Intelligent tourism destination tourist carrying capacity dynamic evaluation method and system

Abstract

The present application relates to the technical field of smart tourism and passenger flow management, and particularly relates to a smart tourism destination tourist carrying capacity dynamic evaluation method and system, the method comprising: collecting entrance ticket scanning data, device signal density data and video passenger flow statistical data to form multi-source passenger flow data through ticket scanning gate, wireless network probe and video passenger flow camera respectively; reconstructing a node real-time tourist density distribution map through a spatial interpolation algorithm; obtaining a node engineering carrying threshold parameter and performing upstream and downstream coupling conduction modulation on the node engineering carrying threshold parameter based on an upstream node congestion index to obtain a node dynamic carrying threshold; performing a three-hour passenger flow prediction based on historical passenger flow time series data and external passenger flow influence factors, comparing the node predicted passenger flow with the node dynamic carrying threshold to trigger a shunt early warning; and generating a low-density tour route suggestion based on the node real-time tourist density distribution map and the shunt early warning and pushing the low-density tour route suggestion to a tourist terminal.

Description

Technical Field

[0001] This invention relates to the field of smart tourism and passenger flow management technology, specifically to a method and system for dynamically assessing the carrying capacity of smart tourist destinations. Background Technology

[0002] Current tourist flow management in scenic areas generally relies on entrance ticket quotas, which can only control the total number of people entering the entire scenic area at any given moment. This approach fails to detect the real-time congestion levels at different functional areas within the scenic area, and is even less capable of handling sudden surges in visitor numbers caused by unforeseen weather or social media influencer photo opportunities. Furthermore, as a complex spatial system with multiple interconnected nodes, the flow of tourists between these nodes exhibits a clear upstream-downstream transmission characteristic. During periods of critical congestion, upstream congestion can compress the effective service time window of downstream nodes by extending path travel time, causing the tourist arrival time distribution to collapse from a uniform flow to a pulsed flow. This phenomenon is the fundamental dynamic mechanism triggering stampedes and other safety accidents, but current industry solutions generally overlook it.

[0003] Chinese patent application CN104899650B discloses a method for predicting tourist flow in scenic areas based on multi-source data analysis. This method employs an eight-step process: determining the starting point and time period, predicting tourist flow, first correction, second correction, daily correction, adjustment, generation, and system parameter correction. It integrates real-time data and OTA pre-sale data, incorporating factors such as weather, traffic, and online public opinion to predict tourist flow on a daily basis. However, this method only predicts the total tourist flow of the entire scenic area, failing to reconstruct the real-time tourist density distribution of each functional area within the scenic area. Furthermore, it does not address the dynamic mechanism of whether the node carrying capacity threshold changes dynamically with upstream congestion transmission. During peak holiday periods, its judgment on whether each node is overloaded still relies on static thresholds determined once during the engineering design phase. It cannot identify the compression of the downstream effective service time window caused by upstream congestion spillover. The warning timing for critical congestion periods is either too early, damaging management credibility, or too late, missing the optimal diversion window.

[0004] Chinese invention patent application CN114202103A discloses a machine learning-based method for predicting tourist flow in scenic areas during holidays. This method uses tourist flow as the target variable and integrates tourist flow attributes, online popularity attributes, scenic area reservation attributes, weather attributes, and time attributes as feature variables. It trains the prediction model using support vector regression and applies unexpected factors as correction factors to perform secondary corrections on the prediction results. However, this approach also focuses on the single scalar output of total scenic area tourist flow. Its correction mechanism is a statistical weighted adjustment of external interference factors, without involving physical transmission modeling of node-level dynamic carrying capacity thresholds. Therefore, it cannot reveal the causal relationship between changes in the distribution of tourist arrival times and the compression of node carrying capacity, and thus cannot capture the crucial early warning window of the shock wave collapse precursor during the critical congestion period.

[0005] Chinese invention patent application CN121544085A discloses a method for assessing and warning of recreational pressure in hotspot areas based on tourists' spatiotemporal trajectories. This method identifies dynamic hotspot areas by processing tourists' spatiotemporal trajectories and geographic information data, constructs a spatiotemporal graph network to model the pressure propagation relationship, and uses a pre-trained spatiotemporal graph prediction model to obtain the predicted pressure index and propagation path for future periods, thereby generating tiered warnings and personalized tour guide strategies. However, the core technical path of this scheme is a completely data-driven black-box modeling of spatiotemporal graph neural networks. The node carrying capacity threshold in this scheme is still implicitly a static reference quantity fixed during the model training phase, and the dynamic compression mechanism of the carrying capacity threshold during the critical congestion period is masked by the end-to-end learning of the neural network. Furthermore, this scheme lacks a self-checking and re-weighting mechanism for potential physical distortions that may occur in multi-source data acquisition units during critical periods, and the timing of warnings remains dependent on the reliability of the input data itself.

[0006] In summary, the existing scenic area visitor flow monitoring schemes, which trigger early warnings by comparing the instantaneous number of visitors with a one-time static carrying capacity threshold, fail to identify the key dynamic mechanism by which congestion at upstream nodes extends the path time to downstream nodes, thereby compressing the effective service time window of downstream nodes and causing the dynamic carrying capacity threshold of nodes to be far below the engineering design threshold. As a result, the timing of early warnings during the critical congestion period systematically deviates from the optimal diversion window. There is an urgent need to propose an intelligent scenic area carrying capacity assessment technology scheme with the upstream and downstream coupling and transmission modulation of the dynamic carrying capacity threshold of nodes as its core. Summary of the Invention

[0007] To address the bottleneck of existing scenic area visitor flow monitoring technologies, which rely on comparing instantaneous visitor numbers with static engineering thresholds to trigger early warnings, resulting in a systematic deviation of the critical congestion period warning timing from the optimal diversion window, this invention provides a method and system for dynamically assessing the carrying capacity of intelligent tourist destinations. This method involves dynamically weighting and fusing three heterogeneous data sources—entrance ticket scanning, park wireless network probes, and video visitor flow statistics—using spatial interpolation to reconstruct real-time visitor density distribution at the node level. It also involves applying upstream-downstream coupling and conduction modulation to the node engineering carrying capacity threshold parameters based on upstream node congestion indices to obtain the node dynamic carrying capacity threshold. Finally, it solves the global collaborative early warning sequence based on visitor flow continuity constraints by comparing the predicted visitor flow with the node dynamic carrying capacity threshold. Under the premise of fixed data acquisition hardware and scenic area topology, this method achieves dynamic and accurate assessment of scenic area carrying capacity and optimal diversion early warning at the level of visitor flow spatiotemporal distribution morphology and fluid continuity physical constraints.

[0008] The technical solution of this invention is: a dynamic assessment method for the carrying capacity of intelligent tourist destinations, comprising the following steps: S1, collecting entrance ticket scanning data, device signal density data, and video passenger flow statistics respectively through ticket scanning gates, park wireless network probes, and video passenger flow cameras deployed at the scenic area entrance, wherein the entrance ticket scanning data, device signal density data, and video passenger flow statistics constitute multi-source passenger flow data; S2, reconstructing a real-time passenger density distribution map of nodes in each functional area of ​​the scenic area using a spatial interpolation algorithm based on the multi-source passenger flow data; S3, obtaining node engineering carrying capacity threshold parameters, and adjusting the node engineering carrying capacity threshold based on upstream node congestion indicators. S4. Parameters are coupled and modulated upstream and downstream to obtain the node dynamic carrying capacity threshold; external passenger flow influencing factors are obtained, including holiday calendar factors, weather forecast factors, and social media popularity factors; based on historical passenger flow time series data and the external passenger flow influencing factors, passenger flow prediction for the next three hours is performed to output the node predicted passenger flow, and the node predicted passenger flow is compared with the node dynamic carrying capacity threshold. For nodes whose predicted passenger flow exceeds the node dynamic carrying capacity threshold, a diversion warning is triggered; S5. Based on the node real-time tourist density distribution map and the diversion warning, a low-density tour route suggestion is generated and pushed to the tourist terminal.

[0009] This invention also provides a dynamic assessment system for the carrying capacity of intelligent tourist destinations, including a multi-source data acquisition unit, a density distribution reconstruction unit, a dynamic carrying capacity threshold coupling unit, a prediction and early warning unit, and a diversion route push unit. The multi-source data acquisition unit includes a ticket scanning gate deployed at the scenic area entrance, a wireless network probe deployed within the scenic area, and video passenger flow cameras deployed in various functional areas of the scenic area. The ticket scanning gate is used to collect ticket scanning data at the entrance, the wireless network probe is used to collect device signal density data, and the video passenger flow cameras are used to collect video passenger flow statistics. The density distribution reconstruction unit is connected to the multi-source data acquisition unit and is used to reconstruct a real-time tourist density distribution map of each functional area of ​​the scenic area using a spatial interpolation algorithm. The dynamic carrying capacity threshold coupling unit... The unit is used to obtain the node engineering carrying capacity threshold parameter and perform upstream and downstream coupling conduction modulation on the node engineering carrying capacity threshold parameter based on the upstream node congestion index to output the node dynamic carrying capacity threshold; the prediction and early warning unit is connected to the dynamic carrying capacity threshold coupling unit and is used to perform passenger flow prediction for the next three hours based on historical passenger flow time series data and the external passenger flow influencing factors to output the node predicted passenger flow, compare the node predicted passenger flow with the node dynamic carrying capacity threshold, and trigger diversion early warning for nodes whose prediction exceeds the node dynamic carrying capacity threshold; the diversion route push unit is connected to the density distribution reconstruction unit and the prediction and early warning unit and is used to generate low-density tour route suggestions based on the node real-time tourist density distribution map and the diversion early warning, and push the low-density tour route suggestions to the tourist terminal.

[0010] The beneficial effects of this invention are as follows: First, this invention obtains the dynamic carrying capacity threshold of a node by performing upstream and downstream coupling and conduction modulation on the node engineering carrying capacity threshold parameter based on the upstream node congestion index. This transforms the triggering benchmark for diversion warnings from the static engineering threshold commonly used in the industry to a dynamic threshold that reflects the real-time transmission pressure of upstream congestion spillover. The mechanism is that upstream node congestion changes the time distribution of tourists arriving at the node by extending the path time, thus compressing the instantaneous service capacity of the node. Therefore, the true carrying capacity of the node is a dynamic variable modulated by the upstream state. Compared with the schemes of CN104899650B and CN114202103A, which only compare a single scalar passenger flow with a static reference value, this invention can identify critical congestion spillover transmission nodes and trigger warnings 5 ​​to 15 minutes in advance, thus avoiding the dilemma of "early warnings damaging credibility or late warnings missing the diversion window" under the static threshold paradigm. Second, this invention explicitly transforms the precursor of shock wave collapse into a quantifiable early warning signal by performing morphological entropy index calculation and mutation rate identification on the tourist arrival time distribution. The mechanism is that when the arrival time distribution collapses from a uniform flow to a pulsed flow, its Shannon entropy drops sharply from near the maximum entropy value, and the mutation rate exhibits a peak characteristic. This characteristic can be detected much earlier than the density threshold is exceeded. This mechanism is submerged in the end-to-end learning of the spatiotemporal graph neural network data-driven black box in CN121544085A, while this invention formalizes it into a white-box interpretable physical quantity monitoring index. Third, this invention identifies probe collapse intervals by discriminating the deviation direction of the change rate between device signal density data and video passenger flow statistics, and performs dynamic reweighting of three-source confidence scores. The mechanism is that during the critical congestion period, a large number of tourists actively access the scenic area's public wireless network, causing the MAC detection frames of passive probe detection to drop sharply due to randomization and degradation, forming data distortion that is the opposite of the actual congestion. Conventional fusion algorithms misjudge this distortion as a decrease in the number of people. This invention uses the probe collapse interval marker as a bypass warning signal for reverse utilization, forming a dual-channel collaboration with the diversion warning based on the node's dynamic carrying threshold, improving the overall warning accuracy by more than 18%. Fourth, this invention constructs a set of passenger flow continuity equations by analyzing the path connection topology between scenic area nodes and performs global early warning timing optimization using the dynamic carrying capacity threshold of the nodes as the upper bound constraint. The mechanism is that independent early warnings from multiple nodes have a mutual cancellation effect in the passenger flow field (upstream early warnings drive tourists to divert downstream, which in turn causes the downstream to reach the critical point earlier). After the passenger flow continuity equations are embedded in the early warning decision as a physical constraint of fluid conservation, the model is upgraded from a pure statistical black box to a physical constraint gray box. The generalization ability and multi-node collaborative response time are significantly better than the unconstrained statistical early warning scheme. This effect is a nonlinear collaborative gain composed of the above three innovations, and the overall optimal diversion response time is shortened by more than 20%. Attached Figure Description

[0011] Figure 1This is a schematic diagram of the overall process of the intelligent tourism destination dynamic assessment method of the present invention.

[0012] Figure 2 This is a schematic diagram of the overall architecture of the intelligent tourism destination visitor carrying capacity dynamic assessment system of the present invention. Detailed Implementation

[0013] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. These embodiments are merely illustrative of the invention and are not intended to limit the scope of protection of the invention.

[0014] This embodiment uses a 5A-level mountain scenic area with 24 functional area nodes as the application scenario. The scenic area receives approximately 30,000 visitors per day, with peak visitor flow exceeding 80,000 during holidays. The scenic area entrance has 4 main entrances and 2 auxiliary entrances. Within the park, a complex spatial network is formed by typical nodes such as cable car stations, viewing platforms, boardwalk intersections, and rest plazas. The overall process of the dynamic assessment method for the carrying capacity of intelligent tourism destinations described in this embodiment is as follows: Figure 1 As shown, it includes five main steps, S1 to S5.

[0015] Step S1: Multi-source visitor flow data collection. Ticket scanning gates are deployed at the six entrances / exits of the scenic area. Each entrance / exit is equipped with 3 to 5 parallel gates to handle peak throughput during holidays. These gates identify visitors' real-name reservation information based on QR codes and add a millisecond-level timestamp to the scanning moment, generating 1 to 8 entry ticket scanning data entries per second. Each data entry includes three fields: a unique visitor identifier, the scanning time, and the gate number. A total of 64 wireless network probes are deployed along the main tourist routes and at key node locations within the park. The deployment density is approximately one probe per 3000 square meters of node space. These probes operate in a dual-band passive listening mode (2.4GHz and 5GHz), sniffing and collecting detection frames sent by surrounding devices. Device signal density data is output every 5 seconds. Each data entry includes three fields: probe number, detection time, and the number of non-duplicate devices detectable at that time. Video visitor flow cameras are deployed at the central locations of 24 functional area nodes within the scenic area. These cameras are equipped with wide-angle lenses and edge computing modules, and locally execute deep learning-based pedestrian detection and deduplication algorithms. They output video visitor flow statistics every 10 seconds, with each data entry containing three fields: camera number, statistical time, and the number of unique visitors identified in the video footage within the statistical period. The entrance ticket scanning data, the device signal density data, and the video visitor flow statistics are transmitted back to the scenic area's monitoring center data aggregation server via the scenic area's 5G private network, forming multi-source visitor flow data. This multi-source visitor flow data is then aligned with a unified timestamp and stored in a time-series database.

[0016] Step S2: Reconstruct the real-time visitor density distribution map of the node. Before performing spatial interpolation, a three-source confidence dynamic weighting step is performed to address potential probe data distortion during critical congestion periods. For the scenic area's... Nodes ( , (Total number of nodes in the scenic area), calculate the signal density data of the devices mentioned above. With the video visitor statistics Rate of change between adjacent time windows:

[0017] ,

[0018] ,

[0019] Further calculation of the deviation direction discrimination factor:

[0020] ,

[0021] in: For nodes At any moment The rate of change of the device signal density data is a scalar with a value range of [value range missing]. The unit is persons / s, which is obtained by dividing the difference in device signal density data between two adjacent time windows by the sampling interval, and represents the instantaneous change trend and rate of the device signal density data. For nodes At any moment The rate of change of the video passenger flow statistics is a scalar, and its value range is [value range missing]. The unit is persons / s, which is obtained by dividing the difference in video passenger flow statistics between two adjacent time windows by the sampling interval, and represents the instantaneous change trend and rate of the video passenger flow statistics. For nodes At any moment The device signal density data is a scalar with a value range of [value range missing]. The unit is persons, and it is determined by the wireless network probe at the node. Collected from [location]; For nodes At any moment The video passenger flow statistics are scalars, and their values ​​range from [value range missing]. The unit is persons, and it is determined by the video passenger flow camera at the node. Collected from [location]; The sampling interval is denoted as , and is a scalar with a value of 10, in seconds, and is determined by the system configuration. For nodes At any moment The deviation direction discrimination quantity is a scalar with a value range of . The unit is The formula, calculated from this, indicates whether the direction of change in the device signal density data is consistent with the direction of change in the video passenger flow statistics. Indicates that the two change in the same direction. This indicates that the two change in opposite directions. Dimensional consistency verification: Left side of the equation. Units are ; right side of the equation Units are The dimensions on both sides are consistent.

[0022] Identifying probe collapse interval markers based on deviation direction discriminant:

[0023] ,in: For nodes At any moment The probe collapse interval marker is a scalar Boolean value with a value of Dimensionless, obtained by calculation using this formula. This indicates that the current time window is determined to be the probe collapse interval. The normal range indicates whether the signal density data of the device has reverse distortion due to the degradation of the MAC detection frame caused by the large-scale active access of tourists to the scenic area's public wireless network; The preset probe collapse judgment threshold is a scalar with a value range of [value range missing]. The unit is In this embodiment, Determined by on-site calibration. If the value is too large, the collapse judgment will be too lenient, leading to false positives; if the value is too small, the collapse judgment will be too strict, leading to false negatives. This is an indicator function that takes the value 1 when the condition within the square brackets is true and 0 when it is false.

[0024] Then, the multi-source passenger flow data is dynamically reweighted using three-source confidence levels:

[0025] ,

[0026] in: For nodes At any moment The three-source fusion weight vector is a three-dimensional vector, and the value range of each component is [value range missing]. Furthermore, the sum of the three components is always 1, dimensionless, and calculated by this formula, corresponding to the fusion weights of the entrance ticket scanning data, the device signal density data, and the video passenger flow statistics data, respectively. The baseline weight vector for the normal range is calibrated by offline cross-validation of historical data. The corrected weight vector for the probe collapse interval is used to reset the fusion weight of the device signal density data to 0 in order to eliminate distorted signals; The definition is the same as the formula mentioned above.

[0027] After further weighting, the estimated fusion density is obtained:

[0028] ,in: For nodes At any moment The estimated number of tourists after merging is a scalar, with a value range of [value missing]. The unit is persons, which is calculated by this formula and used as the input observation for subsequent spatial interpolation algorithms; For nodes At any moment The estimated passenger flow obtained by allocating the entrance ticket scanning data according to the average visitor movement path is a scalar quantity with a value range of [value missing]. The unit is persons; , The definition is the same as the formula mentioned above; , , The definition is the same as the aforementioned weight vector.

[0029] Dimensional verification: Left side of the equation The unit is persons; each term on the right side of the equation is the product of a dimensionless weight and a unit variable of persons, with the unit being persons. The sum of the three terms is still in persons, and the dimensions on both sides are consistent.

[0030] After obtaining the fused density estimates for each node, an inverse distance-weighted spatial interpolation algorithm is performed on the continuous space outside each functional area of ​​the scenic area to reconstruct the real-time visitor density distribution map of the nodes. For any location on the scenic area's plane... Its tourist density is estimated to be:

[0031] ,in: For the plane coordinates of the scenic area At the moment The estimated tourist density, in persons; To determine the number of neighboring nodes participating in the interpolation, this embodiment takes... ; For the target position to the number Euclidean distance between neighboring nodes, in meters; As the inverse distance weighting index, this embodiment takes... The spatial interpolation results of the fused density estimates of all nodes are visualized in the form of a heatmap to present the spatial differentiation of crowding levels.

[0032] Step S3: Node Dynamic Bearing Threshold Upstream and Downstream Coupling Conductive Modulation. Before performing upstream and downstream coupling conductive modulation, the arrival time distribution morphological entropy index and its mutation rate for each node are calculated based on the entrance ticket scanning data. For the scenic area... Each node extracts data from the time-series database showing tourists' travel history from its upstream node within the past 5 minutes. Path time sequence for reaching this node via scenic route Divide the path time interval evenly into A time bucket of equal width (in this embodiment, we take...) ), count the arrival frequency within each time bucket Then, the arrival frequency probability distribution is calculated:

[0033] The Shannon entropy is calculated on the arrival frequency probability distribution to obtain the morphological entropy index of the arrival time distribution.

[0034] ,in: For nodes At any moment The arrival time distribution morphotropy index is a scalar with a value range of [value range missing]. The unit is bits, calculated using this formula, and represents the time distribution of tourists arriving at this node. Close to the maximum value This indicates that the arrival times are approximately uniformly distributed. A value close to 0 indicates that the arrival times are highly concentrated and distributed in a pulse pattern; The number of time buckets is a scalar with a value of 20, which is dimensionless and is determined by the system configuration. For nodes At any moment No. The probability of arrival frequency within each time bucket is a scalar, and its value ranges from 1 to 2. , dimensionless, is obtained by dividing the arrival frequency within the time bucket by the sum of the arrival frequencies of all time buckets; For nodes At any moment No. The arrival frequency within each time bucket, in persons; It is a logarithmic function with base 2; when Time Agreement .

[0035] Dimensional verification: The unit on the left side of the equation is bits; the unit on the right side of the equation is bits. middle Dimensionless The unit is bits, the product unit is bits, and the summation unit is still bits, with consistent dimensions on both sides.

[0036] Further calculate the mutation rate of the arrival time distribution morphotropy index:

[0037] ,in: For nodes At any moment The mutation rate is a scalar, and its value range is [value range missing]. The dimensionless value is obtained by dividing the absolute value of the difference between the arrival time distribution morphological entropy index and the previous time window value using this formula. It characterizes the instantaneous change amplitude of the arrival time distribution morphological entropy index. The appearance of a spike is a precursor to shock wave collapse. The sliding time window interval is a scalar with a value of 300 and a unit of seconds, determined by the system configuration. To prevent extremely small positive numbers with a denominator of zero, the value is taken as... Dimensionless; For absolute value operators. Shock collapse precursor detection: when... Time-determined node At any moment The occurrence of shock wave collapse precursors, wherein the preset morphological entropy abrupt change threshold is mentioned. The range of values ​​is In this embodiment, The value is determined by the P95 quantile of the statistical distribution of morphological entropy mutation rate before known congestion events in historical data. Dimensional verification: Left side of the equation. Dimensionless; the numerator and denominator on the right side of the equation are in bits, the ratio is dimensionless, and the dimensions on both sides are consistent.

[0038] After the arrival time distribution morphotropic index and its mutation rate are calculated, upstream and downstream coupling conduction modulation of the node dynamic carrying threshold is performed. For the scenic area... Each node extracts a set of all its upstream nodes from the path connection topology between the nodes in the scenic area. Calculate each upstream node The upstream node congestion index:

[0039] ,in: For nodes At any moment The upstream node congestion index is a scalar with a value range of [value range missing]. Dimensionless, calculated by this formula, characterizing the upstream node. The degree of congestion relative to its engineering load-bearing threshold parameter; For nodes At any moment The real-time number of visitors, in persons; For nodes The node engineering bearing capacity threshold parameter is a scalar quantity, with the unit being "persons," and is comprehensively determined by the scenic area engineering design stage based on the node area, facility capacity, and ecological protection red line. Dimensional verification: The left side of the equation is dimensionless; the right side of the equation, persons / persons, is dimensionless, and the dimensions of the left and right sides are consistent.

[0040] A shock wave correction term is introduced to couple the abrupt change rate of the arrival time distribution morphotropic index into the upstream congestion spillover factor:

[0041] ,in: For nodes At any moment The shock wave correction term is a scalar with a range of values ​​of . , dimensionless, is calculated by this formula, and represents whether shock wave collapse precursor is detected. When not detected, the baseline value is 1; when detected, it is linearly amplified according to the proportion exceeding the threshold. This is the shock wave amplification factor, with a value range of [value range missing]. Dimensionless, in this embodiment, we take The bearing capacity of downstream nodes under shock wave collapse events is inverted and calibrated based on historical data. , The definition is the same as the formula mentioned above.

[0042] The upstream congestion spillover factor is calculated based on the upstream node congestion index and the shock wave correction term:

[0043] ,in: For nodes At any moment The upstream congestion spillover factor is a scalar with a value range of [value range missing]. Dimensionless, calculated by this formula, characterizing this node. It is subjected to the combined pressure of congestion spillover from all upstream nodes; For the set of upstream nodes The cardinality (i.e., the number of upstream nodes) is a positive integer and dimensionless; upstream node Up to this node The path propagation weight is a scalar with a value range of . And for each satisfy Dimensionless, representing the upstream node in the path connection topology between the scenic area nodes. Up to this node The distance is obtained by normalizing the reciprocal of the distance; , The definition is the same as the formula mentioned above; The summation range is the set of upstream nodes. All nodes in the equation. Dimensional verification: The left side of the equation is dimensionless; the product and sum of dimensionless variables on the right side of the equation, divided by the dimensionless base, is still dimensionless, indicating that the dimensions of the left and right sides are consistent.

[0044] Finally, the dynamic carrying threshold of the node is obtained:

[0045] ,in: For nodes At any moment The node dynamic carrying threshold is a scalar with a value range of [value range missing]. The unit is persons, which is calculated by this formula and serves as the dynamic comparison benchmark for triggering the diversion warning in the subsequent step S4. The definition is the same as the formula mentioned above; The definition is the same as the formula mentioned above; Let be the collapse-coordinated down-adjustment coefficient, and let be a scalar with a value range of . Dimensionless, in this embodiment, we take ; The definition is the same as the aforementioned formula, when the probe collapse interval marker and the diversion warning are triggered simultaneously within adjacent time windows of the same node. Otherwise, in the calculation of the dynamic carrying threshold of this node Dimensional verification: The unit on the left side of the equation is persons; the product of the unit variable persons and the two dimensionless factors on the right side of the equation is still persons, so the dimensions on both sides are consistent.

[0046] Through the aforementioned upstream and downstream coupling conduction modulation, the node The carrying capacity threshold is no longer a static value that is fixed once during the engineering design phase, but a dynamic variable that is modulated in real time by the congestion level of its upstream nodes, the shock wave collapse precursors of the arrival time distribution of upstream nodes, and the collapse state of the probe of this node. It can accurately reflect the phenomenon of compression of the service capacity of this node caused by upstream spillover during the critical congestion period.

[0047] Step S4: Prediction of Passenger Flow in the Next Three Hours and Multi-Node Collaborative Early Warning. Historical passenger flow time-series data for the past 30 days is extracted from the time-series database, and the external passenger flow influencing factors are obtained. The holiday calendar factor has a discrete value between 0 and 3, corresponding to non-holidays, short holidays, Golden Week, and extreme festival days, respectively; the weather forecast factor is obtained by weighted aggregation after normalization of temperature, precipitation probability, and wind force, with weights of 0.4, 0.4, and 0.2, respectively; the social media popularity factor is obtained by normalizing the logarithmic transformation of the number of posts and reposts of relevant social media topics related to the scenic area in the past 24 hours. In this embodiment, four indicators are used for weighted aggregation: the number of topic searches, the number of related image and text posts, the number of original video releases, and the number of reposts and comments on mainstream social media platforms.

[0048] The Long Short-Term Memory (LSTM) network is used as the backbone of the time series prediction model, and the input feature vector dimension is [missing information]. Output the predicted passenger flow for each node every 10 minutes for the next 3 hours. ,in The unit is min. The LSTM model uses a two-layer structure with a hidden layer dimension of 64. The training loss function is mean squared error, and the Adam optimizer is used. The initial learning rate is... During the training phase, historical visitor flow data from the past 12 months of the scenic area were used as training samples, with approximately 50,000 or more training samples per node. These samples were divided into training, validation, and test sets in a 7:2:1 ratio. A dropout rate of 0.2 and an early stopping strategy were employed to prevent overfitting. The trained LSTM model was deployed to the data aggregation server at the scenic area's monitoring center and continuously absorbed newly arriving visitor flow data through incremental learning to adapt to the long-term evolution of visitor flow patterns. In the online prediction phase, the most recent 24-hour visitor flow data was used as a sliding window for model input, triggering a prediction every 10 minutes to output the predicted visitor flow sequence for each node. During holiday transition periods, the system used the discrete levels of the holiday calendar factors as attention enhancement signals to improve the model's sensitivity to holiday effects.

[0049] The predicted passenger flow of the node is compared with the dynamic carrying capacity threshold of the node for each future time. With each node The system determines whether a diversion warning needs to be triggered. However, if each node triggers a diversion warning independently, the timing of warnings between nodes will cancel each other out. An early warning from an upstream node will drive tourists to divert downstream, causing downstream nodes to reach their critical point earlier. To eliminate this cancellation effect, this embodiment constructs a system of passenger flow continuity equations for the path connection topology between the scenic area nodes to perform multi-node collaborative warning optimization.

[0050] For each node of the scenic area The following continuity equation describes the dynamics of visitor accumulation:

[0051] ,in: For nodes At any moment The real-time number of visitors, in persons; For nodes The set of downstream nodes; For a moment From upstream node Inflow into this node The passenger flow rate is a scalar, and its value range is [value range missing]. The unit is persons / min, which is obtained by topological inversion of the entrance ticket scanning data along the path connection between the scenic area nodes; For a moment From this node Flowing to downstream nodes Passenger flow rate, unit and value range are the same .

[0052] Dimensional verification: Left side of the equation The unit is persons / min; the unit of the summation term on the right side of the equation is persons / min, and the dimensions of the left and right sides are consistent.

[0053] The passenger flow continuity equations for future times Accumulated passenger flow at nodes is used for point-based prediction:

[0054] ,in: For nodes In the future Passenger flow predicted by the passenger flow continuity equation system, in persons; For nodes The warning trigger delay variable is defined in minutes; the other variables are defined as before. In the numerical implementation, the above integral is discretized into a summation with a step size of 10 minutes.

[0055] Construct a global collaborative early warning timing optimization problem: the objective function is to minimize the total early warning trigger delay of all nodes, and the dynamic carrying capacity threshold of the nodes is used as the hard constraint upper bound for passenger flow accumulation at each node.

[0056] ,

[0057] ,

[0058] ,in: Let be a global vector composed of the early warning trigger delay variables of each node. A dimensional vector, where each component takes values ​​ranging from 1 to 2. The unit is min, and the output of the constraint optimization solution constitutes the global collaborative early warning time series. The objective function, expressed in min, is obtained by summing the warning trigger delay variables of all nodes. The total number of scenic area nodes is taken in this embodiment. Dimensionless; This is the upper bound for the warning trigger delay, with a value of 180 in minutes, and is determined by the system configuration. The definition is the same as the formula mentioned above; The definition is the same as the aforementioned integral prediction formula. This constrained optimization problem is solved in real time on the data aggregation server of the scenic area monitoring center using a sequential quadratic programming (SQP) solver, with a single solution taking no more than 2 seconds. Dimension verification: The objective function on the left side of the equation is in units of min; both sides of the constraint conditions are in units of persons, ensuring consistent dimensions.

[0059] The obtained global collaborative early warning timing sequence This is the globally optimal multi-node collaborative early warning triggering time series: for the... Each node, the diversion warning at time Triggered. When the probe collapses into a marked region. At that time, the collapse collaborative early warning channel independently outputs an early warning signal at the scenic area management terminal, which is presented in parallel with the diversion early warning based on the node's dynamic carrying threshold, forming a dual-channel collaborative early warning system.

[0060] Step S5: Low-density tour route generation and tourist terminal push. Based on the real-time tourist density distribution map of the nodes and the diversion warning, low-density tour route suggestions are generated. This embodiment employs a multi-objective weighted path search method. The objective function includes three weighted terms: the sum of the inverses of the densities of all nodes on the path (weight 0.5), the total path time (weight 0.3), and the matching degree of the tourist's visited node preferences (weight 0.2). Specifically, starting from the node where the tourist is currently located, and ending at the scenic area exit node or the target node actively specified by the tourist, a weighted A* path search algorithm is executed on the path connection topology between the scenic area nodes. The heuristic function of the A* search uses the product of the straight-line distance and the expected density of the target node as normalization. During the search process, the total cost of the above three weighted objectives is calculated cumulatively for each candidate path, and finally, the top 3 candidate paths with the smallest total cost are selected as alternative solutions for output. When the diversion warning is triggered at the node where the tourist is currently located, the path search will forcibly avoid that node; when the diversion warning is triggered at any node on the path from the tourist's current location to the scenic area exit, the path search will introduce an additional penalty for that node to include it in the avoidance range.

[0061] For each tourist using the scenic area's app or mini-program, a geofence is established based on their terminal location data. This geofence is a circular area with a radius of 200 meters centered on a warning node. The scenic area's monitoring center data aggregation server performs collision detection on the real-time coordinates of all registered tourist terminals every 5 seconds. When a tourist terminal enters the geofence within 200 meters of the warning node, the system pushes a low-density tour route suggestion via application messaging. The push includes a list of nodes along the suggested alternative route, a real-time density heatmap thumbnail of nodes along the current route, the estimated arrival time and total tour time for each candidate route, and the congestion level and avoidance suggestions for the current warning node. Tourists can choose to accept, ignore, or customize the alternative on their terminal. Accepting the alternative automatically activates voice navigation and pushes introductions of attractions along the route at each node. Ignoring or customizing the alternative records the tourist's choice and continuously monitors it. If the tourist actually enters the core area of ​​the warning node, the system will trigger a secondary push with a stronger warning level. All tourist terminal location data, selection behavior, and actual tour trajectory are transmitted back to the scenic area monitoring center's data aggregation server via the tourist terminal application. This data is used to continuously optimize the weight configuration and heuristic function strategy of the multi-objective weighted path search.

[0062] In this embodiment, the acquisition of device signal density data is based on the passive counting of device detection frames by the park's wireless network probes. During the acquisition phase, the wireless network probes perform irreversible hashing on the device hardware identifiers and retain only the aggregated access counts for each functional area. They do not store or restore any device identifiers that can identify a specific individual. The device signal density data participates in subsequent fusion as a regional population statistic and does not constitute the identification or tracking of a specific individual. The video visitor flow statistics are obtained by the video visitor flow camera at the terminal side, which only transmits the count results and does not transmit the original facial images. The acquisition of visitor terminal location data and the collision detection of the geofence are only performed on visitors who have actively installed the scenic area application or mini-program and authorized the processing of their location information in accordance with the "Personal Information Protection Law of the People's Republic of China" upon first launch. The scenic area monitoring center's data aggregation server follows the principle of minimum necessity and only collects the real-time coordinates necessary to implement the low-density tour route suggestion push. After completing a single geofence collision detection and route push, the visitor terminal location data is deleted or anonymized according to a preset retention period. Visitors can withdraw their authorization and stop the processing of their location information at any time on their visitor terminal. Through the collaborative design of anonymization, individual consent, minimum necessity, and time-limited deletion, this invention achieves dynamic assessment and traffic diversion guidance of carrying capacity while meeting compliance requirements for personal information protection.

[0063] This completes a full execution cycle of the intelligent tourism destination visitor carrying capacity dynamic assessment method. The system continuously executes the entire process from S1 to S5 in 10-minute cycles, forming a dynamic, refined, and closed-loop assessment and control of the scenic area's carrying capacity. On the scenic area management terminal, the real-time visitor density distribution map of the nodes is rendered in real time by overlaying a two-dimensional heat map with the scenic area map base map. The diversion warning and the collapse collaborative warning channels are indicated by different colors and icons flashing at the corresponding node locations on the map. Scenic area managers can click on any node on the terminal to view its historical carrying capacity threshold curve, upstream congestion spillover factor change curve, arrival time distribution morphological entropy mutation rate curve, and three-source confidence dynamic reweighted status record, forming a three-layer penetrating regulatory view of "macro situation - meso transmission - micro mechanism".

[0064] This embodiment provides a dynamic assessment system for the carrying capacity of intelligent tourist destinations, corresponding to the above method. The overall architecture of the system is as follows: Figure 2 As shown, it includes five core units: a multi-source data acquisition unit, a density distribution reconstruction unit, a dynamic load threshold coupling unit, a prediction and early warning unit, and a diversion route push unit.

[0065] The multi-source data acquisition unit includes three types of hardware devices: ticket scanning gates, wireless network probes, and video passenger flow cameras. The ticket scanning gates are deployed at the scenic area entrance; in this embodiment, 3 to 5 parallel gates are deployed at each of the six entrances. These gates identify tourists' real-name reservation information based on QR codes and add a millisecond-level timestamp to the scanning moment, used to collect entrance ticket scanning data. The wireless network probes, totaling 64, are deployed along the main tourist routes and at key node locations, operating in a dual-band passive monitoring mode (2.4GHz and 5GHz) to collect device signal density data. The video passenger flow cameras are deployed at the key nodes of the 24 functional areas within the scenic area, equipped with wide-angle lenses and edge computing modules to locally execute pedestrian detection and deduplication algorithms, used to collect video passenger flow statistics. The entrance ticket scanning data, the device signal density data, and the video passenger flow statistics are transmitted back to the scenic area's monitoring center data aggregation server via the scenic area's 5G private network to form multi-source passenger flow data.

[0066] The density distribution reconstruction unit is a software module running on the data aggregation server of the scenic area monitoring center, connected to the multi-source data acquisition unit. The density distribution reconstruction unit first performs dynamic reweighting of the multi-source visitor flow data using three-source confidence levels. It calculates the rate of change of the device signal density data and the rate of change of the video visitor flow statistics data, and uses their product as a deviation direction discrimination quantity. When the deviation direction discrimination quantity is less than a preset probe collapse judgment threshold and the rate of change of the device signal density data is negative while the rate of change of the video visitor flow statistics data is positive, it is determined to be a probe collapse interval and marked as such. For time windows with probe collapse interval markings, the fusion weight of the device signal density data is reset to 0, and the fusion weights of the entrance ticket scanning data and the video visitor flow statistics data are both increased to 0.5. Then, the reweighted multi-source visitor flow data is reconstructed using an inverse distance weighted spatial interpolation algorithm to reconstruct a real-time visitor density distribution map of each functional area of ​​the scenic area, and the spatial differentiation of congestion is presented in real-time as a heat map for the scenic area management terminal to view.

[0067] The dynamic carrying capacity threshold coupling unit is a software module running on the data aggregation server of the scenic area monitoring center. The dynamic carrying capacity threshold coupling unit first obtains the node engineering carrying capacity threshold parameters from the scenic area configuration database; then, based on the entrance ticket scanning data, it calculates the arrival time distribution morphological entropy index of each node, dividing the path time sequence from the upstream node to this node into 20 equal-width time buckets within a 5-minute sliding time window, statistically analyzing the arrival frequency distribution of each time bucket and calculating Shannon entropy; next, it calculates the mutation rate of the arrival time distribution morphological entropy index between adjacent sliding time windows. When the mutation rate exceeds a preset morphological entropy mutation threshold, it is determined to be a precursor to shock wave collapse, and the shock wave correction term of the upstream node is increased; then, for each node, it extracts the upstream node congestion index from all upstream nodes in the upstream node set, the ratio of the real-time number of tourists at the upstream node to the upstream node engineering carrying capacity threshold parameter, and combines this with the shock wave correction term for weighted aggregation to obtain the upstream congestion spillover factor; finally, it multiplies the node engineering carrying capacity threshold parameter by the difference between it and the upstream congestion spillover factor, and couples it with the collapse collaborative down-adjustment coefficient of the probe collapse interval marker to output the node dynamic carrying capacity threshold.

[0068] The prediction and early warning unit is a software module running on the data aggregation server of the scenic area monitoring center, connected to the dynamic carrying capacity threshold coupling unit. The prediction and early warning unit obtains external passenger flow influencing factors (including holiday calendar factors, weather forecast factors, and social media popularity factors) from an external data interface, extracts historical passenger flow time-series data for the past 30 days from the time-series database, and uses a two-layer LSTM time-series prediction model to output the predicted passenger flow of each node every 10 minutes for the next 3 hours. Then, it constructs a passenger flow continuity equation system for the path connection topology between the scenic area nodes, using the node dynamic carrying capacity threshold as the upper bound of the passenger flow continuity equation system. With the objective function of minimizing the total delay time of the global early warning sequence, it uses a sequential quadratic programming solver to perform constraint optimization and output the global collaborative early warning sequence. It triggers the diversion early warning for each node sequentially according to the global collaborative early warning sequence, and outputs the diversion early warning and the collapse collaborative early warning channel driven by the probe collapse interval marker in parallel to the scenic area management terminal.

[0069] The diversion route recommendation unit is a software module that works collaboratively with the scenic area monitoring center's data aggregation server and the tourist terminal application, connecting the density distribution reconstruction unit and the prediction and early warning unit. Based on the real-time tourist density distribution map of the nodes and the diversion early warning, the diversion route recommendation unit generates low-density tour route suggestions for each tourist holding the scenic area's APP or mini-program using a multi-objective weighted path search method. The objective function includes the weighted sum of the inverses of the densities of each node on the path (weight 0.5), the total path travel time (weight 0.3), and the tourist's preference matching degree for already visited nodes (weight 0.2). Furthermore, a geofence is established based on the tourist terminal's location data. When the tourist terminal enters the geofence within 200m of the early warning node, the low-density tour route suggestion is pushed via application message. The push content includes suggested alternative routes, a real-time density heatmap thumbnail of the nodes along the current route, and the estimated arrival time.

[0070] The five units achieve real-time data interaction and collaborative work through the scenic area's data bus. The scenic area's data bus is built on a dual-link redundant architecture of a 5G private network and wired fiber optic cables at the physical layer. The transmission layer uses a message middleware supporting millions of concurrent connections. This message middleware establishes independent publish / subscribe topics based on node numbers to achieve decoupled asynchronous communication between units. The scenic area monitoring center's data aggregation server adopts a dual-machine hot standby configuration, equipped with multi-core processors, massive memory, and high-speed solid-state storage. During peak holiday periods, it can stably handle the concurrent writing of thousands of multi-source passenger flow data points per second and the parallel computing requests of each unit's algorithm modules. The deployment density of the wireless network probes is approximately one probe per 3000 square meters, based on the node's spatial area. The deployment height is 3m to 4m above the ground to expand the effective detection range. The deployment height of the video passenger flow cameras is 4m to 6m above the ground, with a downward angle adjusted to 30 to 45 degrees to balance head detection accuracy and coverage area. The effective statistical radius of each video passenger flow camera is 30m to 50m. The ticketing and QR code scanning gates are equipped with low-light QR code recognition modules and facial recognition-assisted verification modules to address nighttime visits and photo misuse scenarios. The entire system operates in a 10-minute cycle, performing refined, adaptive, and closed-loop assessment and control of dynamic changes in the scenic area's carrying capacity. After deployment and field application verification, compared to traditional static threshold early warning schemes during peak holiday periods, the system provides an average of 8.6 minutes earlier warnings, improves warning accuracy by 18.3%, shortens the overall diversion response time by 22.7%, reduces the incidence of critical congestion events at nodes by 41%, and increases the average tourist satisfaction score from 3.6 to 4.4 out of 5.

[0071] The embodiments of the present invention are not limited to the specific embodiments described above. Those skilled in the art can make various equivalent changes or substitutions based on the technical solutions of the present invention, and all such changes or substitutions should be included within the protection scope of the present invention.

Claims

1. A method for dynamically assessing the carrying capacity of smart tourism destinations, characterized in that, Includes the following steps: S1. Ticket scanning data, device signal density data, and video passenger flow statistics are collected by ticket scanning gates, wireless network probes, and video passenger flow cameras deployed at the scenic area entrance, respectively. The ticket scanning data, device signal density data, and video passenger flow statistics constitute multi-source passenger flow data. S2. Reconstruct the real-time tourist density distribution map of each functional area of ​​the scenic spot using the spatial interpolation algorithm based on the multi-source tourist flow data; S3. Obtain the node engineering load capacity threshold parameter, and perform upstream and downstream coupling conduction modulation on the node engineering load capacity threshold parameter based on the upstream node congestion index to obtain the node dynamic load capacity threshold. S4. Obtain external passenger flow influencing factors, including holiday calendar factors, weather forecast factors, and social media popularity factors; perform passenger flow prediction for the next three hours based on historical passenger flow time series data and the external passenger flow influencing factors to output node predicted passenger flow, and compare the node predicted passenger flow with the node dynamic carrying capacity threshold, triggering diversion warning for nodes whose prediction exceeds the node dynamic carrying capacity threshold; S5. Based on the real-time tourist density distribution map of the node and the diversion warning, generate a low-density tour route suggestion and push the low-density tour route suggestion to the tourist terminal.

2. The method for dynamic assessment of tourist carrying capacity of smart tourism destinations according to claim 1, characterized in that, The upstream-downstream coupling conduction modulation of the node engineering carrying capacity threshold parameter based on the upstream node congestion index in step S3 specifically includes: Establish a path connection topology between scenic area nodes, extract all upstream node congestion indicators for each node, the upstream node congestion indicators include the ratio of the real-time number of tourists at the upstream node to the upstream node's engineering carrying capacity threshold parameter; calculate the upstream congestion spillover factor based on the upstream node congestion indicators, and multiply the node's engineering carrying capacity threshold parameter by the difference between the upstream congestion spillover factor and the node's dynamic carrying capacity threshold.

3. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 2, characterized in that, The upstream congestion spillover factor is calculated based on the mutation rate of the arrival time distribution morphological entropy index, which is obtained by: Based on the entrance ticket scanning data, the path time sequence of tourists from the upstream node to this node is obtained. The path time sequence is divided into equal-width time buckets within a 5-minute sliding time window, and the arrival frequency distribution within each time bucket is statistically analyzed. Shannon entropy is calculated on the arrival frequency distribution to obtain the arrival time distribution morphological entropy index. The absolute value of the difference between the arrival time distribution morphological entropy index and the arrival time distribution morphological entropy index of the previous time window is divided by the arrival time distribution morphological entropy index of the previous time window to obtain the mutation rate. When the mutation rate exceeds the preset morphological entropy mutation threshold, it is determined to be a precursor to shock wave collapse, and the upstream congestion spillover factor is increased.

4. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 3, characterized in that, Before reconstructing the real-time visitor density distribution map of each functional area of ​​the scenic spot using a spatial interpolation algorithm on the multi-source visitor flow data in step S2, a three-source confidence dynamic reweighting step is also included: The rate of change of the device signal density data and the rate of change of the video passenger flow statistics are calculated separately, and the product of the rate of change of the device signal density data and the rate of change of the video passenger flow statistics is used as the deviation direction discrimination value. When the deviation direction discrimination value is less than the preset probe collapse judgment threshold and the rate of change of the device signal density data is negative and the rate of change of the video passenger flow statistics is positive, it is determined that there is a probe collapse interval in the current time window and the probe collapse interval is marked. For the time window with the probe collapse interval mark, the fusion weight of the device signal density data is reset to 0, and the fusion weight of the entrance ticket scanning data and the video passenger flow statistics are both increased to 0.

5. The reweighted multi-source passenger flow data is used for the spatial interpolation algorithm.

5. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 4, characterized in that, Step S4 further includes triggering a collapse collaborative early warning channel based on the probe collapse interval marker, and the collapse collaborative early warning channel and the diversion early warning based on the node dynamic bearing threshold are output to the scenic area management terminal in parallel; When the probe collapse interval marker and the diversion warning are triggered simultaneously within adjacent time windows of the same node, the node dynamic load threshold will be further reduced by 15% to 25%.

6. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 5, characterized in that, The triggering of the diversion warning in step S4 is also based on the optimization of multi-node passenger flow continuity constraints: A system of passenger flow continuity equations is constructed for the path connection topology between the scenic area nodes. In the system of passenger flow continuity equations, each node satisfies the constraint that the difference between the inflow rate and the outflow rate is equal to the visitor accumulation rate of that node. The dynamic carrying capacity threshold of each node is used as the upper bound of the constraint of the system of passenger flow continuity equations. The constraint optimization solution is performed with the objective function of minimizing the total delay time of the global early warning sequence, and the global collaborative early warning sequence is output. The diversion early warning is triggered sequentially according to the global collaborative early warning sequence.

7. The method for dynamic assessment of tourist carrying capacity of smart tourism destinations according to claim 1, characterized in that, The generation of the low-density tour route suggestion in step S5 is based on multi-objective weighted path search. The objective function of the multi-objective weighted path search includes the sum of the inverses of the densities of each node on the path and the weighted sum of the total time taken on the path and the matching degree of the tourist's preferences for the nodes already visited, with weight ratios of 0.5, 0.3, and 0.2, respectively.

8. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 1, characterized in that, In step S4, the holiday calendar factor takes a discrete value between 0 and 3, corresponding to non-holidays, short holidays, Golden Week, and extreme festival days, respectively; the weather forecast factor is obtained by weighted aggregation of normalized temperature, precipitation probability, and wind force; the social media popularity factor is obtained based on the logarithmic transformation of the number of posts and reposts of social media topics related to the scenic spot in the past 24 hours.

9. The method for dynamically assessing the carrying capacity of smart tourism destinations according to claim 1, characterized in that, In step S5, the method for pushing the low-density tour route suggestion to the tourist terminal is as follows: a geofence is established based on the tourist terminal's location data, and when the tourist terminal enters the geofence within 200m of the warning node, the low-density tour route suggestion is pushed through an application message.

10. A dynamic assessment system for the carrying capacity of intelligent tourist destinations, used to implement the dynamic assessment method for the carrying capacity of intelligent tourist destinations as described in any one of claims 1-9, characterized in that, include: The multi-source data acquisition unit includes a ticket scanning gate deployed at the entrance of the scenic area, a wireless network probe deployed in the scenic area, and video passenger flow cameras deployed in various functional areas of the scenic area. The ticket scanning gate is used to collect entrance ticket scanning data, the wireless network probe is used to collect device signal density data, and the video passenger flow cameras are used to collect video passenger flow statistics. The entrance ticket scanning data, the device signal density data, and the video passenger flow statistics constitute multi-source passenger flow data. The density distribution reconstruction unit is connected to the multi-source data acquisition unit and is used to reconstruct the real-time tourist density distribution map of each functional area of ​​the scenic spot from the multi-source tourist flow data using a spatial interpolation algorithm. The dynamic load capacity threshold coupling unit is used to obtain the node engineering load capacity threshold parameter, and perform upstream and downstream coupling conduction modulation on the node engineering load capacity threshold parameter based on the upstream node congestion index to output the node dynamic load capacity threshold. The prediction and early warning unit, connected to the dynamic carrying capacity threshold coupling unit, is used to obtain external passenger flow influencing factors composed of holiday calendar factors, weather forecast factors, and social media popularity factors. Based on historical passenger flow time series data and the external passenger flow influencing factors, it performs passenger flow prediction for the next three hours to output the node predicted passenger flow. The node predicted passenger flow is compared with the node dynamic carrying capacity threshold, and a diversion early warning is triggered for nodes whose prediction exceeds the node dynamic carrying capacity threshold. The diversion route push unit connects the density distribution reconstruction unit and the prediction and early warning unit. It is used to generate low-density tour route suggestions based on the real-time tourist density distribution map of the node and the diversion early warning, and push the low-density tour route suggestions to the tourist terminal.

Images