Tourism project planning and visualization method and system based on digital twinning

Abstract

This invention relates to the field of intelligent tourism management technology, and discloses a tourism project planning and visualization method and system based on digital twins. The method includes: establishing a 3D twin map and extracting the bridge modal layer; acquiring sensing data to reconstruct cableway arrival flow, bridge deck density, and queue volume; extracting step frequency resonance parameters based on the above parameters; constructing candidate schemes, using the above parameters to build a comprehensive fitness level and updating the position; performing forward simulation mapping on the candidate schemes to generate a 3D situation map; decoding and outputting the execution sequence and display sequence according to the optimal scheme; and finally collecting measured deviations to update system parameters and retrieve archives. This method effectively avoids the hidden dangers caused by the superposition of passenger flow pulses and facility vibrations, and realizes dynamic optimization of tour routes and future situation prediction.

Description

Technical Field

[0001] This invention relates to the field of intelligent tourism management technology, and more specifically, to a method and system for tourism project planning and visualization based on digital twins. Background Technology

[0002] With the evolution of digital twin technology, the management and control of tourist attractions is gradually moving towards data synchronization and dynamic planning. In complex tour systems such as mountainous and canyon scenic areas that include passenger cableways, glass suspension bridges, and cliff walkways, route design involves not only path selection but also dynamic constraints from facility capacity and environmental conditions. Current digital twin and path planning algorithms for scenic areas typically simplify tour routes into conventional road networks with constant capacity and travel time. However, in this interconnected tour system, passenger cableways, limited by cabin transport and station distribution, inject intermittent passenger flow pulses into downstream sections; simultaneously, large-span transparent suspension bridges are prone to low-damped vibrations under pedestrian loads. When the mass passenger flow released by the cableway converges into the bottleneck at the bridgehead, density waves are generated, and the transparent, high-altitude visual stimulation can trigger fear, pauses, and non-uniform aggregation among tourists, which in turn alters the gait frequency of the group, causing significant modal responses in the bridge structure. Existing control systems often overlook the dynamic coupling effect caused by the interplay of cableway pulses, tour facility vibrations, and visual dwell time. They rely solely on current headcount statistics for post-event warnings, which leads to tour routes output by algorithms such as swarm intelligence optimization easily causing local congestion and decreased comfort during actual operation. Furthermore, the displayed images are significantly outdated and distorted compared to the actual situation. Summary of the Invention

[0003] This invention provides a method and system for tourism project planning and visualization based on digital twins, which solves the technical problems mentioned in the background art.

[0004] Firstly, a digital twin-based tourism project planning and visualization method is applied to mountain tourism systems that include passenger cableways, glass suspension bridges, and cliffside walkways, including: A three-dimensional twin map of the mountain tourism system was established, and a bridge modal layer was extracted to characterize the structural dynamics, so as to synchronize the response benchmarks of the physical structure and the virtual model. By acquiring passenger flow perception data and deconstructing the pulse interference caused by the intermittent release of the cableway, the cableway arrival flow, bridge deck density, and bridgehead queuing volume representing queuing pressure can be reconstructed. Based on the bridge modal layer, the bridge deck density, and the queuing volume at the bridgehead, structural vibration, visual stress, and passenger flow impulses are coupled to extract the step frequency resonance potential that reflects the risk of human-structure interaction. Construct candidate solutions, quantify the conflict between safety and experience using the step frequency resonance potential, build a comprehensive fitness, and then update the position of the candidate solutions. For the candidate schemes, a digital twin forward simulation is performed to map the physical evolution laws and generate a three-dimensional situation map representing the risk spatial distribution. Based on the candidate scheme that achieves the optimal overall fitness, the decoding outputs a planning execution sequence and a display sequence for guiding on-site management; Collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

[0005] Secondly, a digital twin-based tourism project planning and visualization system, executing any one of the digital twin-based tourism project planning and visualization methods, includes: The twin graph construction module is used to build a three-dimensional twin graph of the mountain tour system and extract the bridge modal layer to characterize the structural dynamic characteristics, so as to synchronize the response benchmarks of the physical structure and the virtual model. The passenger flow status reconstruction module is used to acquire passenger flow perception data, deconstruct the pulse interference generated by the intermittent release of the cableway, and thus reconstruct the cableway arrival flow, bridge deck density, and bridgehead queuing volume that characterizes queuing pressure. The resonance potential extraction module is used to extract the step frequency resonance potential that reflects the risk of human-structure interaction based on the bridge modal layer, the bridge deck density and the queuing amount at the bridgehead, coupled with structural vibration, visual stress and passenger flow pulse; The scheme iteration and update module is used to construct candidate schemes, quantify the conflict between safety and experience using the step frequency resonance potential, construct a comprehensive fitness, and then update the position of the candidate schemes. The forward simulation mapping module is used to perform digital twin forward simulation on the candidate scheme, and map the physical evolution law to generate a three-dimensional situation map representing the risk spatial distribution. The sequence decoding output module is used to decode and output a planning execution sequence and a display sequence for guiding on-site management based on the candidate scheme that optimizes the overall fitness. The state closed-loop update module is used to collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

[0006] The beneficial effects of this invention include: deeply coupling passenger flow pulses, bridge modal vibrations, and visual dwelling effects to construct specialized gait frequency resonance parameters, enabling the optimization algorithm to proactively avoid local congestion and resonance risks caused by the superposition of intermittent passenger flow and high-altitude transparent facilities when retrieving tour routes. Furthermore, this invention transforms route planning into a three-dimensional situation map generated by forward simulation, achieving an intuitive preview of the risk evolution process, and updating system state parameters through a closed-loop update of measured deviations, effectively improving the management safety and display realism of the mountain tour system. Attached Figure Description

[0007] Figure 1 This is a flowchart of a tourism project planning and visualization method based on digital twins according to the present invention. Detailed Implementation

[0008] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0009] Example 1: As Figure 1 As shown, a digital twin-based tourism project planning and visualization method is applied to a mountain tourism system that includes passenger cableways, glass suspension bridges, and cliff walkways, including: A three-dimensional twin map of the mountain tourism system was established, and a bridge modal layer was extracted to characterize the structural dynamics, so as to synchronize the response benchmarks of the physical structure and the virtual model. By acquiring passenger flow perception data and deconstructing the pulse interference caused by the intermittent release of the cableway, the cableway arrival flow, bridge deck density, and bridgehead queuing volume representing queuing pressure can be reconstructed. Based on the bridge modal layer, the bridge deck density, and the queuing volume at the bridgehead, structural vibration, visual stress, and passenger flow impulses are coupled to extract the step frequency resonance potential that reflects the risk of human-structure interaction. Construct candidate solutions, quantify the conflict between safety and experience using the step frequency resonance potential, build a comprehensive fitness, and then update the position of the candidate solutions. For the candidate schemes, a digital twin forward simulation is performed to map the physical evolution laws and generate a three-dimensional situation map representing the risk spatial distribution. Based on the candidate scheme that achieves the optimal overall fitness, the decoding outputs a planning execution sequence and a display sequence for guiding on-site management; Collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

[0010] A 3D twin graph is constructed, comprising visitor nodes, edges, a topology matrix, and a time-varying state set. Visitor nodes refer to the various functional nodes in the mountain tour system, including the upper and lower cable car stations, the ends of the glass suspension bridge, the entrances and exits of the cliffside walkway, and viewing platforms. Edges refer to the paths connecting the various visitor nodes, including cable car lines, bridge deck access areas, and walkway access areas. The topology matrix uses an adjacency matrix to represent the connections between visitor nodes. A value of 1 indicates a one-way direct path between the two nodes, a value of 2 indicates a two-way direct path, and a value of 0 indicates no direct path. Nodes are numbered sequentially from entrance to exit according to the tour order. Edge weights are initially the path length and are subsequently updated to reflect dynamic time consumption. The time-varying state set uses a dictionary data structure, with the key being the parameter name and the value being the parameter value and corresponding timestamp. It stores the real-time state parameters of each visitor node and edge, including visitor density, traffic speed, and structural vibration parameters. All data sources for the 3D twin map adopt the WGS84 coordinate system. The coordinate transformation parameters are obtained through on-site control point measurements, and the transformation error is less than 0.1 meters.

[0011] The preprocessed bridge deck acceleration signal is subjected to frequency domain transformation to obtain the acceleration spectrum, thereby eliminating broadband background noise. Time window. A Hanning window was used, with a window length of 1024 sampling points and an overlap rate of 50% between adjacent time windows. Fourier transform operator. This is used to perform a discrete Fourier transform on the acceleration signal within a time window to obtain the corresponding frequency domain signal. The formula for calculating the acceleration spectrum is: ; in The frequency is represented by the acceleration spectrum, which is the square of the magnitude of the frequency domain signal and is used to characterize the distribution of vibration energy at different frequencies. Wideband background noise mainly originates from environmental wind vibration, sensor electronic noise, etc., and can be distinguished from the narrowband peak signal generated by structural resonance through frequency domain transformation.

[0012] Locate the structural resonance point and extract its peak frequency. The structural resonance point corresponds to a local maximum point in the acceleration spectrum. The peak frequency threshold is set to be greater than 3 times the standard deviation of the background noise, and the minimum frequency interval between adjacent peaks is 0.2 Hz. Peaks with frequencies that are integer multiples of the fundamental frequency and amplitudes less than 1 / 3 of the fundamental frequency are identified as higher-order harmonic interference and are excluded. Peak frequency For the first The resonance frequencies corresponding to the first mode are obtained by searching the acceleration spectrum for the resonant frequencies that satisfy the condition. The frequency values ​​are obtained. For glass suspension bridges, the peak frequencies corresponding to the first three vertical vibration modes are usually extracted, and their frequency range is generally between 0.5Hz and 3Hz.

[0013] The damping attenuation rate was calculated using the ratio of attenuation peaks. Attenuation peaks were obtained through free vibration tests on the bridge deck. During the test, a drop-load excitation method was used at the mid-span to induce vertical vibration on the bridge deck. The drop weight was 50 kg, and the drop height was 0.5 m. Three excitations were performed, each 5 minutes apart. Vibration attenuation signals were collected for 30 seconds after each excitation. A set number of cycles was used for the interval. Two attenuation peaks and The first on the free vibration decay curve The first peak and the second The amplitude of each peak. The value is set to 3 to ensure sufficient difference in the amplitude of the attenuation peaks and improve the accuracy of the damping attenuation rate calculation. The formula for calculating the damping attenuation rate is: ; Damping attenuation rate is used to characterize the rate at which structural vibration energy decays.

[0014] Identify modal damping that reflects the energy dissipation characteristics of a bridge. Modal damping This is a dimensionless parameter used to characterize the energy dissipation capacity of a structure, and its calculation formula is as follows: ; For long-span glass suspension bridges, the modal damping is typically between 0.005 and 0.02. The Modal Confidence Criterion (MAC) is used to verify modal validity; a MAC value greater than 0.8 is considered a valid mode. The energy percentage of each mode is calculated, and mode extraction stops when the cumulative modal energy percentage exceeds 95%.

[0015] Peak frequencies and modal damping are integrated into a bridge modal layer and updated to a time-varying state set to characterize the structure's sensitivity to pedestrian loads. The bridge modal layer consists of structured data containing peak frequencies and modal damping for each modal order, with dimensions corresponding to the extracted modal orders. The same applies. The bridge modal layer is updated to the time-varying state set, with an update cycle of 1 hour, or updated immediately when the peak value of the bridge deck acceleration signal changes by more than 20% of the previous update value. The peak value statistics duration is 10 seconds. Signals greater than 3 times the historical peak value are considered abnormal peaks and are ignored to ensure that the virtual model and the physical structure's response benchmark remain synchronized.

[0016] By multiplying the number of passengers in each batch by the time weight and summing the results, the arrival data is smoothed and discretized to obtain the cableway arrival flow reflecting the upstream injection pulse. (Batch set) Includes all cabin trips made by the cable car within the specified time frame. Number of passengers per trip. For the first The actual passenger count for each cabin batch is obtained from the counting data at the cable car entrance gates. Time weighting. Using the Gaussian kernel function, its expression is: ,in This is the time bandwidth parameter, which is half the cable car departure interval, ranging from 20 to 60 seconds. Arrival time. For the first The arrival times of each passenger cabin batch at the upper cable car station. The formula for calculating cable car arrival flow is: ; The cableway arrival flow is the number of tourists arriving at the upper station per unit time, measured in people per second. When a batch of arrivals exceeds the current calculation time window, it is included in the calculation of the next time window. In the event of a cableway malfunction, the arrival flow is set to 0 until the cableway resumes operation.

[0017] Tourist locations are calculated using spatial weights, and bridge deck density, representing the intensity of space occupancy, is obtained by combining this with local width. (Tourist locations) Location data was obtained through the fusion of a video surveillance system and a Bluetooth positioning system deployed on the bridge deck. The video positioning weight was 0.6, and the Bluetooth positioning weight was 0.4. The positioning accuracy was no less than 1 meter, and the time synchronization error was less than 100ms. Abnormal location points were filtered as follows: points with speeds greater than 5 m / s or less than 0.1 m / s were considered abnormal and replaced with linear interpolation of adjacent points. Location data missing for no more than 3 consecutive frames were completed using linear interpolation; data missing for more than 3 frames was marked as lost. Spatial weighting. Using the Gaussian kernel function, its expression is: ,in This is the spatial bandwidth parameter, taken as twice the positioning accuracy, ranging from 1 meter to 3 meters. Local width. For the bridge deck at location The actual traffic width at the location is in meters. The formula for calculating bridge deck density is: ; Bridge deck density is the number of visitors per unit area, expressed as people per square meter. When visitor flow distribution exhibits multi-peak characteristics, the Epanechnikov kernel can be used instead of the Gaussian kernel, and its expression is: ,when If it is 0, then it is 0; otherwise, it is 0.

[0018] By combining visitor speed, spatial weights, and a minimum constant, the bridge deck speed, reflecting overall movement efficiency, is extracted. Visitor speed For the first The instantaneous velocity of each tourist is calculated by dividing the position difference between two consecutive frames of positioning data by the time interval, and the unit is meters per second. The origin of the bridge deck coordinate system is the upstream bridgehead, the x-axis points downstream along the longitudinal direction of the bridge deck, the y-axis points to the left along the transverse direction of the bridge deck, and the z-axis is perpendicular to the bridge deck and points upwards. The longitudinal velocity is obtained by projecting the tourist's velocity vector onto the x-axis. Projecting onto the y-axis yields the lateral velocity. The raw velocity is smoothed using a moving average filter with a window size of 5 frames and a step size of 1 frame. Anomaly velocity thresholds are defined as those greater than 3 times the average velocity or less than 0.1 times the average velocity; outliers are replaced by the average value within the window. Minimal constant. The value is taken as 1e-6 to avoid the denominator being zero. The formula for calculating the bridge deck speed is: ; Bridge surface speed is position The average speed of tourists at the location, expressed in meters per second.

[0019] Spatial integration of the bridge deck density within the queuing area yields the bridge approach queuing volume, characterizing the bottleneck accumulation scale. (Queuing area) The designated queuing area at the upstream end of the glass suspension bridge has an initial length of 50 meters and a width equal to the bridge deck width. When the queue length exceeds 80% of the current area, the queuing area extends backward by 20 meters, up to a maximum of 150 meters. When the queue length is less than 30% of the current area, the queuing area shortens forward by 20 meters. The queue length at the bridge end is calculated using the trapezoidal rule for numerical integration, with an integration step size of 1 meter and endpoint values ​​used for boundary points. The formula for calculating the queue length at the bridge end is: ; The queue length at the bridgehead is the total number of tourists in the queuing area, expressed in people.

[0020] Modal response kernels are constructed by combining peak frequency and modal damping. The first-order modal response kernel is used to characterize the amplification factor of the vibration response of a structure at different frequencies. Each mode is assigned a weight according to its energy proportion. The weights of the first mode are ,in For the first The energy of a modal response kernel is the integral value of the kernel over the step frequency range. The formula for calculating the modal response kernel is: ; The order of the extracted structural modes is usually taken as 3.

[0021] By integrating the step frequency energy spectrum with the frequency, incoherent walking interference is eliminated, yielding the step frequency resonance term reflecting the energy coupling strength. The step frequency signal is separated from the bridge acceleration mixture signal using Independent Component Analysis (ICA). The ICA algorithm used is FastICA, with 1000 iterations and a convergence threshold of 1e-6. The separated signal is then filtered through a narrowband filter from 1Hz to 3Hz to extract the step frequency signal. Step frequency energy spectrum. Frequency domain analysis of the step frequency signal revealed a time window length of 10 seconds, an overlap rate of 50%, and a frequency resolution of 0.1 Hz. A Hanning window was used for windowing. Frequency range. The frequency range is set to 1Hz to 3Hz, covering the normal walking frequency range for humans. When there are no pedestrians on the bridge, the gait frequency resonance term is set to 0. The gait frequency energy spectrum is normalized to the [0,1] interval, with the normalization benchmark being the historical maximum gait frequency energy spectrum value. The formula for calculating the gait frequency resonance term is: The step frequency resonance term is a dimensionless parameter used to characterize the energy coupling strength between the visitor walking load and the structural mode.

[0022] By combining bridge deck density, normalized visual depth, and lateral sway, the fear-susceptibility effect induced by transparent high-altitude environments is characterized, and the visual persistence term is calculated. Visual depth Location of the bridge deck Vertical height from the ground, in meters. Normalized apparent depth is... ,in The maximum height of the bridge deck, dimensionless. Absolute value of lateral velocity. This represents the absolute value of the velocity component of the tourist in the direction perpendicular to the bridge surface, expressed in meters per second. Velocity dispersion. For position The standard deviation of the lateral velocity of tourists is expressed in meters per second, and the calculation window is a rectangular window with a time of 5 seconds and a spatial dimension of 10 meters. The absolute value of the longitudinal velocity is also considered. This represents the absolute value of the velocity component of the tourist traveling in the direction parallel to the bridge surface, expressed in meters per second. The formula for calculating the visual dwell time term is: ;in The length of the bridge is denoted by . The visual dwell time term is a dimensionless parameter used to characterize the intensity of the fear-dwelling effect induced by a transparent high-altitude environment, and its value normally ranges from 0 to 3.

[0023] The positive part of the time derivative of bridge approach queuing volume is used to characterize the intensity of flow recirculation. The time derivative is calculated by differencing the queuing volume at the bridgehead at two consecutive moments, with a time interval of 1 second. Only the portion of the increased queue size is retained to characterize the intensity of the flow reversal, i.e., the speed at which tourists continuously flow into the queue area from upstream.

[0024] A queuing accumulation term reflecting the degree of pulse accumulation is extracted by combining bridge deck density and bridge deck speed. The queuing accumulation term is a dimensionless parameter used to characterize the degree of accumulation of cableway pulse passenger flow at the bottleneck at the bridgehead. Under normal circumstances, its value ranges from 0 to 5; a value greater than 5 is considered severe congestion. Its calculation formula is as follows: The denominator in the formula is the traffic capacity of the bridge deck per unit time, expressed in person / second.

[0025] By nonlinearly combining the gait frequency resonance term, visual dwell time term, and queue accumulation term, a gait frequency resonance potential characterizing the intensity of multiple physical, psychological, and behavioral disturbances is derived. This gait frequency resonance potential is a dimensionless parameter that comprehensively reflects the superimposed effects of structural vibration risk, tourist psychological stress risk, and passenger congestion risk. Its calculation formula is as follows: ; The nonlinear combination approach amplifies the overall risk level when multiple risks coexist. The risk thresholds for the step-frequency resonance potential are set as follows: a safety threshold of 0.5, a warning threshold of 1.0, and a danger threshold of 2.0. The corresponding control measures are normal passage, warning to slow down, and flow restriction.

[0026] A random sorting key is used to generate the visit order, and combined with dynamic time consumption, the planned arrival time reflecting the spatiotemporal trajectory is obtained. The random sorting key is a random number between 0 and 1, and each visit node corresponds to a random sorting key. The visit order of the visit nodes is determined according to the size of the random sorting key. (Node preorder edge set) For the first Among the candidate solutions, the one that reached the first The set of all edges that must be traversed before reaching a given node. Dynamic time consumption. To pass through the edge The required time is calculated using the following formula: ,in Let be the length of the side. Let be the average velocity of the edge. For a cableway, This refers to the cableway's operating speed. For bridge decks and walkways, ,in The free-flow velocity is set to 1.2 m / s. (Entry time offset) For the first Among the candidate solutions, the first one is... The offset of each node's entry time relative to the base time, in seconds. The formula for calculating the planned arrival time is: ; The planned arrival time is the In the candidate scheme, tourists arrive at the... The time of each node is in seconds.

[0027] Based on the integral of dynamic time consumption, step frequency resonance potential, congestion cost, and experience benefits, a multi-objective optimization function is constructed and optimization objectives are extracted. The multi-objective optimization function contains four optimization objectives, namely: total tour time... Total Risk Total congestion cost Total experience loss .in For the first The candidate routes are encoded using integers, with each integer corresponding to a visitor node's number. The encoding length equals the total number of visitor nodes. Each node can only appear once and must include key nodes such as the upper cable car station, the glass suspension bridge, and the lower cable car station. One-way edges can only be accessed in one direction. The total visit time is the total time required to complete the entire tour route, expressed in seconds, and is calculated using the following formula: ; The total risk is the sum of the time integrals of the resonant potentials of all edges along the tour route, and its calculation formula is as follows: ; The total congestion cost is the sum of the time integrals of the relative congestion levels along all sides of the tour route. Relative congestion level is the ratio of actual density to the maximum permissible density. The value is 2 people per square meter, and the calculation formula is as follows: ; The total experience loss is the sum of the experience losses at all nodes along the tour route. The experience loss is 10 minus the experience gains. For tourists at the node Arrival time The corresponding experience rating ranges from 0 to 10. The base score is determined by the node type: 8 for viewing platforms, 5 for bridgeheads, and 3 for cable car stations. The relationship between experience benefits and arrival time is as follows: ,in The best times to arrive are usually 10:00 AM and 3:00 PM. The attenuation coefficient has a value of [value missing]. The formula for calculating the total experience loss is: ; The optimization objective is normalized by applying data normalization to obtain a normalized objective. Linear normalization is used to map the value of each optimization objective to a value between 0 and 1. Normalized Objective The calculation formula is: ,in and These are the th in the current population. The minimum and maximum values ​​of the optimization objectives.

[0028] The randomness of manually assigned weights is eliminated by utilizing the information entropy of the normalized objective, and the weights of each optimization objective are adaptively allocated. (Sample proportion) For the first The first candidate solution The normalization objective value accounts for the th among all candidate solutions in the current population. The proportion of the sum of normalized target values ​​is calculated using the following formula: ; in Population size. Information entropy. Used to characterize the The dispersion of an optimization objective is determined by its information entropy. The smaller the information entropy, the greater the contribution of the optimization objective to distinguishing the merits of candidate solutions. The calculation formula is as follows: ; When population size When the normalized objective value is the same for all candidate solutions, the information entropy is set to 0. When the normalized objective value is the same for all candidate solutions, the information entropy is set to 1. Weights For the first The adaptive weights for each optimization objective, with the sum of all weights being 1, are calculated using the following formula: ; in To optimize the number of targets.

[0029] Calculate the overall fitness. For the first The sum of the weighted normalized objective values ​​of the candidate solutions; the smaller the overall fitness, the better the candidate solution. The calculation formula is as follows: ; Inertia weights are generated using the comprehensive fitness variance. Inertia weights Used to control the degree to which the speed of the previous iteration affects the speed of the current iteration. and The first Second and third The variance of the overall fitness of the population at each iteration. A larger variance in overall fitness results in a larger inertia weight, which is beneficial for global search. Conversely, a smaller variance in overall fitness results in a smaller inertia weight, which is beneficial for local search. The formula for calculating the inertia weight is: ; Initial inertia weights Set it to 0.9.

[0030] The learning factor is calculated by combining the positional difference. Individual extreme value location. For the first The candidate solution is in the... The optimal position obtained before the next iteration. Population extreme position. For the entire population in the first The optimal position obtained before the next iteration. Learning factor. and These are the individual learning factor and the social learning factor, respectively, and their sum is 1. A larger individual learning factor indicates that the candidate solution is more inclined to learn from its own historical best position. A larger social learning factor indicates that the candidate solution is more inclined to learn from the population's historical best position. The formula for calculating the learning factor is: ; Correct the search direction and update the positions of candidate solutions. (Random diagonal matrix) and The diagonal elements are random numbers between 0 and 1, and the off-diagonal elements are 0, used to increase the randomness of the search and avoid the algorithm getting trapped in local optima. Initial velocity vector The value range is [-1, 1]. Maximum speed. Set to 2, when the speed exceeds When clamped to The boundaries of the position vector define the range of encoded values; if the value exceeds the boundary, it is clamped to the boundary. Update evolution speed. For the first The candidate solution is in the... The velocity vector at the next iteration is calculated using the following formula: ; The position update formula is used to calculate the first position. The candidate solution is in the... The new position at the next iteration is calculated using the following formula: ; Population size of particle swarm optimization algorithm The fitness threshold is set to 50, the maximum number of iterations is 100, and the fitness convergence threshold is 1e-6. Iteration terminates if the fitness remains unchanged for 20 consecutive generations. Encoding validity checks are performed after each position update. Invalid encodings include duplicate nodes, missing key nodes, and reverse-directed edges. The repair methods are to randomly replace duplicate nodes with non-existent nodes, supplement missing key nodes, and adjust reverse-directed edges to be forward-directed. When multiple Pareto optimal solutions exist, the solution with the lowest total risk is selected as the final optimal solution.

[0031] By combining free-running speed and passenger flow suppression parameters, passenger flow is predicted to simulate the evolution of congestion density waves in the tour chain. (Unloaded free-running speed) Location of the bridge deck The free walking speed of tourists when there are no other tourists, measured in meters per second, typically between 1.2 and 1.5 meters per second. Density suppression parameter. This parameter characterizes the degree to which passenger flow density inhibits walking speed, with values ​​ranging from 0.1 to 0.5 square meters per person. Passenger flow diffusion parameter. Used to characterize the spatial diffusion speed of passenger flow, with values ​​ranging from 0.5 to 2 square meters per second. Source / inflow term. This refers to the number of new tourists per unit area per unit time, expressed in people per square meter. The passenger flow (in seconds) mainly originates from cable car arrivals and outflows from nodes. The partial differential equation for passenger flow conservation describes the spatial and temporal evolution of passenger flow, and its expression is: ; ; Numerical solution is performed using a first-order upwind scheme, with a time step of [missing information]. Set to 0.5 seconds, space step. Set to 1 meter. The Courrant condition is satisfied. ,in The velocity is set to 1.5 m / s to ensure numerical stability. The initial passenger flow density distribution and velocity distribution are the measured distributions. The inflow boundary is the arrival flow from the upper cableway station. ,Right now ,in This refers to the upstream bridgehead location. The outflow boundary is the free outflow condition at the downstream bridgehead, i.e. The maximum load capacity boundary is people per square meter, when the density exceeds At that time, the speed is set to 0. After a visitor stops at a node, the visitor is allocated to each edge according to the proportion of the travel capacity of the downstream edge, and the travel capacity is... When the density of a certain edge exceeds When that time comes, stop allocating passenger flow to that side.

[0032] By combining the mass, damping, and stiffness matrices with walking loads, the dynamic equations are solved to obtain predicted displacements, thereby assessing the vibration comfort of the structure. (Mass matrix) Damping matrix and stiffness matrix The finite element model parameters for the glass suspension bridge were obtained from structural design data and on-site modal tests. The damping matrix was constructed using Rayleigh damping. ,in and The Rayleigh damping coefficient is calculated using the damping ratio of the first two modes. , , and These are the angular frequencies of the first two modes. and The corresponding damping ratio is given. The vibration load from pedestrian traffic is a simple harmonic load, where... The average weight of the tourists is taken as 60kg. The value represents the walking acceleration amplitude, ranging from 0.2g to 0.3g. The walking frequency is 1.5Hz to 2.5Hz. The phase angle, with a value from 0 to... Random numbers between these ranges. Wind load. Wind speed data, measured in Newtons, is collected by sensors placed on the bridge deck. The expression for the dynamic equation is: ; via Newmark- Numerical solution is performed using the method, parameters , This is an unconditionally stable format with a time step of 0.01 seconds. Predicted displacement. The vertical vibration displacement of the bridge deck is expressed in meters.

[0033] A normalized mapping based on sample ranking is performed on predicted passenger flow, predicted displacement, and step frequency resonance potential to eliminate dimensional differences and generate a visualization layer. The empirical cumulative distribution function based on sample ranking is also used. Used to map parameters of different dimensions to the range 0 to 1. For a given parameter value , For samples less than or equal to The proportion of the sample size to the total sample size. The formula for calculating the data in the visualization layer is: ; The data values ​​for the visualization layer range from 0 to 3. The larger the value, the higher the risk at that location and at that moment.

[0034] A cumulative situation map is generated by performing spatiotemporal double integration on the visualization layers to characterize the degree of risk accumulation within a future time window, thus forming a three-dimensional situation map. Spatiotemporal double integration refers to integrating the visualization layer data spatially along the bridge deck length and temporally along the visitor time. The spatial resolution of the three-dimensional situation map is 1 meter, the temporal resolution is 1 second, and the future time window length is 30 minutes. The risk value and color correspondence are as follows: 0 to 0.5 is green, 0.5 to 1.0 is yellow, 1.0 to 2.0 is orange, and greater than 2.0 is red. The layer stacking order is: basic terrain layer, passenger flow density layer, structural displacement layer, and risk potential layer. The cumulative situation map data is the total cumulative risk value of the i-th candidate scheme within the future time window, and its calculation formula is: ; The candidate solution that minimizes the overall fitness is extracted as the optimal candidate solution to determine the risk balance point. The candidate solution with the lowest overall fitness in the current population has a tour route that achieves the optimal balance between total tour time, total risk, total crowding cost, and total experience loss.

[0035] The optimal candidate solutions are decoded to generate the optimal item sequence, and the node arrival time is predicted by combining this with the dwell distribution. The optimal item sequence is a sequence of visit nodes arranged in the encoding order of the optimal candidate solutions. Dwell distribution For tourists at the node The probability distribution of the length of stay is calculated using a normal distribution. The length of stay distribution is updated every 24 hours using an exponential moving average method, and the mean of the new length of stay distribution is... ,variance The relationship between residence distribution and real-time density is as follows: In other words, the higher the density, the longer the stay. The arrival time of subsequent nodes is when the tourist arrives at node number 1. The prediction time for each node, in seconds, is calculated using the following formula: ; Entrance gate release instructions are generated using the optimal project sequence and batch size to control the impact of cableway surges on downstream structures. The entrance gate release instruction data represents the number of visitors allowed to pass through the entrance gate per unit time, expressed in people per second. The minimum execution time is 1 second, and the maximum number of visitors allowed in each batch is the cableway cabin capacity, typically 8 people. Release instructions are synchronized with the cableway operation rhythm, releasing one batch at each cableway departure interval. By adjusting the entrance gate release time and the number of visitors allowed, intermittent passenger flow surges on the cableway can be transformed into relatively stable passenger flow, thereby reducing the impact on the downstream glass suspension bridge. The formula for calculating the release instructions is: ; in For batch quantity, For the first Release time for each batch.

[0036] The optimal project sequence, release instructions, and 3D situation map are encapsulated to generate a display sequence for driving the publishing end. This optimal project sequence, combined with the release instructions, serves as the planned execution sequence. The display sequence is multimedia data containing the optimal project sequence, release instructions, and 3D situation map, used to display at the scenic area entrance, visitor center, and mobile terminals, providing tour guidance for tourists. The planned execution sequence consists of control instructions containing the optimal project sequence and release instructions, used to drive the scenic area's turnstiles, cable cars, and broadcasting systems, achieving automatic control of the tour process.

[0037] The measured bias is calculated by subtracting the predicted parameters from the forward simulation output of the digital twin from the measured sensing data, thus quantifying the degree of mismatch between the simulation model and physical reality. The measured sensing data includes bridge deck density. Bridge speed Bridge deck acceleration Queues at the bridgehead Data were collected through a video surveillance system, a positioning system, an accelerometer, and a gate counting system, respectively. Prediction parameters include the corresponding simulation output values. , , and The timestamps of both measured and simulated data are unified to UTC time, with a synchronization error of less than 100ms. When data is missing, the deviation value from the previous moment is used as a replacement; if consecutive missing values ​​exceed 10 seconds, correction is paused. The measured deviation is represented as a column vector, with each element being the difference between the measured and predicted values. The calculation formula is as follows: ; The system state parameters and observation parameters are dimensionless, mapping all parameters to the interval [0,1]. System state vector. Including density suppression parameters Passenger flow diffusion parameters Modal damping After dimensionless transformation, all elements are dimensionless parameters. The Kalman gain is calculated by combining the error covariance and measurement noise to correct for the estimated drift of the system state parameters. Error covariance matrix. Used to characterize the uncertainty of the estimated system state parameters, the initial value is a diagonal matrix with 1e-3 diagonal elements. Observation mapping matrix Used to map system state parameters to the observation parameter space. Measurement noise matrix. The noise level of the measured sensing data is characterized by a diagonal matrix obtained from the sensor's technical parameters and historical test data. The Kalman gain matrix is ​​used to determine the weight of the measured deviation on the correction of the system state parameters; its calculation formula is as follows: ; The system state transition equation is ,in Let be an identity matrix, and assume that the state parameters change slowly over time. The process noise vector and covariance matrix are given. It is a diagonal matrix with diagonal elements of 1e-6. It recursively corrects the system state parameters and updates the error covariance matrix. (Identity matrix) This is the identity matrix with the same dimensions as the error covariance matrix. Through recursive correction, the estimation error of the system state parameters can be continuously reduced, improving the accuracy of the simulation model. The correction formula is as follows: ; ; The execution sequence of the current planning round and the measured deviation are stored in the algorithm retrieval archive. The algorithm retrieval archive is a database used to store historical planning execution sequences and corresponding measured deviation data. Each record contains characteristic information such as the code of the planning execution sequence, comprehensive fitness, measured deviation vector, execution timestamp, weather conditions, total passenger flow, and cable car departure interval.

[0038] By selecting samples with optimal overall fitness and measured deviation, highly reliable initial search seeds are extracted, thereby improving the search efficiency of the next round of planning. Samples with optimal overall fitness and measured deviation are those whose overall fitness is less than a set threshold and whose Mahalanobis distance of the measured deviation is less than a set threshold. Mahalanobis distance measures the deviation between the measured deviation and the measurement noise distribution; its calculation formula is: The similarity matching between the current operating conditions and historical operating conditions uses cosine similarity. The feature vector includes the current time, weather conditions, total passenger flow, and cable car departure interval. Samples with a similarity greater than 0.8 are considered similar. The initial search seed number is 10, and the encoding vectors of the 10 samples with the highest similarity are selected as the initial population for the next round of optimization algorithm.

[0039] The system's real-time performance metrics are: data acquisition cycle of 1 second, data processing cycle of 1 second, planned execution sequence generation cycle of 5 minutes, and generation cycle of 10 seconds in emergency situations. The total latency from data acquisition to the generation of control commands is less than 2 seconds. Modules interact via RESTful APIs, using JSON data format. Each interface defines the name, type, and value range of input and output parameters. Data validation employs parameter type checks and value range checks. Sensor fault detection uses threshold and rate-of-change methods; a fault is identified when sensor data exceeds the normal range or the rate of change exceeds the threshold. In case of a fault, the average value of historical data is used as a substitute, and an alarm is issued simultaneously. When more than 30% of sensors fail, the system degrades to manual control mode. In extreme conditions, entrance gates are closed to stop passage when passenger flow exceeds the maximum capacity. The glass suspension bridge and cableway are closed when wind speed exceeds 15 m / s, and tourists are guided to evacuate. Facility operation is immediately stopped upon malfunction, and emergency evacuation routes are generated. All random numbers are generated using the Mersenne Twister algorithm with a fixed seed of 12345 to ensure algorithm repeatability.

[0040] The mass proportional damping coefficient characterizes the damping component that is proportional to the structural mass, and mainly originates from the frictional damping of the structural supports and the internal damping of the materials. The stiffness proportional damping coefficient represents the damping component that is proportional to the structural stiffness, primarily originating from energy dissipation at structural connections and air damping. The value ranges for both are as follows: , The damping ratio was determined by the first two damping ratios through on-site modal testing. The diagonal elements characterize the degree of random fluctuation of the system state parameters, and the value of 1e-6 is determined based on the slow time-varying characteristics of the mountain tour system. The system state parameters, such as density suppression parameters and modal damping, change by no more than 1% per hour under normal operating conditions. Therefore, the variance of the process noise is set to 1e-6 to ensure the tracking capability of the filter while avoiding parameter oscillations caused by oversensitivity.

[0041] When matching current operating conditions with historical operating conditions, the weights of the feature vectors are time (0.3), weather (0.2), total passenger flow (0.3), and cable car departure interval (0.2). Time has the highest weight because tourist behavior has obvious daily and weekly periodicity; weather has the second highest weight because rainy days significantly prolong the time tourists spend in the city; total passenger flow and cable car departure interval directly determine the intensity of passenger flow pulses.

[0042] The base scores were obtained through standardized visitor surveys with a sample size of no less than 1,000, covering different age groups and genders. The survey used a 5-point Likert scale, linearly mapping visitor satisfaction ratings for different points to a range of 0-10. A base score of 8 for the viewing platform corresponds to "very satisfied," 5 for the bridgehead to "somewhat satisfied," and 3 for the cable car station to "somewhat dissatisfied." The base scores are updated quarterly; adjustments are made when the satisfaction level for a particular point changes by more than 1 point.

[0043] The stay distribution was obtained through statistical analysis of video surveillance data from 7 consecutive days, with the statistical period each day being 8:00-18:00. For each node, the entry and exit times of all tourists were extracted, the stay time was calculated, and outliers (less than 10 seconds, false positives) and stay times (more than 30 minutes, excessive rest) were removed. Then, a normal distribution was fitted to the valid data, and the mean and variance were calculated using maximum likelihood estimation. When the sample size was less than 100, the stay distribution of adjacent nodes was used as a substitute.

[0044] Simulation divergence is detected by a combination of three indicators: passenger flow density exceeding 3 people / square meter for more than 5 seconds; structural displacement exceeding 1.2 times the design allowable value for more than 3 seconds; and step frequency resonance potential exceeding 5 for more than 2 seconds. Upon detection of simulation divergence, the current simulation is immediately terminated, the valid state from the previous time step is loaded as the initial state, and the time step is reduced to half its original size before resuming the simulation. If simulation divergence occurs three consecutive times, a simplified simulation model is switched to, calculating only passenger flow evolution without structural dynamics simulation.

[0045] All system state parameters and observation parameters are independently linearized and dimensionless before entering the Kalman filter. For each parameter... The dimensionless parameters are ,in and These are the historical minimum and maximum values ​​of the parameter. After dimensionless transformation, all elements of the state vector and observation vector are dimensionless parameters in the interval [0,1], and all elements of the error covariance matrix have a dimension of 1.

[0046] Example 2: A tourism project planning and visualization system based on digital twins, comprising executing a tourism project planning and visualization method based on digital twins as described in any one of the examples, including: The twin graph construction module is used to build a three-dimensional twin graph of the mountain tour system and extract the bridge modal layer to characterize the structural dynamic characteristics, so as to synchronize the response benchmarks of the physical structure and the virtual model. The passenger flow status reconstruction module is used to acquire passenger flow perception data, deconstruct the pulse interference generated by the intermittent release of the cableway, and thus reconstruct the cableway arrival flow, bridge deck density, and bridgehead queuing volume that characterizes queuing pressure. The resonance potential extraction module is used to extract the step frequency resonance potential that reflects the risk of human-structure interaction based on the bridge modal layer, the bridge deck density and the queuing amount at the bridgehead, coupled with structural vibration, visual stress and passenger flow pulse; The scheme iteration and update module is used to construct candidate schemes, quantify the conflict between safety and experience using the step frequency resonance potential, construct a comprehensive fitness, and then update the position of the candidate schemes. The forward simulation mapping module is used to perform digital twin forward simulation on the candidate scheme, and map the physical evolution law to generate a three-dimensional situation map representing the risk spatial distribution. The sequence decoding output module is used to decode and output a planning execution sequence and a display sequence for guiding on-site management based on the candidate scheme that optimizes the overall fitness. The state closed-loop update module is used to collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

[0047] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A tourism project planning and visualization method based on digital twins, applied to a mountain tourism system including passenger cableways, glass suspension bridges, and cliff walkways, characterized in that... include: A three-dimensional twin map of the mountain tourism system was established, and a bridge modal layer was extracted to characterize the structural dynamics, so as to synchronize the response benchmarks of the physical structure and the virtual model. By acquiring passenger flow perception data and deconstructing the pulse interference caused by the intermittent release of the cableway, the cableway arrival flow, bridge deck density, and bridgehead queuing volume representing queuing pressure can be reconstructed. Based on the bridge modal layer, the bridge deck density, and the queuing volume at the bridgehead, structural vibration, visual stress, and passenger flow impulses are coupled to extract the step frequency resonance potential that reflects the risk of human-structure interaction. Construct candidate solutions, quantify the conflict between safety and experience using the step frequency resonance potential, build a comprehensive fitness, and then update the position of the candidate solutions. For the candidate schemes, a digital twin forward simulation is performed to map the physical evolution laws and generate a three-dimensional situation map representing the risk spatial distribution. Based on the candidate scheme that achieves the optimal overall fitness, the decoding outputs a planning execution sequence and a display sequence for guiding on-site management; Collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

2. The tourism project planning and visualization method based on digital twins according to claim 1, characterized in that, Establishing the three-dimensional twin map and extracting the bridge modal layer includes: Construct the three-dimensional twin graph containing visit nodes, edges, topology matrix, and time-varying state set; The acceleration spectrum of the bridge deck acceleration signal is obtained by frequency domain transformation in order to eliminate broadband background noise, locate the structural resonance point and extract the peak frequency. The damping attenuation rate is calculated using the attenuation peak ratio, thereby identifying the modal damping that reflects the energy dissipation characteristics of the bridge. The peak frequency and the modal damping are integrated into the bridge modal layer and updated to the time-varying state set to characterize the structure's sensitivity to pedestrian loads.

3. The tourism project planning and visualization method based on digital twins according to claim 1, characterized in that, Acquiring passenger flow perception data to reconstruct the cableway arrival flow, the bridge deck density, and the queue size at the bridgehead includes: Multiply the number of people in each batch by the time weight and sum them up to smooth and discretize the arrival data, thus obtaining the cableway arrival flow that reflects the upstream injection pulse; The location of tourists is calculated using spatial weights, and the bridge deck density, which characterizes the intensity of space occupancy, is obtained by combining the local width. By combining tourist speed, the aforementioned spatial weights, and a minimum constant, the bridge deck speed, which reflects the overall movement efficiency, is extracted. Spatial integration of the bridge deck density within the queuing area yields the bridgehead queuing volume, which characterizes the bottleneck accumulation scale.

4. The tourism project planning and visualization method based on digital twins according to claim 3, characterized in that, Extracting the step frequency resonant potential includes: By combining the peak frequency and the modal damping to construct a modal response kernel, and by integrating it with the frequency of the step frequency energy spectrum, incoherent walking interference is eliminated, and a step frequency resonance term reflecting the energy coupling strength is obtained. By combining the bridge deck density, the rate of change of visual depth, and the lateral sway, the fear-staying effect induced by the transparent high-altitude environment is characterized, and the visual stay term is calculated. The positive part of the time derivative of the bridgehead queuing volume is used to characterize the flow resurgence intensity, and the queuing accumulation term reflecting the pulse accumulation degree is extracted by combining the bridge deck density and the bridge deck velocity. By nonlinearly combining the step frequency resonance term, the visual dwell term, and the queuing accumulation term, the step frequency resonance potential characterizing the intensity of multiple physical-psychological-behavioral disturbances is obtained.

5. The tourism project planning and visualization method based on digital twins according to claim 4, characterized in that, Constructing the candidate solution and updating the position includes: The access order is generated using a random sorting key, and the planned arrival time reflecting the spatiotemporal trajectory is obtained by combining dynamic time consumption. Based on the dynamic time consumption, the integral of the step frequency resonance potential, the congestion cost, and the experience benefits, a multi-objective optimization function is constructed and the optimization objective is extracted. The optimization objective is normalized to obtain a normalized objective. The information entropy of the normalized objective is used to eliminate the randomness of the human weights. The weights of each optimization objective are adaptively allocated and the comprehensive fitness is calculated. The inertial weights are generated using the comprehensive fitness variance and the learning factor is calculated in combination with the position difference to correct the search direction and complete the position update of the candidate solution.

6. The tourism project planning and visualization method based on digital twins according to claim 5, characterized in that, Performing digital twin forward simulation of the candidate scheme and mapping it to generate the three-dimensional situation map includes: By combining free walking speed and passenger flow suppression parameters, passenger flow is extrapolated and predicted to simulate the evolution of congestion density waves in the tour chain. By combining the mass, damping, and stiffness matrices with walking loads, the dynamic equations are solved to obtain predicted displacements, thereby assessing the vibration comfort of the structure. The predicted passenger flow, the predicted displacement, and the step frequency resonance potential are normalized and mapped based on sample sorting to eliminate dimensional differences and generate a visualization layer. The visualization layer is subjected to spatiotemporal double integration to generate a cumulative situation map, which represents the degree of risk accumulation within a future time window, thereby forming the three-dimensional situation map.

7. The tourism project planning and visualization method based on digital twins according to claim 6, characterized in that, Based on the candidate solution that achieves the optimal overall fitness, the planning execution sequence and the display sequence are decoded and output, including: The candidate solution that minimizes the overall fitness is extracted as the optimal candidate solution to determine the risk balance point; The optimal candidate scheme is decoded to generate the optimal project sequence, and the arrival time of the node is predicted by combining the dwell distribution; The optimal project sequence and the number of people in each batch are used to generate an entrance gate release instruction in order to control the impact of cableway pulses on downstream structures; The optimal project sequence, the release command, and the 3D situation map are encapsulated to generate the display sequence used to drive the release end, and the optimal project sequence combined with the release command is used as the planning execution sequence.

8. A tourism project planning and visualization method based on digital twins according to claim 7, characterized in that, Collect the measured deviation, update the system status parameters and the algorithm-retrieved archives, including: The measured deviation is formed by subtracting the predicted parameters output by the digital twin forward simulation from the measured sensing data, in order to quantify the degree of mismatch between the simulation model and physical reality. The Kalman gain is calculated by combining the error covariance and measurement noise to correct the estimated drift of the system state parameters.

9. A tourism project planning and visualization method based on digital twins according to claim 8, characterized in that, Collecting the measured deviation, updating the system status parameters and the algorithm-retrieved archives, also includes: The execution order of the current planning process and the measured deviation are stored in the algorithm retrieval archive. By selecting the sample with the best overall fitness and the best measured deviation, an initial search seed with high reliability is extracted, thereby improving the search efficiency of the next planning round.

10. A tourism project planning and visualization system based on digital twins, executing the tourism project planning and visualization method based on digital twins as described in any one of claims 1-9, characterized in that, include: The twin graph construction module is used to build a three-dimensional twin graph of the mountain tour system and extract the bridge modal layer to characterize the structural dynamic characteristics, so as to synchronize the response benchmarks of the physical structure and the virtual model. The passenger flow status reconstruction module is used to acquire passenger flow perception data, deconstruct the pulse interference generated by the intermittent release of the cableway, and thus reconstruct the cableway arrival flow, bridge deck density, and bridgehead queuing volume that characterizes queuing pressure. The resonance potential extraction module is used to extract the step frequency resonance potential that reflects the risk of human-structure interaction based on the bridge modal layer, the bridge deck density and the queuing amount at the bridgehead, coupled with structural vibration, visual stress and passenger flow pulse; The scheme iteration and update module is used to construct candidate schemes, quantify the conflict between safety and experience using the step frequency resonance potential, construct a comprehensive fitness, and then update the position of the candidate schemes. The forward simulation mapping module is used to perform digital twin forward simulation on the candidate scheme, and map the physical evolution law to generate a three-dimensional situation map representing the risk spatial distribution. The sequence decoding output module is used to decode and output a planning execution sequence and a display sequence for guiding on-site management based on the candidate scheme that optimizes the overall fitness. The state closed-loop update module is used to collect measured deviations, correct the mismatch between the simulation model and actual observations, and update the system state parameters and algorithm retrieval archives.

Images