Simulation scene complexity-based airport high-speed rail transfer streamline optimization system and method

Abstract

The invention provides an airport high-speed rail transfer streamline optimization system and method based on simulation scene complexity. The system comprises a multi-source data import module, a pedestrian parameter identification module, a simulation scene establishment module, a travel feature analysis module, a station passenger flow simulation module, a transfer streamline optimization module and a system setting management module. The multi-source data importing module is used for importing six parts of data including station environment, monitoring videos, mobile phone signaling, sky train time, sky train passenger tickets and questionnaires; and the pedestrian parameter identification module is used for multi-target pedestrian detection and tracking and parameter output. And the simulation scene establishment module is used for scene index system construction, weight determination and membership calculation. And the travel feature analysis module is used for optimizing measure system construction and equipment selection preference and guide identifier preference analysis. And the station passenger flow simulation module is used for building a station simulation model, setting parameters and performing operation control. And the transfer streamline optimization module is used for crowded point location thermodynamic identification, streamline optimization scheme recommendation and improvement effect verification.

Description

Technical Field

[0001] The present invention relates to the technical field of transportation simulation, and particularly to an airport-high speed rail transfer streamline optimization system and method based on simulation scenario complexity. Background Art

[0002] As an important part of the modern integrated transportation system, air-rail intermodal transportation constructs a two-layer three-dimensional travel network of "air + high-speed rail" by organically integrating the advantages of fast long-distance air transportation and the dense medium- and short-distance coverage ability of the high-speed rail network. This multimodal transportation mode not only effectively solves the problem of the radiation radius limitation of a single transportation mode, but also realizes the seamless "door-to-door" connection of cross-regional passenger flow through the coordinated operation of hub airports and high-speed railway stations. Taking the Beijing-Shanghai route as an example, through the linkage between Shanghai Hongqiao Hub and the Beijing-Shanghai High-Speed Railway, passengers can complete the cross-city travel from Beijing to Nanjing within 2 hours, which shortens the travel time by 40% compared with pure air transportation, and reduces the comprehensive travel cost by about 30%.

[0003] The current development of traffic simulation technology shows a significant characteristic of being dominated by a single mode: road traffic simulation focuses on microscopic traffic flow interaction, urban rail transit modeling focuses on network operation efficiency, and the railway transportation system emphasizes dispatching optimization algorithms. This technological segmentation leads to significant shortcomings in the modeling of multimodal transportation scenarios, typically manifested as: the lack of spatio-temporal connection logic across transportation modes, insufficient capacity coupling of hub transfer facilities, and the absence of a dynamic coordination mechanism for timetables of different transportation systems. Especially for the unique "air segment + high-speed rail segment" combined itinerary in air-rail intermodal transportation, which involves cross-system coordination in multiple links such as the check-in system, baggage check, and security inspection process, it is difficult for traditional simulation methods to construct a digital twin of a complete service chain.

[0004] In terms of passenger behavior modeling, existing research generally adopts homogeneous assumptions and ignores the differential characteristics of the passenger group. There are significant differences between business passengers and tourist groups in terms of travel time sensitivity, transfer tolerance, payment preferences, etc. More notably, with the increasing digitization of the travel decision-making process, passenger behavior has formed a double-loop mode of "online pre-decision + offline experience feedback". This dynamic evolution characteristic requires that the simulation model must introduce a real-time behavior correction mechanism.

[0005] The big data era provides a new technical path to solve the above problems. By integrating multi-source heterogeneous data such as the departure data of airlines, the passenger flow heat map of 12306 of the railway, the transaction records of urban traffic cards, and the real-time road conditions of navigation APPs (the data volume can reach the PB level), a passenger portrait system containing 42 feature dimensions can be constructed. Based on the parameter self-optimization algorithm of deep learning, the dynamic calibration of transportation network parameters can be realized, improving the coincidence degree between the simulation results and the actual situation. After introducing machine learning into the passenger flow prediction model, the error range is smaller than that of traditional methods, providing a reliable decision-making basis for the allocation of hub transport capacity resources.

[0006] The essence of the current bottleneck in simulation technology lies in the triple disconnection of data - model - verification. The distributed modeling framework based on federated learning can effectively break data silos. By constructing the parameter mapping relationships of subsystems such as aviation, railway, and urban transportation, it realizes joint simulation across different transportation modes. The air - rail intermodal simulation platform jointly developed by Lufthansa and Deutsche Bahn in Germany has successfully verified the feasibility of this technology - by integrating the volatility characteristics of railway timetables into the aviation delay prediction model through transfer learning, the robustness of the intermodal transportation plan has been improved. The future development direction will focus on the deep integration of digital twin and meta - universe technologies to build an interactive virtual simulation environment to support transportation managers in conducting real - time deduction and effect evaluation of multiple scenarios.

[0007] As one of the key nodes of the air - rail intermodal hub, the airport high - speed railway station has the characteristics of short - term large - passenger - flow surges for arriving passengers transferring from air to high - speed rail and from high - speed rail to air. Especially for many high - speed railway stations of air - rail intermodal hubs in China, which are built underground, the space structure is relatively enclosed, and it is easy to have crowded passenger flows, resulting in unsmooth transfer path flow lines. In existing domestic air - rail intermodal hubs, the conversion rate of high - speed rail - connected passenger flows is relatively low, and the high - speed rail transport capacity advantage has not been fully utilized as expected in construction. Among them, the convenience of transfer walking affects the choice behavior of passengers' connection methods. In addition, as one of the forms of passengers' inter - modal transportation, the reliability of the transfer process affects the success rate of passengers' inter - modal transportation, especially when the transfer reserved time for the two - segment journey is very close to the limit, its importance will be more prominent. Currently, for the research on passenger transfer flow line simulation and optimization, few studies take the connection transfer process of air - rail intermodal as the research object, and there is a lack of authenticity verification for the passenger flow simulation environment and output results. Summary of the Invention

[0008] In view of the above problems in the prior art, the present application proposes an optimization method for the transfer flow line of an airport high - speed railway station based on the complexity of the simulation scenario, specifically including the following steps:

[0009] Step S1: Determine the air - rail intermodal comprehensive transportation hub and obtain multi - source anonymous data, where the multi - source anonymous data includes station environment data, surveillance video data, mobile phone signaling data, air - rail time data, air - rail ticket data, and questionnaire data;

[0010] Step S2: Use the target perception algorithm of the YOLO v5 and Deep SORT combined model to identify pedestrian movement parameters;

[0011] Step S3: Determine the portrait indicators of the simulation scenario library according to the mobile phone signaling data, and calculate the membership degree of the scenario complex characteristics through the entropy - weight cloud model;

[0012] Step S4: Relying on the questionnaire data, quantitatively analyze the helpfulness of facilities and guiding signs using a random forest model to determine the explanatory importance;

[0013] Step S5: Use the station environment data to construct an AnyLogic simulation model, including the physical building structures of the transfer passage, concourse level, and platform level;

[0014] Step S6: Design logic modules, and at the same time, according to the flight and high-speed rail timetables in the air-rail time data, set the moments of the simulated pedestrian flow and the arrival and dissipation of trains;

[0015] Step S7: Run the simulation program to verify the reliability of the simulation model and the accuracy of the results based on the air-rail ticket data;

[0016] Step S8: According to the bottleneck points of the flow line congestion shown in the heat map of the simulation results, appropriately match different importance optimization measures in combination with the membership degree of the complex scene characteristics;

[0017] Step S9: Run the simulation program again after optimization, specifically manifested as the smoothness of the flow line heat map and the complete passage time of pedestrians;

[0018] Step S10: Record and save the above process, and at the same time maintain and update the parameters, set access permissions, and realize the sustainable use of the system.

[0019] In one embodiment, the station environment data import unit is responsible for importing station environment data, the monitoring video data import unit is responsible for importing monitoring video data, the mobile phone signaling data import unit is responsible for importing mobile phone signaling data, the air-rail time data import unit is responsible for importing air-rail time data, the air-rail ticket data import unit is responsible for importing air-rail ticket data, and the questionnaire data import unit is responsible for importing questionnaire data.

[0020] In one embodiment, the basic materials of the station environment data are obtained by on-site research and taking photos, obtaining station structure diagrams, CAD drawings, etc.

[0021] In one embodiment, based on the starting and ending points of the air-rail intermodal transfer path at the airport high-speed rail station, determine the layout points along the line where the monitoring video data needs to be obtained. During the transfer from air to high-speed rail, take one side of the airport end of the air-rail intermodal transfer passage as the starting point, go through a series of processes such as high-speed rail security check and ticket checking and entering the station, and finally end with boarding the high-speed rail on the platform as the ending point. During the transfer from high-speed rail to air, take the arrival of passengers at the platform as the starting point, go through a series of processes such as ticket checking and leaving the station, and finally end on one side of the airport end of the air-rail intermodal transfer passage as the ending point. Select available video data according to conditions such as time range, picture clarity, and shooting angle without obvious occlusion.

[0022] In one embodiment, the mobile signaling data automatically demarcates the research area on the map in the system background and selects the API statistical interfaces for hourly granularity of people flow and people portrait.

[0023] In one embodiment, the airport flight schedules and railway train schedules are respectively crawled, and the ticket data provided by relevant enterprise departments of airports and railway administrations are cleaned, including deleting default values, outliers, and duplicate values.

[0024] In one embodiment, questionnaires are distributed to passengers in the waiting halls of airport high-speed railway stations, and the collection process is carried out synchronously offline and online. The question settings include the degree of help of the optimization measures to the passenger transfer guidance and the satisfaction with the convenience of the air-rail intermodal transfer process.

[0025] In one embodiment, since the walking characteristics of pedestrians in the air-rail intermodal hub are relatively complex, which is reflected in the uneven spatial distribution of walking speed and regional density, taking the along-line monitoring videos as the data source basis for parameter identification, the speed of pedestrians in the target area per unit time can be obtained.

[0026] In one embodiment, video data is loaded, and the multi-target pedestrian detection unit is responsible for using the YOLO v5 algorithm to detect pedestrians as targets, number each target and extract the pedestrian height information.

[0027] In one embodiment, the multi-target pedestrian tracking unit is responsible for predicting the position and state of the pedestrian in the next frame according to the pedestrian position and state using Kalman filtering, and performing cascade matching when the same target is detected in three consecutive frames of images.

[0028] In one embodiment, the multi-target parameter output unit is responsible for calculating the pedestrian walking speed according to the ratio of the actual walking distance of the pedestrian to the adjacent time interval, and thus obtaining the average walking speed of multi-target pedestrians.

[0029] The above technical features can be combined in various suitable ways or replaced by equivalent technical features as long as the purpose of the present invention can be achieved.

[0030] A method for optimizing the transfer streamline of airport high-speed railway stations based on the complexity of simulation scenarios provided by the present invention has at least the following beneficial effects compared with the prior art:

[0031] The present invention is based on six major anonymous data sources, namely, station environment data, surveillance video data, mobile phone signaling data, air-rail timetable data, air-rail ticket data, and questionnaire data. The pedestrian motion parameters are obtained through a video perception algorithm, and a passenger portrait is established in combination with various interfaces of mobile phone signaling data. The scene complexity is calculated using the entropy weight method cloud model membership function. The passenger selection behavior preferences are analyzed using the random forest model through the questionnaire data, so as to select appropriate optimization measures and adjustment means according to the scene characteristics. A passenger transfer simulation model of an airport high-speed railway station is established in combination with the station environment data, and pedestrian motion parameters are input to restore the passenger air-rail intermodal travel process. The arrival and departure times of airplanes and high-speed railways are set using air-rail timetable data, and the reliability of the simulation results is verified with the help of air-rail ticket data. Targeted optimization measures are taken according to the needs of the scene, so as to improve the convenience of air-rail intermodal transfer while controlling the transformation cost, thereby improving the overall service level of the station and the competitiveness of the comprehensive transportation hub. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein:

[0033] Figure 1 Shows a schematic diagram of the system framework of the present invention;

[0034] Figure 2 A logical flow chart of the working method is shown;

[0035] Figure 3 The schematic diagram of the indicator system for the station simulation scenario is shown;

[0036] Figure 4 A schematic diagram of the optimization action system is shown;

[0037] Figure 5 A schematic diagram of the simulation environment is shown;

[0038] Figure 6 The simulation logic module schematic is shown;

[0039] Figure 7 The output function graph of the simulation run is shown;

[0040] Figure 8 A heat map of the simulation run is shown;

[0041] Figure 9 The distribution function graph of passenger arrival time before optimization is shown;

[0042] Figure 10 The optimized distribution function graph of passenger arrival time is shown. DETAILED DESCRIPTION

[0043] The present invention will be further described below in conjunction with the accompanying drawings.

[0044] Refer to Figure 1 This is an implementation example of an optimized transfer flow system for airport and high-speed rail stations based on the complexity of simulation scenarios in the present invention; it includes: a multi-source data import module, a pedestrian parameter recognition module, a simulation scenario establishment module, a travel feature analysis module, a station passenger flow simulation module, and a transfer flow optimization module;

[0045] Refer to Figure 2 This is the process of an optimized transfer flow system for airport and high-speed rail stations based on the complexity of simulation scenarios in the present invention. The steps are as follows:

[0046] Step S1: Determine the integrated air-rail transportation hub, and obtain multi-source anonymous data. The multi-source anonymous data includes station environment data, surveillance video data, mobile phone signaling data, air-rail schedule data, air-rail ticket data, and questionnaire data. There are the following sub-steps:

[0047] S1.1: The station environment data import unit is responsible for importing station environment data, the surveillance video data import unit is responsible for importing surveillance video data, the mobile phone signaling data import unit is responsible for importing mobile phone signaling data, the air-rail schedule data import unit is responsible for importing air-rail schedule data, the air-rail ticket data import unit is responsible for importing air-rail ticket data, and the questionnaire data import unit is responsible for importing questionnaire data;

[0048] S1.2: Take the air-rail transportation of Wuxu Airport Hub in Nanning City, Guangxi Zhuang Autonomous Region as an example, and obtain basic materials of station environment data by means of on-site investigation and photographing, obtaining station structure diagrams, CAD drawings, etc.;

[0049] S1.3: Based on the starting and ending points of the air-rail transfer path of the airport and high-speed rail station, determine the layout points along the line where surveillance video data needs to be obtained. During the process of transferring from air to high-speed rail, take one side of the airport end of the air-rail transfer channel as the starting point, go through a series of processes such as high-speed rail security check and ticket checking to enter the station, and finally take boarding the high-speed rail platform as the ending point. During the process of transferring from high-speed rail to air, take the passenger getting off the train and arriving at the platform as the starting point, go through a series of processes such as ticket checking and leaving the station, and finally end at one side of the airport end of the air-rail transfer channel as the ending point. Select available video data according to conditions such as time range, picture clarity, and shooting angle with no obvious occlusion;

[0050] S1.4: The mobile phone signaling data is provided by the "Wutong Insights - Customer Group Insights" series of products of China Mobile. Manually delimit the research area in the map of the system background, and select the API statistical interface of personnel flow and personnel portrait with an hourly granularity;

[0051] S1.5: Crawl airport flight schedules and railway train schedules information through Ctrip and 12306 platforms respectively;

[0052] S1.6: Clean the ticket data provided by the relevant enterprise departments of the airport and railway bureau, including deleting default values, abnormal values, duplicate values, etc.;

[0053] S1.7: Questionnaires were distributed to passengers in the waiting hall of the airport high-speed rail station. The collection process was carried out simultaneously offline and online. The questions included the degree of help of the optimization measures in guiding passengers to transfer (scored using a five-level Likert scale) and satisfaction with the convenience of the air-rail intermodal transfer process. The specific questions were as follows:

[0054]

[0055]

[0056] Step S2: Use the YOLO v5 and Deep SORT combined model target perception algorithm to identify pedestrian motion parameters. There are the following steps:

[0057] S2.1: Since the movement characteristics of pedestrians in the air-rail transport hub are relatively complex, which is reflected in the uneven spatial distribution of walking speed and regional density, the surveillance video along the line is used as the basis for parameter identification data source, and the speed of pedestrians in the target area per unit time can be obtained:

[0058] S2.2: Load the video data. The multi-target pedestrian detection unit is responsible for detecting pedestrians using the YOLO v5 algorithm, numbering each target and extracting pedestrian height information.

[0059] S2.3: The multi-target pedestrian tracking unit is responsible for predicting the position and state of the pedestrian in the next frame based on the pedestrian's position and state using Kalman filtering, and performing cascade matching when the same target is detected in three consecutive frames of images;

[0060] S2.4: Use the center coordinates of the pedestrian target detection frame in two adjacent frames to calculate the pixel distance moved, and calculate the actual moving distance d1 according to the following formula:

[0061]

[0062] Where d1 is the actual distance between the center points of the pedestrian in the previous and next two frames; d2 is the pixel distance between the center points of the pedestrian in the previous and next two frames; H1 is the actual height of the pedestrian in reality; H2 is the pixel height of the pedestrian in the video.

[0063] S2.5: The multi-target parameter output unit is responsible for calculating the pedestrian walking speed based on the ratio of the pedestrian's actual walking distance and the adjacent time interval, thereby obtaining the average walking speed of the multi-target pedestrians.

[0064] Step S3: Determine the portrait metrics of the simulation scenario library based on the mobile signaling data, and calculate the membership degree of the scenario complexity features through the entropy weight cloud model. The following are the sub-steps:

[0065] S3.1: The scenario metric system construction unit is responsible for constructing the metric system. The mobile signaling data for constructing the simulation scenario library metrics is obtained separately from the personnel flow interface and the group feature interface;

[0066] S3.2: Among them, the natural attributes include gender, age, number belonging place, birth household registration place, permanent residence district / county, work district / county, residential district / county, education level, occupation, and marital status;

[0067] S3.3: Among them, the social attributes include communication fee level, preferred hotel star rating, travel mode preference, APP usage preference, mobile terminal model preference, source place, destination place, arrival mode, and departure mode;

[0068] S3.4: Refer to Figure 3 As shown, organize the metrics into a system, comprehensively consider the mutual influence of multiple factors, and calculate the complexity of different scenarios through the method of calculating the membership degree using the entropy weight cloud model;

[0069] S3.5: Determine the comment set cloud model. According to the normal cloud model, determine the quantitative domain and evaluation domain of the evaluation metrics, determine the cloud model characteristic parameters, and calculate the three cloud digital characteristics of the cloud model, namely the expectation E x , entropy E n and hyperentropy H e :

[0070]

[0071] H e = kE n

[0072] In the formula, X min is the minimum value in the same metric, X max is the maximum value in the same metric, and k is a coefficient, generally taking 0.05.

[0073] S3.6: The scenario metric weight determination unit is responsible for calculating the weights. Directly use the entropy weight method to calculate the weight matrix of each metric

[0074] S3.7: The scenario membership degree calculation unit is responsible for calculating the membership degree. Apply the forward cloud generator for precise numerical calculation, and use the membership degree calculation method between cloud models to solve the membership degrees corresponding to different levels of different data, generating the membership degree matrix Z = (z ij ) mn , and obtain the membership degree matrix Z corresponding to each evaluation scenario through multiple weighted average operations of the cloud generator.

[0075] S3.8: Utilize the obtained weight matrix and the membership degree matrix Z to calculate the fuzzy subset F=(f1, f2, …, f n ) of the evaluation index to be evaluated on the evaluation universe, and

[0076] S3.9: Based on the principle of maximum membership degree, select the maximum level of membership degree in F as the level to which the scenario belongs. Step S4: Relying on the questionnaire data, use the random forest model to quantitatively analyze the help degree of streamline optimization means such as facilities and equipment and guiding signs, and determine the explanatory importance. There are the following sub-steps:

[0077] S4.1: The random forest model establishment unit calculates the impurity of the decision tree node division according to the Gini impurity, so as to measure the feature importance of the influence of independent variables on the dependent variable. The optimization measures in questions 1-16 of the questionnaire are independent variables, and question 17 reflects the help effect on the entire transfer process with the satisfaction as the dependent variable. The optimization measure system construction unit classifies the optimization measures according to facilities and equipment, and guiding and signs. For specific reference, see Figure 4 ;

[0078] S4.2: Use Bootstrap sampling with replacement for the original data set D train to reselect N new sub-data sets D n (n = 1, 2, …, N), and train a decision tree based on each sub-data set D n ;

[0079] S4.3: Randomly select k dimensions of features (k < M) from all M-dimensional features included in the sub-data set D n , find the optimal decision tree division point, and divide the samples into 2 sub-nodes;

[0080] S4.4: Repeat steps S4.2 and S4.3 until all nodes cannot be further divided, and the decision tree T n of the sub-data set D n is constructed;

[0081] S4.5: Repeat S4.4, and finally train N decision trees to form a classifier, input the M-dimensional features of the test set data D test , and predict the classification label results respectively;

[0082] S4.6: Calculate the corresponding quantity of the predicted category of each decision tree in S4.5, and use the one with the highest number of votes as the final predicted category label;

[0083] S4.7: Each time for the variable X iNode splitting is performed, and the Gini impurity indices of the two descendant nodes will be less than that of the parent node. Therefore, using the Gini impurity identification helps to identify important optimization measures to improve the convenience and satisfaction of air-rail intermodal transportation for passengers;

[0084] S4.8: For a node splitting variable X with J categories i Its Gini impurity calculation formula is as follows:

[0085]

[0086] In the formula, G(X i ) is the Gini impurity index of the explanatory variable X i , and P(X i =j) is the probability that the sample of the estimated node X i belongs to category j;

[0087] S4.9: For the explanatory variable to be split, that is, the optimization measure, the importance analysis unit of the measure first calculates the decrease value of the Gini impurity index at the internal tree node, and then calculates the average decrease value of the Gini impurity in all decision trees to obtain the importance measure;

[0088] Step S5: Use the station environment data to build an AnyLogic simulation model, including the physical building structures of the transfer passage, concourse level, and platform level. The following are the sub-steps:

[0089] S5.1: The function of the station simulation model building unit is to establish a physical model of the station. The current Nanning Wuxu International Airport Station has two floors, namely the B1 concourse level and the B2 platform level, both of which are underground building structures. Refer to Figure 5 . Passengers enter the station through the north entrance and undergo security checks. On the east side of the station is Ticket Check 1, which can lead to Platforms 1 and 2, and on the west side of the station is Ticket Check 2, which can lead to Platforms 3 and 4. Passengers leave the station through the west exit after ticket checking. There are a total of 4 stairways and escalators connecting the two floors;

[0090] S5.2: Supplement other space marking modules, draw a service desk in the center of the concourse level, and simplify unnecessary building structures and service facilities, such as areas that do not affect the transfer process, such as convenience stores and restrooms;

[0091] S5.3: Set two tracks with a gauge of 1.5 meters, and represent the arrival and departure states during the operation of the high-speed rail by determining specific positions on the tracks, visually reflecting the generation and disappearance of the high-speed rail;

[0092] S5.4: The function of the simulation model parameter setting unit is to set the relevant simulation parameters. According to the national standards and relevant specifications, the conveying speed of the escalator is set to 0.5 m / s, and its corresponding throughput is 8,100 person-times per hour. The average time for security inspection is 13.5 seconds per person, the ticket selling service time of the self-service ticket vending machine is 45.7 seconds per person, the service time for inbound ticket checking is 2.9 seconds per person, the service time for outbound ticket checking is 3.2 seconds per person, the pedestrian size follows a random distribution of 0.3 - 0.5 m, the walking speed follows a random distribution of 0.8 - 1.6 m / s, and the length of the 8-car formation train is 210 m;

[0093] Step S6: Design the logic module, and at the same time, according to the flight and high-speed rail timetables in the sky-rail time data, set the moments of arrival and dissipation of the simulated pedestrian flow and trains. The following are the sub-steps:

[0094] S6.1: The function of the simulation model operation control unit is to design the logic module and be able to run normally. Refer to the Figure 6 schematic diagram of the simulation logic module, including three parts: passengers entering the station, passengers leaving the station, and the arrival and departure of trains;

[0095] S6.2: First, use the pedSource module in the pedestrian library to generate a crowd in the airport passage and then use the pedService module to simulate the queuing waiting for pedestrian security inspection after arriving at the high-speed rail station. This module is used to define a service with an electronic queue and select the shortest path;

[0096] S6.3: Then use the selectOutput module to select the ticket gate. According to the passenger flow situation, the probability of pedestrians choosing either of the two ticket gates is 0.5;

[0097] S6.4: In the structure of the four escalators and four staircases in the high-speed rail station, the pedEscalator module is selected for both. The running speed of the staircase is set to 0, and the escalator is preferentially selected with a probability of 90%;

[0098] S6.5: After passengers arrive at the platform layer, use the pedWait module to realize the waiting of pedestrians on the platform layer;

[0099] S6.6: When the high-speed rail arrives, use the freeAll() function to simulate the departure of pedestrians taking the high-speed rail;

[0100] S6.7: Similarly for outbound, use the pedService module to reflect the waiting for outbound ticket checking. Considering that the outbound gate is located far from the airport side of the transfer passage, add the pedGoto module to play a guiding role in path movement;

[0101] S6.8: For the arrival and departure of trains, the trainSource event represents the high-speed train entering the station, the trainMoveTo event represents the acceleration, constant speed, and deceleration states of the high-speed train running along the set track, the delay event represents the high-speed train staying at the platform, and the trainDispose event represents the high-speed train leaving the station;

[0102] S6.9: During the generation and disappearance of pedestrians, the timeMeasureStart and timeMeasureEnd modules are used to count the complete time of the pedestrian intermodal transfer between air and rail;

[0103] Step S7: Run the simulation program, and verify the reliability of the simulation model and the accuracy of the results according to the air-rail ticket data. There are the following sub-steps:

[0104] S7.1: Refer to Figure 7 the output function images, including the arrival time distribution, departure time distribution, change in the number of people on the concourse level, change in the number of people on the platform level, transfer channel density, density in the Nanning direction on the platform level, and density in the Chongzuo direction on the platform level;

[0105] S7.2: Combine the number of people and time at the entrance and exit turnstiles in the air-rail ticket data to verify the reliability of the output results of the simulation model. Using the air-rail ticket data as the true value and the simulation results as the predicted value, it is determined through the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE):

[0106]

[0107] In the formula, Y i is the true value; is the predicted value; η is the total number of samples.

[0108] Step S8: According to the flow line congestion bottleneck points shown in the heat map of the simulation results, combine the membership degree of the complex scene features to appropriately match different optimization measures with different importance levels. There are the following sub-steps:

[0109] S8.1: Refer to Figure 8 the heat map of the simulation run. According to the heat identification unit of the congestion points, it can be judged that the closer the color is to red, the greater the flow density in the passenger transfer area and the more congested the flow line;

[0110] S8.2: Recommend units according to the streamline optimization plan and match them with the results of scenario complexity and optimization measures. If the scenario complexity is I, select all optimization measures with the cumulative explanation importance reaching 25%. If the scenario complexity is II, select all optimization measures with the cumulative explanation importance reaching 50%. If the scenario complexity is III, select all optimization measures with the cumulative explanation importance reaching 75%. If the scenario complexity is IV, select all optimization measures with the cumulative explanation importance reaching 100%.

[0111] Step S9: Run the simulation program again after optimization, specifically manifested as the smoothness of the streamline heat map and the complete passing time of pedestrians. The following are the sub-steps:

[0112] S9.1: For crowded bottleneck points, optimize guiding signs, facility layouts, etc. through preference selection and verify and compare the improvement effects through simulation;

[0113] S9.2: According to the unit for verifying the improvement effect through simulation, refer to Figure 9 and Figure 10 , where the abscissa of the function image is time (unit: second) and the ordinate is the cumulative probability value. It can be seen the changes in the inbound time distribution before and after. The overall time for passengers to enter the station after optimization is significantly shortened.

[0114] Step S10: Record and save the above process, and at the same time maintain and update the parameters and set access permissions to achieve the sustainable use of this system. The following are the sub-steps:

[0115] S10.1: The hardware management unit saves the above usage records to the local storage hard disk, including multi-source anonymous data, simulation parameter settings, simulation operation results, etc.;

[0116] S10.2: This system is for the staff of aviation and railway transportation enterprises. The permission control unit sets two types of identities, administrators and visitors, according to the user type. The access permission of the administrator is all functions of the system, including adding, deleting, modifying user names and permission function settings. The access permission of the visitor is only for browsing and querying;

[0117] S10.3: The maintenance and update unit updates the parameter results after simulation optimization. At the same time, to ensure the sustainable use of the system, the system equipment is regularly maintained and upgraded.

[0118] While the present invention has been described herein with reference to particular embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that numerous modifications may be made to the exemplary embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein may be combined in ways different from those described in the original claims. It should also be understood that the features described in connection with separate embodiments may be used in other described embodiments.

Claims

1. A method for optimizing airport high-speed rail transfer streamlines based on simulation scene complexity, characterized in that: The following steps are involved: Step S1: determine an air-rail intermodal integrated transportation hub and obtain multi-source anonymous data, wherein the multi-source anonymous data includes station environment data, surveillance video data, mobile phone signaling data, air-rail timetable data, air-rail ticket data, and questionnaire data; Step S2: Use the YOLO v5 and Deep SORT combined model target perception algorithm to identify pedestrian motion parameters; Step S3: Determine the simulation scene library portrait index based on the mobile phone signaling data, and calculate the scene complex feature membership through the entropy weight cloud model; Step S4: Based on the questionnaire data, the random forest model is used to quantitatively analyze the helpfulness of facilities and guide signs to determine the importance of explanation; Step S5: Use the station environment data to build an AnyLogic simulation model, including the physical building structures of the transfer corridor, concourse level, and platform level; Step S6: Design a logic module and set the arrival and departure times of simulated pedestrian flows and trains according to the flight and high-speed rail schedules in the air-rail timetable data; Step S7: Run the simulation program to verify the reliability of the simulation model and the accuracy of the results based on the air-rail ticket data; Step S8: According to the streamline congestion bottleneck points displayed in the simulation result heat map, the optimization means of different importance are appropriately matched in combination with the membership of the scene complex features; Step S9: After optimization, the simulation program is run again, which is specifically manifested in the smoothness of the streamline heat map and the complete pedestrian passage time; Step S10: The above process is recorded and saved, and parameters are maintained and updated, and access rights are set.

2. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: Based on the starting and ending points of the air-rail intermodal transfer route of the airport high-speed railway station, determine the points along the route where surveillance video data is to be obtained.

3. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: The mobile phone signaling data automatically delineates the research area on the map in the system background, and selects the hourly granularity personnel flow and personnel portrait API statistical interface.

4. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: The airport flight schedule and railway train timetable information are crawled separately, and the airport and railway ticket data are cleaned, including deleting default values, abnormal values, and duplicate values.

5. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: Questionnaires were distributed to passengers in the waiting hall of the airport high-speed railway station, and the collection process was carried out simultaneously offline and online. The questions included the degree to which the optimization measures helped guide passengers' transfers, and the satisfaction with the convenience of the air-rail transfer process.

6. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: The surveillance videos along the route are used as the data source for parameter identification to obtain the speed of pedestrians in the target area per unit time.

7. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: After loading the video data, the multi-target pedestrian detection unit is responsible for detecting pedestrians using the YOLO v5 algorithm, numbering each target and extracting pedestrian height information.

8. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: The multi-target pedestrian tracking unit is responsible for predicting the position and state of the pedestrian in the next frame based on the pedestrian's position and state using Kalman filtering, and performing cascade matching when the same target is detected in three consecutive frames.

9. The airport high-speed rail transfer streamline optimization method based on simulation scene complexity according to claim 1 is characterized in that: The multi-objective parameter output unit is responsible for calculating the pedestrian's walking speed based on the ratio of the pedestrian's actual walking distance and the adjacent time interval, thereby obtaining the average walking speed of the multi-objective pedestrians.

10. An airport high-speed rail transfer streamline optimization system based on simulation scene complexity, characterized in that: include: Multi-source data import module, pedestrian parameter identification module, simulation scenario establishment module, travel characteristics analysis module, station passenger flow simulation module, transfer flow optimization module; Execute the airport high-speed rail transfer streamline optimization method based on simulation scene complexity as described in any one of claims 1-9.

Images