number of people entering and number of people leaving , calculate the real-time crowd density And update the density data to the weighted graph G=(V,E). According to the real-time density, the system dynamically adjusts the weight of the path edge. ; When the density of a certain area exceeds the capacity When , increase the weight of the relevant edge To reduce the recommendation priority; when the density of a certain area is lower than the threshold DensityThreshold, reduce the weight In order to attract more users, the system optimizes path planning by dynamically adjusting weights, guiding users to avoid high-density areas and divert them reasonably, thereby improving the visiting experience and venue operation efficiency.

[0016] As an embodiment of the present invention, in step 7: when the system detects the user's current location Deviating from the original path or a certain area Real-time density Exceeding capacity When congestion occurs, the system adjusts the weights Use Dijkstra algorithm to re-plan the path, and the new path starts from the user's current location Departure, the end point is the exit area , and cover the collection of high-interest projects that have not yet been visited , weight calculation takes into account the path distance , real-time crowd density and interest matching score : New path generated by the system Ensure maximum coverage of high-interest projects, avoid congested areas, and meet the user's remaining visit time limit; At the same time, the system updates the user's visit order, prompts the route change and provides new estimated arrival time and coverage project details.

[0017] The beneficial effects of the present invention are as follows: The present invention proposes an intelligent optimization navigation method based on dynamic crowd flow prediction and user interest modeling. By integrating static area information, real-time dynamic data and user interest portraits, a weighted graph model is constructed, and combined with dynamic weight adjustment and path planning algorithms, real-time optimization and personalized recommendation of visiting routes are achieved, effectively improving the flexibility of the system and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Overall structure diagram of the method.

[0019] Figure 2 Figure G=(V,E) schematic diagram. DETAILED DESCRIPTION

[0020] The present invention will be further described below in conjunction with the accompanying drawings and examples.

[0021] The present invention proposes an intelligent optimization navigation method based on dynamic crowd flow and user interests. The method analyzes the popularity distribution of life experience projects, the regional distribution of projects in the venue, and the interest preferences of the masses, combined with the number of reservations and real-time visit data, to recommend the best study tour route for the masses. The system can monitor the crowd density in each area in real time, and predict future peak areas to achieve crowd diversion and dynamic adjustment. During the visit, the system will collect the user's actual visiting path and behavior changes, and combine user feedback to continuously optimize the route generation algorithm to make it more in line with the visiting habits and needs of the masses.

[0022] This method integrates crowd flow data, interest data, and behavior data, and generates efficient dynamic navigation and diversion strategies through the collaborative work of prediction models and behavioral learning algorithms, thereby avoiding congestion in the venue, improving the visiting experience, and providing a scientific basis for venue operations and an intelligent solution for optimizing path planning.

[0023] The present invention is an intelligent optimization navigation model, whose input is a variety of heterogeneous data, including numerical data (such as crowd density, regional area), categorical data (such as project theme, user interest preference), time series data (such as appointment number change trend, visiting path data), etc., and the output is a personalized study tour route.

[0024] The overall structure of the method is as follows Figure 1 As shown in the figure, the system first initializes the static information through the venue's static area information and user interest portraits, and then conducts the first crowd monitoring and prediction, using this as the initial crowd flow and density information to build a weighted graph model. On this basis, the crowd density, user interest matching score and the distance between venue nodes are combined with a dynamic weight adjustment mechanism to dynamically optimize the path weight and generate a preliminary path. When the system starts running, the system uses the prediction module to continuously analyze the trend of future regional density, integrates the information of congestion risk areas into the path planning process, adjusts the optimal recommended path that meets user needs through the path planning algorithm, and regenerates the optimal path based on the latest dynamic information when the user deviates from the path.

[0025] Here are the detailed steps: Step 1: At the beginning of the system operation, you first need to load the basic static information of the venue and deploy the necessary sensors. The sensors are image collection devices with monitoring functions and infrared counters with counting functions to provide support for subsequent data collection and analysis. The structured information of the venue is extracted through the database, including the collection of all areas. , where each region Including specific location, area , Maximum capacity and topic tags , such as "Life Science" or "Mysteries of the Human Body". This regional information is constructed as a weighted graph ,like Figure 2 As shown, the node set Represents all regions, edge sets Represents the connection relationship between regions. The initial weight of the edge is the physical distance between regions. At the same time, the system loads all the experience items in the venue. , each project Include its hashtag , average experience duration , Maximum number of people that can be received at one time , and the region where the project is located This information lays an important foundation for the generation and optimization of visiting routes.

[0026] Step 2: Then, the system collects the reservation information of users on that day. Extract user interests and visit plans, and generate a preliminary interest profile for each user Specifically, each user The interest profile is based on the interest score of each item. These scores are usually floating point values ​​from 0 to 1, which reflect the user's preference for the project. In addition, the system also obtains the user's scheduled arrival time and total planned visit duration , in order to generate a personalized time-optimized path. To achieve real-time dynamic monitoring, the system deploys a collection of sensors at the main channels and project entrances of the venue. , and start the equipment to collect data and transmit it to the data center for subsequent dynamic crowd flow prediction and path planning.

[0027] Step 3: These sensors collect data in real time and leave The personnel data of each area is calculated according to the formula: ; Calculate the real-time density of people in the area. Real-time crowd density are recorded continuously and generate historical time series: ; Input to the ARIMA model for future density forecasting. The shape of the input sequence is (m,1), where m represents the number of historical time steps, which is usually a fixed value. Before the sequence is input, ensure that the data is a stationary time series. If there is a trend or seasonal fluctuation, it is processed through d times difference operation.

[0028] The difference formula is The number of differences d is a model parameter, which is determined according to the stationarity of the actual data.

[0029] The ARIMA model consists of three parts: autoregression (AR), difference (I) and moving average (MA). The autoregression part predicts the current density through the values ​​of the previous p time steps, and its mathematical expression is: ; in is the autoregression coefficient, is the noise term. The moving average part adjusts the predicted value through historical errors, and the formula is ,in is the moving average coefficient, is the error value at the past time step.

[0030] After combining autoregression (AR), differencing (I) and moving average (MA), the prediction formula of the ARIMA model is: ; The model parameter p represents the autoregressive order, d is the number of differences, and q is the moving average order. The model fits the parameters by minimizing the mean square error between the prediction error and the true value.

[0031] Predict the density value of the next k time steps, and the model generates prediction sequences in sequence Each predicted value Based on the historical sequence values ​​and the prediction error calculation, the prediction process is iterated step by step, using the prediction value of the previous step as the input of the next step. The shape of the output prediction sequence is (k, 1).

[0032] After the prediction is completed, the system performs congestion detection on the density prediction results. If there is any parameter j greater than 0 and less than or equal to k in the prediction time series, , then the area Marked as a congestion risk area. This mark is directly input into the path planning module as an important basis for dynamically adjusting weights, helping the system to optimize user paths in real time, guide users to avoid high-density areas, and improve visit efficiency and experience. Through the precise prediction and dynamic analysis of the ARIMA model, the system can identify potential risks in advance and ensure the smoothness and safety of venue operations.

[0033] Step 4: The system first calculates the user The interest matching degree of each item in the item set P This matching degree is achieved by combining the user's interest profile , the project's theme tag and weight factor The specific formula is ,in is the weight of the project feature, which can be set by the venue staff. It is the user's interest score for the project topic. Through calculation, the system obtains the user's interest score for all projects and provides a priority basis for the path planning stage.

[0034] Combining the static layout of the venue and the real-time dynamic data, the system constructs a weighted graph G=(V,E). The node set V=A represents all the venue areas, and the edge set Represents the path between regions. The weight of each edge in the weighted graph It consists of three parts: physical distance , Real-time crowd density , and the user's interest matching score for projects in the area .

[0035] The weight formula is: ,in They represent the importance of path distance, crowd density and interest matching respectively.

[0036] Based on the weighted graph G=(V,E), the system uses the Dijkstra algorithm to Generate from the entry area To the exit area The optimal path .

[0037] First, the system performs a Initialize path weights , where the starting node The weight of the node is set to 0, and the weights of all other nodes are initialized to infinity. At the same time, a priority queue Queue is initialized to dynamically track the nodes to be processed. The initial state only contains the starting node The system also maintains a path dictionary , record the current optimal path to each node, initially only record the path to the starting node. The algorithm extracts the node with the smallest current path weight from the priority queue , traverse all neighbor nodes connected to it .

[0038] For each edge , the system calculates from via arrive The cumulative path weight is: ,in is the path weight of the current node: ; It is the weighted value of the edge, which comprehensively considers the physical distance between regions, the density of human traffic, and the user's interest matching in the projects in the region.

[0039] If the calculated Less than the current record , then update the node The path weight of the path and update the path dictionary ,Will Add to the optimal path sequence to form a new path Then, the node Join the priority queue The algorithm repeats this process, continuously extracting the node with the smallest weight from the priority queue, and updating the path weight and path information of the adjacent nodes until all nodes are processed (that is, the priority queue is empty) or the exit node is empty. The path weights are no longer updated.

[0040] When the algorithm ends, the path dictionary records That is from arrive The optimal path. You can set multiple groups , generating a variety of preliminary paths that meet the conditions for each user, avoiding congestion caused by the same path initialization. The generated preliminary path contains a node sequence (areas that users pass through in sequence), a set of recommended items (users' high-interest items covered by the path), and provides a total weight value of the path to evaluate the optimization effect of the path. In this way, the system can effectively integrate user interests and real-time dynamics of the venue to provide users with efficient and personalized visiting plans.

[0041] Step 5: After generating the preliminary path, the system evaluates the candidate paths to select the optimal path that meets the user's needs and the venue's real-time conditions, and finally generates the user's recommended path This process includes multiple evaluation dimensions, including project coverage, time limits, and congestion avoidance.

[0042] First, the system calculates the coverage ratio of each candidate path to measure the degree to which the path satisfies the user's interest needs. The ratio of the set of items covered in P' to the set of all items that the user is interested in P is as follows: ; in, Is the user's project The interest matching score reflects the user's preference for the item. The system uses this indicator to prioritize the paths that cover the highest proportion of the user's high-interest items to ensure that the path recommendations can meet the user's interest needs as much as possible.

[0043] Next, the system evaluates the time constraints of the path and calculates the total time of the path. , ensuring that the total duration of the path does not exceed the user's planned visit duration The total path time is the average experience time of all items on the path. and the travel time of the path edges Add up and you get: ; in, Is the user in the project The average length of stay, Is the user from the region Move to area The system will exclude routes whose total duration exceeds the user's planned visit time and only retain candidate routes that meet the time limit.

[0044] Finally, the system combines the predicted future density of the region The system detects congestion on the path and removes the path that may pass through the high-density area in the future. Specifically, the system analyzes each node in the path. Density value at prediction time step t+k , if the density value exceeds the area capacity ,Right now , the node is marked as a congested area and the candidate paths containing the node are eliminated. In this way, the system can effectively avoid congested areas and improve the user's visiting efficiency and experience.

[0045] Based on the above three evaluation conditions, the system sorts the candidate paths and gives priority to the path with the highest coverage, meeting the time limit and avoiding congested areas as the final recommended path. The final output path not only meets the user's interest needs, but also fully considers the dynamic flow of people in the venue, providing users with an efficient and personalized visiting plan.

[0046] Step 6: During the visit, the system continuously collects the real-time number of people entering and leaving each area, thereby calculating the real-time crowd density , and use the density data to dynamically adjust the weights of the path edges in the weighted graph G=(V,E) The initial calculation of edge weights takes into account distance, user interest matching, and the density of the current area. As density data is updated, the system adjusts the weights of path edges based on real-time density values. Exceeding the area capacity To avoid excessive congestion, the system will increase the weight of the relevant edges. , to reduce the priority of the path. Specifically, the weight increases in proportion to the degree to which the regional density exceeds the capacity, that is: ; in Is the coefficient for controlling the increase in congestion. On the contrary, when the real-time density of the area is lower than the preset diversion threshold DensityThreshold, the system will reduce the weight of the relevant edges to encourage users to flow to these low-density areas. The reduction in weight is proportional to the degree to which the regional density is lower than the threshold, that is: ; Where β is the coefficient that controls the diversion reduction. In order to ensure the rationality of the path and avoid over-adjustment, the weight of the edge will be subject to upper and lower limits. and Through this dynamic adjustment mechanism, the system can guide users to avoid high-density areas and encourage them to flow to low-density areas, thereby effectively diverting the crowd and improving the venue's operating efficiency and user experience.

[0047] Step 7: When the system detects the user's current location Deviating from the original path or a certain area Real-time density Exceeding capacity When congestion occurs, the system adjusts the weights Use Dijkstra algorithm to re-plan the path. The new path starts from the user's current location Departure, the end point is the exit area , and cover the collection of high-interest projects that have not yet been visited Based on the above information and the updated graph G=(V,E), refer to the Dijkstra algorithm in step 4 and comprehensively consider the path distance , real-time crowd density and interest matching score Update the corresponding weight information to generate a new path . The new path generated by the system Ensure that high-interest projects are covered to the maximum extent, congested areas are avoided, and the remaining visit time limit of the user is met. At the same time, the system updates the user's visit order, prompts the route change and provides a new estimated arrival time and coverage project details to optimize the visit efficiency and ensure the user experience.

Claims

1. An intelligent optimization navigation method based on dynamic crowd flow and user interests, characterized in that: The following methods are included: Step 1: Deploy sensors at each main channel and project entrance. The system collects static area information of the venue and first initializes the static information; Step 2: The system collects the reservation information of each user, generates the interest profile of each user, makes the first crowd flow prediction, and uses the initial crowd flow and density information to build a weighted graph model; Step 3: The system calculates the real-time crowd density in each area based on the crowd flow data collected by sensors in each area in real time, records it in real time, generates a time series collection, and inputs it into the ARIMA model for future density prediction; Step 4: The system calculates the interest matching degree of each user for the project, combines the real-time crowd density and the distance between areas, and uses the Dijkstra path planning algorithm to generate the initial optimal recommended path from the entrance area to the exit area according to the weight; Step 5: After generating the preliminary optimal recommended path, the system further evaluates the path in real time, sorts the candidate paths according to the three conditions of coverage, time limit and congestion avoidance, and finally selects the optimal path as the user's recommended route; Step 6: During the navigation process, the system dynamically adjusts the weights of the path edges in the weighted graph according to the real-time crowd density, and adjusts the priority of the navigation path in real time; Step 7: During the navigation process, when the user deviates from the path or congestion occurs, the system replans the path based on the adjusted weights using the Dijkstra algorithm, regenerates the optimal path in real time, and avoids congested areas.

2. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 1 is characterized in that: The method comprises the following steps: in step 1: at the initial stage of system operation: loading the basic static information of the venue; the sensor is an image collection device and an infrared counter; Extract the structured information of the venue from the database, including the collection of all areas: , where each region Including specific location, area , Maximum capacity and topic tags ; Constructing region information as a weighted graph , where the node set Represents all regions, edge sets Represents the connection relationship between regions. The initial weight of the edge is the physical distance between regions. ; Then, the system loads all the experience items in the venue , each project Include its hashtag , average experience duration , Maximum number of people that can be received at one time , and the region where the project is located .

3. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 2 is characterized in that: In step 2: the system obtains the user's reservation information Extracting interest preferences , Estimated time of arrival and planned visit duration , generate a preliminary interest profile for each user .

4. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 3 is characterized in that: In step 3, the system collects graphic information of each venue through an image collection device and an infrared counter to generate a collection , transmitted to the system's data center, the system collects the real-time number of people entering and number of people leaving , combined with the area , and according to the formula: ； Calculate the current crowd density , updated every minute , and store it as a time series , and then predict the future density through the ARIMA model , and mark the predicted density exceeds the regional capacity congestion risk areas.

5. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 4 is characterized in that: The prediction formula of the ARIMA model is: ； The model parameter p represents the autoregressive order, d is the number of differences, and q is the moving average order. The model fits the parameters by minimizing the mean square error between the predicted error and the true value; To predict the density value of k time steps in the future, the model generates a prediction sequence in sequence: ； Each predicted value According to the historical sequence values ​​and the prediction error calculation, the prediction process is iterated step by step, using the prediction value of the previous step as the input of the next step; the shape of the output prediction sequence is (k, 1).

6. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 5 is characterized in that: After the prediction is completed, the system performs congestion detection on the density prediction results; If there is any parameter j greater than 0 and less than or equal to k in the forecast time series, such that , then the area Marked as a congestion risk area; this mark is directly input into the path planning module as the basis for dynamically adjusting the weight. The system optimizes the user path in real time and guides users to avoid high-density areas.

7. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 6 is characterized in that: In step 4: based on the user in the user set U Interest profile and the topic labels of the items in the item set P and weight factor , the system calculates each user's interest matching degree for the project ; Combined with real-time crowd density and distance between regions The system uses the Dijkstra path planning algorithm according to the weights Generate the optimal recommended path from the user's entry area to the exit area .

8. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 7 is characterized in that: In step 5: After generating the preliminary optimal recommended path, the system further evaluates the path and selects the path that meets the following conditions as the final recommended path : First, the system calculates the path based on the set of items covered by the path and the set of all items that the user is interested in. Project coverage in ,give priority to the path that covers the highest proportion of high-interest projects; Secondly, based on the average experience time of the project and the travel time of the path edges Calculate the total time of the path to ensure the total duration of the path Not exceed the user's planned visit duration ; Finally, combined with the prediction results , eliminating the future predicted density congested area paths; The system sorts candidate routes according to three conditions: coverage, time limit and congestion avoidance, and finally selects the best route as the user's recommended route.

9. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 8, characterized in that: In step 6: the system continuously collects data from various areas through sensors Real-time number of people entering and number of people leaving , calculate the real-time crowd density And update the density data to the weighted graph G=(V,E). According to the real-time density, the system dynamically adjusts the weight of the path edge. ; When the density of a certain area exceeds the capacity When , increase the weight of the relevant edge To reduce the recommendation priority; when the density of a certain area is lower than the threshold DensityThreshold, reduce the weight In order to attract more users, the system optimizes path planning by dynamically adjusting weights, guiding users to avoid high-density areas and divert them reasonably, thereby improving the visiting experience and venue operation efficiency.

10. The intelligent optimization navigation method based on dynamic human flow and user interests according to claim 9, characterized in that: In step 7: when the system detects the user's current location Deviating from the original path or a certain area Real-time density Exceeding capacity When congestion occurs, the system adjusts the weights Use Dijkstra's algorithm to re-plan the route. The new route starts from the user's current location. Departure, the end point is the exit area , and cover the collection of high-interest projects that have not yet been visited , weight calculation comprehensively considers the path distance , real-time crowd density and interest matching score ; New path generated by the system Ensure maximum coverage of high-interest projects, avoid congested areas, and meet the user's remaining visit time limit; At the same time, the system updates the user's visit order, prompts the route change and provides new estimated arrival time and coverage project details.

Images