Urban center employment crowd commuting bottleneck identification method based on multiple agents

Abstract

The invention discloses an urban central employment crowd commuting bottleneck identification method based on multiple agents. Comprising the steps of urban center multi-source database integration and space sand table construction, cloud computing platform employment population-built environment spatial-temporal feature analysis, cloud computing platform employment population-built environment decision rule translation, multi-agent model construction and commuting behavior simulation, and urban center employment population commuting bottleneck identification. And commuting bottleneck data visualization and cyclic interaction feedback are carried out. According to the method, high-precision space-time multi-source data is acquired, and a multi-agent modeling environment is utilized to carry out behavior analysis, rule translation and analog simulation on commuting activities of urban center employment crowds, so that systematized, refined and interactive employment crowd commuting bottleneck identification is realized; and a reliable basis is provided for improving the commuting efficiency of urban central employment crowds.

Description

Technical Field

[0001] The present invention relates to the field of urban planning, and in particular to a method for identifying commuting bottlenecks of employed people in a city center based on multiple intelligent agents. Background Art

[0002] The city center is an important spatial carrier for the high concentration of production and life of the urban population. Among them, the "commuting" formed by urban residents' commuting is one of the important daily spatial activities. Its temporal and spatial demand and distribution characteristics are not only an important basis for urban comprehensive transportation planning, but also related to urban land use, spatial structure and facility layout. With the continuous acceleration of urbanization, the commuting demand of employed people in the city center has increased significantly, resulting in an increasingly prominent commuting bottleneck problem, which not only affects the efficiency of the urban transportation system, but also has a profound impact on the quality of life of residents, the economic development of the city and environmental protection. In view of this, identifying and alleviating the commuting bottleneck of employed people in the city center has become an important research content of urban planning and management.

[0003] At present, the common commuting bottleneck research methods mainly focus on macro-level traffic flow analysis and traffic improvement measures for local sections. Traditional methods include traffic flow statistics, commuting time analysis and traffic simulation models. These methods have certain effects in identifying commuting problems, but they have certain limitations in dealing with the complex and changing urban traffic environment:

[0004] 1. Coarse temporal and spatial granularity: Analysis based on monthly or annual data makes it difficult to capture real-time changes and subtle differences during peak commuting hours;

[0005] 2. Lack of dynamic adaptability: It cannot reflect the dynamic changes of traffic conditions in real time and is difficult to deal with sudden traffic incidents and abnormal commuting patterns;

[0006] 3. Local optimization, ignoring the whole: Focusing on the optimization of a single node or area, ignoring the overall coordination and optimization of the entire commuter network, making it difficult for the solution to achieve significant results on a global scale.

[0007] With the widespread application of spatiotemporal positioning (LBS) big data such as mobile phone signaling, the ability to recognize the spatial activity status of urban residents has been unprecedentedly improved. On the other hand, multi-agent technology is highly adaptable and flexible, and can simulate and predict complex traffic flow situations, thereby providing scientific decision-making support for the precise identification of commuting bottlenecks, dynamic adaptation to urban traffic changes, and immediate global optimization responses. Summary of the invention

[0008] In order to address the deficiencies mentioned in the above-mentioned background technology, the purpose of the present invention is to provide a method for identifying commuting bottlenecks of employed people in urban centers based on multi-agents. This method obtains high-precision spatiotemporal multi-source data and utilizes a multi-agent modeling environment to perform behavioral analysis, rule translation and simulation on the commuting activities of employed people in urban centers, thereby achieving systematic, refined and interactive commuting bottleneck identification of employed people, providing a reliable basis for improving the commuting efficiency of employed people in urban centers.

[0009] The technical solution adopted by the present invention is: a method for identifying commuting bottlenecks of employed people in urban centers based on multi-agents, comprising the following steps:

[0010] Step 1: City center multi-source database integration and space sandbox construction

[0011] The cellular communication module takes each information collection device as the center, sets a 250m×250m grid as the information collection unit, and collects the LBS data generated by the use of the mobile communication network by people passing through each grid range in the past 14 days, that is, 10 working days, including ID, gender, age, time, and longitude and latitude attributes. The acquired LBS data is preprocessed: invalid data users with less than 3 records per day are deleted; based on the maximum speed limit of 60km / h on urban roads, data with a displacement of more than 1km within 1 minute is cleaned; and user data whose mobile trajectory points have not been to the city center are eliminated.

[0012] Input the city center boundary data, city center road data, land use data, building form data, business POI data, oblique photography data and pre-processed city center crowd travel trajectory data into the cloud computing platform to automatically build a multi-source spatiotemporal database of the city center. Import the data into the GIS geographic information platform to build a multi-source data sandbox for the city center crowd.

[0013] Step 2: Analysis of spatiotemporal characteristics of cloud computing platform employment population and built environment

[0014] The travel trajectory data of the target city center population in step 1 is retrieved on the cloud computing platform, and the city center population is clustered using the ST-DBSCAN technology to extract population attribute characteristics of 8 indicators in three dimensions: natural attributes, social attributes, and regional attributes. People who appear in the city center for three consecutive days or more in a week are identified as permanent residents of the city, among which permanent residents aged between 26 and 65 are defined as employed people.

[0015] The travel trajectory data, land use data and business POI data of the commuting population in the city center are retrieved to identify the work-residence and commuting chain of the employed population in the city center. The spatial threshold is defined as 500m, the stay time threshold is 30 minutes, the location where the individual stays during the daytime working hours is defined as the place of employment, and the time period is 8:00-18:00, and the location where the individual stays for a long time at night is defined as the place of residence, and the time period is 19:00-6:00; a trip is identified between two stops in different places, and the travel routes that are repeated for more than 3 days in a week are determined as commuting routes.

[0016] The road data, land use data, building form data and business POI data were mapped and reclassified to obtain a set of characteristic indicators of spatiotemporal activity preferences of employed people, which were divided into four categories according to the supporting facilities, road network, land use and environmental quality characteristics. The influence weight of built environment factors in the urban center on the commuting paths of employed people was calculated by geographically weighted regression analysis, and the characteristic value of the influence of each built environment factor on the commuting behavior of employed people was determined.

[0017] Step 3: Cloud computing platform employment population-built environment decision rule translation

[0018] Based on the attribute characteristics and commuting chain of the employed population in the city center in step 2, the fixed-time activity points and spatiotemporal activity behavior preferences of the employed population are determined. The activity choices of the employed population are calculated using a reinforcement learning algorithm, and the behavioral decision-making of the employed population is translated into rules using an iterative q function, which represents the sum of the rewards expected to be obtained by the real-time travel decision-making actions of multiple agents in the present and future. The q function formula is

[0019] Among them, (s, α) represents the immediate report given by the decision action, Represents the delayed reward brought by the decision action in the future; γ represents the discount factor for the delayed reward. By adjusting the parameter α=1, the convergence speed of the q value is made fastest to achieve the maximum function value of the reward signal, so that the intelligent agent can find the optimal commuting path in an unknown environment.

[0020] Based on the influence weight of the built environment factors in the city center on the commuting routes of employed people in step 2, the influence weight of the built environment factors is translated into rules using the attraction utility function.

[0021] Among them, t is the moment when the individual makes a choice; E{U t (d)} is the expected utility of the selected limb d∈D at time t; S is the selected limb D i The set of possible limbs to choose from; X k(d|s) is the characteristic value of the d-th attribute of the d-selection limb in the given state s∈S; θ k For X k Correlation coefficient; is the probability of state t appearing at time s.

[0022] Step 4: Multi-agent model construction and commuting behavior simulation

[0023] The multi-agent model simulation unit imports the model parameters translated from the employment population-built environment decision rules in step three, and uses rule-based agent modeling to construct a multi-agent model of urban center employment population commuting, simulates the activity behaviors of urban center employment population in commuting states under different built environment conditions, and generates accurate spatiotemporal data of employment population commuting trajectories.

[0024] The spatiotemporal data of simulated commuting trajectories and actual commuting trajectories output by the model were collected, and the root mean square error (RMSE) and R between the model output and the actual data were calculated using the K-fold cross-validation method. 2 , set the RMSE tolerance threshold to 0.15, R 2 The allowable error threshold is 0.7. If RMSE and R 2 If the error is within the preset tolerance range, the model is considered to have a good fitting effect; if not, return to step 3 to re-optimize and correct the model parameters until the model is confirmed.

[0025] Step 5: Identify commuting bottlenecks for people employed in the city center

[0026] The minimum recognition time interval is set to 1 minute in the crowd commuting bottleneck recognition unit, and a 250m×250m grid area is defined as the minimum recognition area with each information collection device as the center. The commuting trajectory points of the urban center employed population simulated by the multi-agent model in step 4 are retrieved to calculate the commuting population density of each recognition area. The population density threshold is set to (μ±2σ), and the cumulative time that the population density of each recognition area exceeds the threshold is calculated. If the cumulative time exceeds 10 minutes, the recognition area is judged to be the commuting bottleneck area of ​​the urban center employed population.

[0027] Step 6: Visualization of commuting bottleneck data and loop feedback

[0028] The human-computer interaction device uses the Dashboard information visualization platform to visualize the commuting simulation results of various types of urban center employment groups in step 4, predict the commuting spatiotemporal behavior of the employed population in real time, and transmit the population density, duration, and spatial distribution data of the commuting bottleneck areas of the employed population in each urban center identified in step 5 to the smart display screen. Managers use PCVR devices to establish a virtual simulation environment, analyze the commuting bottleneck situation in the virtual scene, and use interactive functions to make quick decisions and optimize the bottleneck areas.

[0029] Furthermore, the information collection device is equipped with a cellular communication module, a high-precision GPS device, a high-resolution panoramic camera and an environmental sensor, and has a crowd spatiotemporal information collection unit, a built environment data collection unit and an information data transmission unit, which are arranged at fixed points at intervals of 500m within the city center; the crowd spatiotemporal information collection unit communicates with the base station through a cellular communication module compatible with GSM, CDMA, WCDMA and LTE communication standards, and uses cellular positioning technology to collect the longitude and latitude data of the crowd passing through a grid of 250m×250m and record the timestamp, and uses a high-precision GPS device to improve the positioning accuracy; the built environment data collection unit collects road data, land use data, building form data, business POI data and oblique photography data in the city center through a high-resolution panoramic camera and environmental sensors; the information data transmission unit transmits the data collected by the crowd spatiotemporal information collection unit and the built environment data collection unit to the cloud computing platform;

[0030] The cloud computing platform is equipped with a DGX computer workstation, and has an employment population-built environment spatiotemporal feature analysis unit, an employment population-built environment rule translation unit, a multi-agent model simulation unit, and a population commuting bottleneck identification unit; the employment population-built environment spatiotemporal feature analysis unit has a population feature identification module and a built environment factor characterization module, wherein the population feature identification module is used to cluster the employment population in the city center, identify the spatiotemporal features of the population and extract the commuting activity chain, and the built environment factor characterization module is used to calculate the weighted average of the impact of each built environment factor in the city center on the commuting behavior of the employment population; the employment population-built environment rule translation unit is used to translate the ontology decision rules and the environmental impact mechanism and output the model parameters; the multi-agent model simulation unit is used to construct a multi-agent model of the employment population commuting in the city center and simulate the commuting behavior of the population; the population commuting bottleneck identification unit is used to predict and identify the commuting bottleneck area according to the commuting trajectory of the employment population simulated by the multi-agent model;

[0031] The human-computer interaction device is equipped with a dashboard information visualization platform, a smart display screen, and a PCVR device, and has a data visualization unit and a human-computer interaction unit; the data visualization unit is used to visualize the prediction model and the commuting bottleneck area data of the employed population; the human-computer interaction unit is used for management personnel to adjust model parameters, simulate the commuting behavior of the employed population under different conditions, and immediately propose optimization plans.

[0032] Furthermore, in the step one, a multi-source data sandbox of the city center crowd is constructed. The cloud computing platform collects the latitude and longitude data and timestamp data of the crowd passing through a 250m×250m unit grid through communication with the base station based on the cellular communication module, and the road data, land use data, building form data and business POI data of the city center area collected by the high-resolution panoramic camera and environmental sensors. The above data are stored in the shp format to construct a multi-source spatiotemporal database of the city center. The data format is unified and input into the GIS geographic information platform, and spatially overlapped with the oblique photography data to construct a multi-source data sandbox of the city center crowd.

[0033] Furthermore, the population attribute characteristics of the three dimensions of natural attributes, social attributes, and regional attributes, totaling 8 indicators, in step 2 are shown in Table 2:

[0034] Table 2 Crowd attribute characteristics

[0035]

[0036] Furthermore, the characteristic index set of the spatiotemporal activity preference of the employed population in step 2 is shown in Table 3:

[0037] Table 3. Characteristics of the temporal and spatial activity preference indicators of employed people

[0038]

[0039] Furthermore, in step three, the behavioral decision-making of employed people is rule-translated. The state of the commuting individual is composed of four dimensions: commuting start time, commuting duration, current grid position, and current road traffic volume. The commuting state of the individual is obtained at a time interval of 10 minutes. One day can be divided into 144 time steps. For each commuting state, the commuting individual has two choices: stay in the current commuting activity or transfer to the next commuting activity. The state of the commuting individual changes dynamically with the decision-making of the commuting individual. The road congestion level is reflected in the reward function based on the commuting state, so that the intelligent agent can find the optimal commuting path in an unknown environment.

[0040] Furthermore, the attraction utility function in step 3 performs rule translation on the influence weight of built environment factors, and uses the co-evolutionary logit model to deduce the influence of built environment factors on the commuting modes of commuting individuals in the city center. The commuting modes of commuting individuals are extracted into four categories: walking, non-motorized vehicles, motor vehicles, and public transportation. The built environment factors are used as influencing factors, and the multinomial logit (MNL) model is used to determine the probability of the commuting mode being selected. The formula is: Among them, P{I t (d|s)} is the choice of D at time t i Select limb d∈D i The probability of being selected; I(d|s) is the probability of selecting limb d in determining D i The probability of being selected under the choice state s.

[0041] Furthermore, the spatiotemporal data of the commuting trajectories of employed people in step 4 include user ID, commuting time period, commuting route, commuting timestamp and longitude and latitude information of their corresponding positions.

[0042] Furthermore, in step six, the commuting simulation results of various types of urban center employed people are visualized, and the surrounding LBS and built environment data are collected by using an information collection device. The spatiotemporal data of the commuting trajectories of the urban center employed people are calculated in conjunction with the cloud computing platform, and the commuting data of the predicted commuting bottleneck areas are converted into visualized distribution cloud maps and vector maps, which are fed back to the smart display screen.

[0043] Furthermore, the step six utilizes interactive functions to make rapid decisions and optimizations on bottleneck areas, that is, the information collection device can collect real-time monitoring data and synchronize it to the cloud computing platform, verify the results of the predictions of the collected previous data, and realize the precise and instant information collection and interactive feedback of the commuting bottlenecks of the commuting population in the city center. The management personnel use the PCVR equipment to make decision responses to the commuting bottleneck areas through voice commands or touch screen control, realize the prevention of commuting bottlenecks and crowd management in the city center area, and improve the commuting efficiency of the employed population in the city center.

[0044] Beneficial effects of the present invention:

[0045] 1. Process efficiency: The present invention realizes the full process digital operation of the commuting situation of the employed people in the city center from data collection, feature analysis, rule translation, model building, behavior simulation, bottleneck identification to visual interaction through the information collection device equipped with cellular communication module, high-precision GPS equipment, high-resolution panoramic camera and environmental sensor, cloud computing platform equipped with DGX computer workstation, dashboard information visualization platform, smart display large screen and PCVR equipment. This system solves the inefficiency problem of traditional commuting monitoring, establishes a scientific and reasonable prediction and recognition model system, realizes the complete process of large-scale and full-coverage information collection, intelligent analysis, and decision-making closed loop in a short time, and greatly improves the coordination of refined analysis and dynamic monitoring of commuting bottlenecks.

[0046] 2. Multidimensional analysis: Starting from the population itself and the built environment, the present invention selects multidimensional factors to analyze the population feature identification module and the built environment factor characterization module of the model respectively, so as to comprehensively understand the attribute characteristics and commuting behavior of the employed population, accurately identify the impact of built environment factors on commuting, and provide scientific and accurate decision-making support for the construction of a commuting model for employed people in the city center, effectively improving the level of urban management and the commuting experience of employed people.

[0047] 3. Interactive visualization: The present invention uses a large smart display screen and PCVR equipment to assist city managers and personnel in identifying commuting bottleneck areas for people working in the city center more intuitively and clearly, providing two-dimensional image and three-dimensional immersive presentations, and realizing interactive decision-making for dynamic identification and optimization of commuting bottlenecks for people working in the city center. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of the method implemented by the present invention;

[0049] Figure 2 It is a composition diagram of the identification system unit of the present invention;

[0050] Figure 3 It is a flow chart of analyzing the spatiotemporal characteristics of commuting population in the city center of the present invention;

[0051] Figure 4 It is an information interaction device composed of Dashboard information visualization platform and PCVR. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0053] A multi-agent-based method for identifying commuting bottlenecks for employed people in urban centers, such as Figure 1-4 As shown, the specific steps include:

[0054] 1. Build a multi-source database integration and space sandbox in the city center, set up a cellular communication module with each information collection device as the center, set a 250m×250m grid as the information collection unit, and collect the LBS data generated by the use of mobile communication networks by people passing through each grid range in the past 14 days, that is, 10 working days, including ID, gender, age, time and longitude and latitude attributes. Pre-process the acquired LBS data: delete invalid data users with less than 3 records per day; use the maximum speed limit of 60km / h on urban roads as the standard, clean up data with a displacement of more than 1km within 1 minute; and eliminate user data whose mobile trajectory points have not been to the city center.

[0055] The information collection device is equipped with a cellular communication module, a high-precision GPS device, a high-resolution panoramic camera and an environmental sensor, and has a crowd spatiotemporal information collection unit, a built environment data collection unit and an information data transmission unit, which are arranged at fixed points at intervals of 500m within the city center; the crowd spatiotemporal information collection unit communicates with the base station through a cellular communication module compatible with GSM, CDMA, WCDMA and LTE communication standards, and uses cellular positioning technology to collect the longitude and latitude data of the crowd passing through a grid of 250m×250m and record the timestamp, and uses a high-precision GPS device to improve the positioning accuracy; the built environment data collection unit collects road data, land use data, building form data, business POI data and oblique photography data in the city center through a high-resolution panoramic camera and environmental sensors; the information data transmission unit transmits the data collected by the crowd spatiotemporal information collection unit and the built environment data collection unit to the cloud computing platform;

[0056] The LBS data fields are shown in Table 1:

[0057] Table 1 LBS data fields

[0058]

[0059] The method for constructing a multi-source data sandbox of the city center crowd is based on the cloud computing platform, which collects the latitude and longitude data and timestamp data of the crowd passing through a 250m×250m grid through communication with the base station through a cellular communication module, and the road data, land use data, building form data and business POI data of the city center area collected by a high-resolution panoramic camera and environmental sensors. The above data are stored in the shp format to construct a multi-source spatiotemporal database of the city center, and the data format is unified and input into the GIS geographic information platform, and spatially overlapped with the oblique photography data to construct a multi-source data sandbox of the city center crowd.

[0060] 2. Analysis of spatiotemporal characteristics of the employed population and the built environment on the cloud computing platform. The travel trajectory data of the target city center population in step 1 is retrieved on the cloud computing platform, and the ST-DBSCAN technology is used to cluster the city center population, extracting population attribute characteristics of 8 indicators in three dimensions: natural attributes, social attributes, and regional attributes. People who appear in the city center for three consecutive days or more in a week are identified as the city's permanent population, among which the permanent population aged between 26 and 65 is defined as the employed population.

[0061] The cloud computing platform is equipped with a DGX computer workstation, and has an employment population-built environment spatiotemporal feature analysis unit, an employment population-built environment rule translation unit, a multi-agent model simulation unit, and a population commuting bottleneck identification unit; the employment population-built environment spatiotemporal feature analysis unit has a population feature identification module and a built environment factor characterization module, wherein the population feature identification module is used to cluster the employment population in the city center, identify the spatiotemporal features of the population and extract the commuting activity chain, and the built environment factor characterization module is used to calculate the weighted average of the impact of each built environment factor in the city center on the commuting behavior of the employment population; the employment population-built environment rule translation unit is used to translate the ontology decision rules and the environmental impact mechanism and output the model parameters; the multi-agent model simulation unit is used to construct a multi-agent model of the employment population commuting in the city center and simulate the commuting behavior of the population; the population commuting bottleneck identification unit is used to predict and identify the commuting bottleneck area according to the commuting trajectory of the employment population simulated by the multi-agent model;

[0062] The travel trajectory data, land use data and business POI data of the commuting population in the city center are retrieved to identify the work-residence and commuting chain of the employed population in the city center. The spatial threshold is defined as 500m, the stay time threshold is 30 minutes, the location where the individual stays during the daytime working hours is defined as the place of employment, and the time period is 8:00-18:00, and the location where the individual stays for a long time at night is defined as the place of residence, and the time period is 19:00-6:00; a trip is identified between two stops in different places, and the travel routes that are repeated for more than 3 days in a week are determined as commuting routes.

[0063] The road data, land use data, building form data and business POI data were mapped and reclassified to obtain a set of characteristic indicators of spatiotemporal activity preferences of employed people, which were divided into four categories according to the supporting facilities, road network, land use and environmental quality characteristics. The influence weight of built environment factors in the urban center on the commuting paths of employed people was calculated by geographically weighted regression analysis, and the characteristic value of the influence of each built environment factor on the commuting behavior of employed people was determined.

[0064] The crowd attribute characteristics of 8 indicators in the three dimensions of natural attributes, social attributes, and regional attributes in step 2 are shown in Table 2:

[0065] Table 2 Crowd attribute characteristics

[0066]

[0067] The characteristic index set of the temporal and spatial activity preference of the employed population in step 2 is shown in Table 3:

[0068] Table 3. Characteristics of the temporal and spatial activity preference indicators of employed people

[0069]

[0070]

[0071] 3. Translate the decision rules of the employed population in the cloud computing platform to the built environment. Based on the attribute characteristics and commuting chain of the employed population in the city center in step 2, determine the fixed-time activity points and spatiotemporal activity behavior preferences of the employed population. Use the reinforcement learning algorithm to calculate the activity choices of the employed population, and use the iterative q function to translate the rules of the behavioral decisions of the employed population, which represents the sum of the rewards expected to be obtained by the real-time travel decision actions of multiple intelligent agents in the present and future. The q function formula

[0072] Among them, (s, α) represents the immediate report given by the decision action, Represents the delayed reward brought by the decision action in the future; γ represents the discount factor for the delayed reward. By adjusting the parameter α=1, the convergence speed of the q value is made fastest to achieve the maximum function value of the reward signal, so that the intelligent agent can find the optimal commuting path in an unknown environment.

[0073] Based on the influence weight of the built environment factors in the city center on the commuting routes of employed people in step 2, the influence weight of the built environment factors is translated into rules using the attraction utility function.

[0074] Among them, t is the moment when the individual makes a choice; E{U t (d)} is the expected utility of the selected limb d∈D at time t; S is the selected limb D i The set of possible limbs to choose from; X k (d|s) is the characteristic value of the d-th attribute of the d-selection limb in the given state s∈S; θ k For X k Correlation coefficient; is the probability of state t appearing at time s.

[0075] The behavioral decision-making of the employed population is rule-translated. The state of the commuting individual is composed of four dimensions: commuting start time, commuting duration, current grid position, and current road traffic volume. The commuting state of the individual is obtained at a time interval of 10 minutes. One day can be divided into 144 time steps. For each commuting state, the commuting individual has two choices: stay in the current commuting activity or transfer to the next commuting activity. The state of the commuting individual changes dynamically with the decision of the commuting individual. The road congestion is reflected in the reward function based on the commuting state, so that the intelligent agent can find the optimal commuting path in an unknown environment.

[0076] The attraction utility function is used to translate the influence weights of built environment factors. The co-evolutionary logit model is used to deduce the influence of built environment factors on the commuting modes of commuting individuals in the city center. The commuting modes of commuting individuals are extracted into four categories: walking, non-motorized vehicles, motor vehicles, and public transportation. The built environment factors are used as influencing factors, and the multinomial logit (MNL) model is used to determine the probability of the commuting mode being selected. The formula is: Among them, P{I t (d|s)} is the choice of D at time t i Select limb d∈D i The probability of being selected; I(d|s) is the probability of selecting limb d in determining D i The probability of being selected under the choice state s.

[0077] 4. Multi-agent model construction and commuting behavior simulation. The multi-agent model simulation unit imports the model parameters translated from the employment population-built environment decision-making rules in step 3, and uses rule-based agent modeling to construct a multi-agent model of commuting of the employed population in the city center, simulates the activity behavior of the employed population in the city center under different built environment conditions during commuting, and generates accurate spatiotemporal data of the commuting trajectory of the employed population.

[0078] The spatiotemporal data of simulated commuting trajectories and actual commuting trajectories output by the model were collected, and the root mean square error (RMSE) and R between the model output and the actual data were calculated using the K-fold cross-validation method. 2 , set the RMSE tolerance threshold to 0.15, R 2 The allowable error threshold is 0.7. If RMSE and R 2 If the error is within the preset tolerance range, the model is considered to have a good fitting effect; if not, return to step 3 to re-optimize and correct the model parameters until the model is confirmed.

[0079] The spatiotemporal data of the commuting trajectory of the employed population includes user ID, commuting time period, commuting route, commuting timestamp and longitude and latitude information of their corresponding locations.

[0080] 5. Identification of commuting bottlenecks for employed people in the city center: set the minimum identification time interval to 1 minute in the commuting bottleneck identification unit, define a 250m×250m grid area as the minimum identification area with each information collection device as the center, retrieve the commuting trajectory points of employed people in the city center simulated by the multi-agent model in step 4, and calculate the commuting population density of each identification area. Set the population density threshold to (μ±2σ), and calculate the cumulative time that the population density of each identification area exceeds the threshold. If the cumulative time exceeds 10 minutes, the identification area is judged to be a commuting bottleneck area for employed people in the city center.

[0081] 6. Visualization of commuting bottleneck data and cyclic interactive feedback. The human-computer interaction device uses the Dashboard information visualization platform to visualize the commuting simulation results of various types of urban center employment groups in step 4, predict the commuting spatiotemporal behavior of the employed population in real time, and transmit the population density, duration, and spatial distribution data of the commuting bottleneck areas of the employed population in each urban center identified in step 5 to the smart display screen. Managers use PCVR devices to establish a virtual simulation environment, analyze the commuting bottleneck situation in the virtual scene, and use interactive functions to make quick decisions and optimize the bottleneck area.

[0082] The human-computer interaction device is equipped with a dashboard information visualization platform, a smart display screen, and a PCVR device, and has a data visualization unit and a human-computer interaction unit; the data visualization unit is used to visualize the prediction model and the commuting bottleneck area data of the employed population; the human-computer interaction unit is used for management personnel to adjust model parameters, simulate the commuting behavior of the employed population under different conditions, and immediately propose optimization plans;

[0083] The commuting simulation results of various types of urban center employees are visualized, and the surrounding LBS and built environment data are collected by using the information collection device. The spatiotemporal data of the commuting trajectory of the urban center employees are calculated in combination with the cloud computing platform, and the commuting data of the predicted commuting bottleneck area is converted into a visualized distribution cloud map and vector map, which are fed back to the smart display screen;

[0084] The interactive function is used to make rapid decisions and optimizations on bottleneck areas. That is, the information collection device can collect real-time monitoring data and synchronize it to the cloud computing platform, verify the results of the predictions of the previous data collected, and realize the detailed and instant information collection and interactive feedback of the commuting bottlenecks of the commuting population in the city center. The management personnel use the PCVR equipment to make decision responses to the commuting bottleneck areas through voice commands or touch screen control, realize the prevention of commuting bottlenecks and crowd management in the city center area, and improve the commuting efficiency of the employed population in the city center.

[0085] Example

[0086] The technical solution of the present invention will be described in detail below by taking the central urban area of ​​Nanjing as an example.

[0087] (1) Taking the central urban area of ​​Nanjing as the target area, we obtained the three-dimensional vector data of the central urban area of ​​Nanjing and the travel path data of residents, and generated a multi-source data sandbox of the crowd in the central urban area of ​​Nanjing, including:

[0088] (1.1) A cellular communication module compatible with GSM, CDMA, WCDMA, and LTE communication standards is used to communicate with the base station. Using cellular positioning technology, a 250m×250m grid is set as an information collection unit to collect LBS data generated by the use of mobile communication networks by people passing through each grid range in the past 14 days, i.e., 10 working days, including ID, gender, age, time, and latitude and longitude attributes.

[0089] The LBS data fields are shown in Table 1:

[0090] Table 1 LBS data fields

[0091]

[0092] (1.2) Preprocess the acquired LBS data: delete invalid data users with less than 3 records per day; use the maximum speed limit of 60km / h on urban roads as the standard to clean data with a displacement of more than 1km within 1 minute; and eliminate user data whose movement trajectory points have not reached the city center.

[0093] (1.3) Obtain the boundary data of the central area of ​​Nanjing from the Nanjing Planning Department, and use high-resolution panoramic cameras and environmental sensors to collect the road network, building outlines, building heights and land use data of the central area of ​​Nanjing, including the CAD / SHP file of the boundary of the central urban area of ​​Nanjing, the CAD / SHP file of the road network of the central urban area of ​​Nanjing, the CAD / SHP file of the current land use blocks, and the CAD / SHP file of the current closed buildings and the number of floors (height).

[0094] (1.4) The CAD files of the current 3D vector data of the central urban area of ​​Nanjing, including the boundaries, road network, land use and building form, are exported as SHP data of closed polylines; the data coordinates are converted by “projection and transformation” operations, including switching between projection coordinates and geographic coordinates and conversion between different coordinate systems. The 3D vector data of the central urban area of ​​Nanjing is unified into the 2000 National Geodetic Coordinate System, and a multi-source spatiotemporal database of the urban center is automatically constructed. The data is then imported into the GIS geographic information platform; the data is spatially overlapped with the pre-processed LBS data to construct a multi-source data sandbox of the urban center population.

[0095] (2) Using cloud computing platforms to analyze the spatiotemporal characteristics of the employed population and built environment in the central urban area of ​​Nanjing, including:

[0096] (2.1) Calculate the attribute characteristic values ​​of the employed population in the central urban area of ​​Nanjing, including natural attributes, social attributes, and regional attributes. Use ST-DBSCAN technology to cluster and identify the attributes of the population in the central urban area of ​​Nanjing, extract the attribute characteristics of the population in three dimensions of natural attributes, social attributes, and regional attributes, and a total of 8 indicators. The population that appears in the central urban area for three consecutive days or more in a week is identified as the permanent population of the city, among which the permanent population aged between 26 and 65 is defined as the employed population.

[0097] The crowd attribute characteristics are shown in Table 2:

[0098] Table 2 Crowd attribute characteristics

[0099]

[0100] (2.2) Identify the work-residence and commuting chains of employed people in the central urban area of ​​Nanjing, including travel distance, facility use, and behavior chain type. Import the travel trajectory data, land use data, and business POI data of people in the central urban area of ​​Nanjing preprocessed in step (1) into the cloud computing platform, and determine the spatial threshold and residence time threshold as 500m and 30 minutes respectively. Define the location where individuals stay during the daytime working hours as the place of employment, and the time period is 8:00-18:00. Define the location where individuals stay for a long time at night as the place of residence, and the time period is 19:00-6:00. Identify a trip between two stops at different places, and the travel route that appears more than 3 days in a week is determined as the commuting route.

[0101] (2.3) The road data, land use data, building form data and business POI data were mapped and reclassified to obtain a set of characteristic indicators of spatiotemporal activity preferences of employed people, which were divided into four categories according to the supporting facilities, road network characteristics, land use characteristics and environmental quality characteristics. The influence weights of the built environment factors in the central area of ​​Nanjing on the commuting routes of employed people were calculated by the geographically weighted regression analysis method, and the characteristic values ​​of the influence of each built environment factor on the commuting behavior of employed people were determined.

[0102] The characteristic index set of the temporal and spatial activity preference of the employed population is shown in Table 3:

[0103] Table 3. Characteristics of the temporal and spatial activity preference indicators of employed people

[0104]

[0105] (3) Translate the behavioral decision-making of employed people in the central urban area of ​​Nanjing and the weight rules of the built environment factors on the commuting routes of employed people, including:

[0106] (3.1) Extract the individual commuting status in the central urban area of ​​Nanjing and identify the dynamic changes in the decision-making of commuting individuals. The status of commuting individuals is composed of four dimensions: commuting start time, commuting duration, current grid position, and current road traffic volume. The commuting status of individuals is obtained at a time interval of 10 minutes. One day can be divided into 144 time steps. For each commuting status, commuting individuals have two choices: stay in the current commuting activity or transfer to the next commuting activity. The status of commuting individuals changes dynamically with the decision-making of commuting individuals.

[0107] (3.2) Reinforcement learning algorithms are used to calculate the activity choices of commuting individuals and to translate the rules of the behavioral decisions of employed people. An iterable q function is used to represent the sum of the rewards expected to be obtained by the real-time travel decision actions of multiple agents in the present and future. The q function formula is Among them, (s, α) represents the immediate report given by the decision action, Represents the delayed reward brought by the decision action in the future; γ represents the discount factor for the delayed reward. By adjusting the parameter α=1, the convergence speed of the q value is made fastest to achieve the maximum function value of the reward signal, so that the intelligent agent can find the optimal commuting path in an unknown environment.

[0108] (3.3) The co-evolutionary logit model is used to deduce the impact of built environment factors on the commuting modes of commuting individuals in the city center. The commuting modes of commuting individuals are extracted into four categories: walking, non-motorized vehicles, motor vehicles, and public transportation. The built environment factors are used as influencing factors, and the multi-logit (MNL) model is used to determine the probability of the commuting mode being selected. The formula is: Among them, P{I t (d|s)} is the choice of D at time t i Select limb d∈D i The probability of being selected; I(d|s) is the probability of selecting limb d in determining D i The probability of being selected under the choice state s.

[0109] (3.4) The attraction utility function is used to translate the influence weights of built environment factors into rules. Among them, t is the moment when the individual makes a choice; E{U t (d)} is the expected utility of the selected limb d∈D at time t; S is the selected limb D i The set of possible limbs to choose from; X k (d|s) is the characteristic value of the d-th attribute of the d-selection limb in the given state s∈S; θ k For X k Correlation coefficient; is the probability of state t appearing at time s.

[0110] (4) Construct a multi-agent model simulation unit to simulate the commuting behavior of employed people in the central urban area of ​​Nanjing, including:

[0111] (4.1) Import the model parameters translated from the decision rules of the employment population-built environment in step (3) into the multi-agent model simulation unit. A multi-agent model of the commuting population of the urban center is constructed by rule-based agent modeling to simulate the activity behavior of the commuting population of the urban center under different built environment conditions, and generate accurate spatiotemporal data of the commuting trajectory of the employment population, including user ID, commuting period, commuting path, commuting timestamp and the latitude and longitude information of the corresponding location.

[0112] (4.2) Collect the spatiotemporal data of simulated commuting trajectories output by the model and the spatiotemporal data of actual commuting trajectories, and use the K-fold cross-validation method to calculate the root mean square error (RMSE) and R between the model output and the actual data. 2 , set the RMSE tolerance threshold to 0.15, R 2 The allowable error threshold is 0.7. If RMSE and R 2 If the error is within the preset tolerance range, the model is considered to have a good fitting effect; if not, return to step (3) to re-optimize and correct the model parameters until the model is confirmed.

[0113] (5) Identify commuting bottlenecks for employed people in the central urban area of ​​Nanjing, including:

[0114] (5.1) The minimum recognition time interval is set to 1 minute in the crowd commuting bottleneck recognition unit, and a grid area of ​​250m×250m is defined as a single minimum recognition area with each information collection device as the center. The commuting trajectory points of the employed population in the city center simulated by the multi-agent model in step (4) are retrieved to calculate the commuting population density of each recognition area.

[0115] (5.2) Set the population density threshold to (μ±2σ) and calculate the cumulative duration that the population density of each identified area exceeds the threshold. If the cumulative duration exceeds 10 minutes, the identified area is judged to be a commuting bottleneck area for the employed population in the city center.

[0116] (6) Visualize and interactively feedback the commuting bottlenecks of commuters in the city center, including:

[0117] (6.1) The commuting simulation results of employed people in the central urban area of ​​Nanjing outputted in step (4) are imported into the Dashboard information visualization platform device to predict the commuting spatiotemporal behavior of employed people in real time.

[0118] (6.2) The population density, duration, and spatial distribution data of the commuting bottleneck area for employed people in the central urban area of ​​Nanjing identified in step (5) are transmitted to the smart display screen to verify the prediction results of the collected previous data, so as to realize the precise and timely information collection and interactive feedback of the commuting bottleneck of the commuting population in the city center.

[0119] (6.3) Managers use PCVR devices to establish a virtual simulation environment, analyze the commuting bottleneck situation in the virtual scene, and make decisions and responses to the commuting bottleneck area through voice commands or touch screen control, thereby achieving commuting bottleneck prevention and crowd management in the city center area, and improving the commuting efficiency of the employed population in the city center.

[0120] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A multi-agent based method for identifying commuting bottlenecks for employed people in urban centers, comprising the following steps: Step 1: City center multi-source database integration and space sandbox construction The cellular communication module takes each information collection device as the center, sets a 250m×250m grid as the information collection unit, and collects the LBS data generated by the use of the mobile communication network by people passing through each grid range in the past 14 days, that is, 10 working days, including ID, gender, age, time and latitude and longitude attributes; the cloud computing platform pre-processes the acquired LBS data: deletes invalid data users with less than 3 records per day; uses the maximum speed limit of 60km / h on urban roads as the standard, cleans data with a displacement of more than 1km within 1 minute; and removes user data whose mobile trajectory points have not been to the city center; Input city center boundary data, city center road data, land use data, building form data, business POI data, oblique photography data and pre-processed city center crowd travel trajectory data into the cloud computing platform to automatically build a multi-source spatiotemporal database of the city center, and import the data into the GIS geographic information platform to build a multi-source data sandbox for the city center crowd; Step 2: Analysis of spatiotemporal characteristics of cloud computing platform employment population and built environment The travel trajectory data of the target city center population in step 1 is retrieved on the cloud computing platform, and the city center population is clustered using the ST-DBSCAN technology to extract the population attribute characteristics of 8 indicators in three dimensions of natural attributes, social attributes, and regional attributes. The population that appears in the city center for three consecutive days or more in a week is identified as the city's permanent population, among which the permanent population aged between 26 and 65 is defined as the employed population; The travel trajectory data, land use data and business POI data of the commuting population in the city center are retrieved to identify the work-residence and commuting chain of the employed population in the city center; the spatial threshold is defined as 500m, the stay time threshold is 30 minutes, the location where the individual stays during the daytime working hours is defined as the place of employment, and the time period is 8:00-18:00, and the location where the individual stays for a long time at night is defined as the place of residence, and the time period is 19:00-6:00; a trip is identified between two stops in different places, and the travel routes that are repeated for more than 3 days in a week are determined as commuting routes; Road data, land use data, building form data and business POI data were mapped and reclassified to obtain a set of characteristic indicators of spatiotemporal activity preferences of employed people, which were divided into four categories according to supporting facilities, road network characteristics, land use characteristics and environmental quality characteristics. The influence weight of built environment factors in the urban center on the commuting paths of employed people was calculated by geographically weighted regression analysis, and the characteristic value of the influence of each built environment factor on the commuting behavior of employed people was determined. Step 3: Cloud computing platform employment population-built environment decision rule translation Based on the attribute characteristics and commuting chain of the employed population in the city center in step 2, determine the fixed-time activity points and spatiotemporal activity behavior preferences of the employed population; The reinforcement learning algorithm is used to calculate the activity choices of the employed population, and the iterative q function is used to translate the rules of the behavioral decisions of the employed population, which represents the sum of the rewards expected to be obtained by the real-time travel decision actions of multiple agents in the current and future. The q function formula q'(s,α)&= Among them, (s, α) represents the immediate report given by the decision action, represents the delayed reward brought by the decision action in the future; γ represents the discount factor for the delayed reward. By adjusting the parameter α=1, the convergence speed of the q value is made fastest to achieve the maximum function value of the reward signal, so that the intelligent agent can find the optimal commuting path in an unknown environment; Based on the influence weight of the built environment factors in the urban center on the commuting routes of employed people in step 2, the influence weight of the built environment factors is translated into rules using the attraction utility function. Among them, t is the moment when the individual makes a choice; E{U t (d)} is the expected utility of the selected limb d∈D at time t; S is the selected limb D i The set of possible limbs to choose from; X k (d|s) is the characteristic value of the d-th attribute of the d-selection limb in the given state s∈S; θ k For X k Correlation coefficient; is the probability of state t appearing at time s; Step 4: Multi-agent model construction and commuting behavior simulation The multi-agent model simulation unit imports the model parameters translated from the employment population-built environment decision rule in step 3, and uses rule-based agent modeling to build a multi-agent model of the employment population commuting in the city center, simulates the activity behaviors of the employment population in the city center under different built environment conditions in the commuting state, and generates accurate spatiotemporal data of the employment population's commuting trajectory; The spatiotemporal data of simulated commuting trajectories and actual commuting trajectories output by the model were collected, and the RMSE and R of the model output and actual data were calculated using the K-fold cross-validation method. 2 , set the RMSE tolerance threshold to 0.15, R 2 The allowable error threshold is 0.

7. If RMSE and R 2 If the error is within the preset tolerance range, the model fitting effect is considered good; if not, return to step 3 to re-optimize and correct the model parameters until the model is confirmed; Step 5: Identify commuting bottlenecks for people employed in the city center The minimum recognition time interval is set to 1 minute in the crowd commuting bottleneck recognition unit, and a 250m×250m grid area is defined as the minimum recognition area with each information collection device as the center. The commuting trajectory points of the employed people in the city center simulated by the multi-agent model in step 4 are retrieved to calculate the commuting population density of each recognition area; the population density threshold is set to (μ±2σ), and the cumulative time that the population density of each recognition area exceeds the threshold is calculated. If the cumulative time exceeds 10 minutes, the recognition area is judged to be the commuting bottleneck area of ​​the employed people in the city center; Step 6: Visualization of commuting bottleneck data and loop feedback The human-computer interaction device uses the Dashboard information visualization platform to visualize the commuting simulation results of various types of urban center employed people in step 4, predict the commuting spatiotemporal behavior of the employed people in real time, and transmit the population density, duration, and spatial distribution data of the commuting bottleneck areas of the employed people in each urban center identified in step 5 to the smart display large screen; in the human-computer interaction unit, managers use PCVR devices to establish a virtual simulation environment, analyze the commuting bottleneck situation in the virtual scene, and use interactive functions to make quick decisions and optimize the bottleneck areas.

2. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 1, characterized in that: The information collection device is equipped with a cellular communication module, a high-precision GPS device, a high-resolution panoramic camera and an environmental sensor, and has a crowd spatiotemporal information collection unit, a built environment data collection unit and an information data transmission unit, which are arranged at fixed points at intervals of 500m within the city center; the crowd spatiotemporal information collection unit communicates with the base station through a cellular communication module compatible with GSM, CDMA, WCDMA and LTE communication standards, and uses cellular positioning technology to collect the longitude and latitude data of the crowd passing through a grid of 250m×250m and record the timestamp, and uses a high-precision GPS device to improve the positioning accuracy; the built environment data collection unit collects road data, land use data, building form data, business POI data and oblique photography data in the city center through a high-resolution panoramic camera and environmental sensors; the information data transmission unit transmits the data collected by the crowd spatiotemporal information collection unit and the built environment data collection unit to the cloud computing platform; The cloud computing platform is equipped with a DGX computer workstation, and has an employment population-built environment spatiotemporal feature analysis unit, an employment population-built environment rule translation unit, a multi-agent model simulation unit, and a population commuting bottleneck identification unit; the employment population-built environment spatiotemporal feature analysis unit has a population feature identification module and a built environment factor characterization module, wherein the population feature identification module is used to cluster the employment population in the city center, identify the spatiotemporal features of the population and extract the commuting activity chain, and the built environment factor characterization module is used to calculate the weighted average of the impact of each built environment factor in the city center on the commuting behavior of the employment population; the employment population-built environment rule translation unit is used to translate the ontology decision rules and the environmental impact mechanism and output the model parameters; the multi-agent model simulation unit is used to construct a multi-agent model of the employment population commuting in the city center and simulate the commuting behavior of the population; the population commuting bottleneck identification unit is used to predict and identify the commuting bottleneck area according to the commuting trajectory of the employment population simulated by the multi-agent model; The human-computer interaction device is equipped with a dashboard information visualization platform, a smart display screen, and a PCVR device, and has a data visualization unit and a human-computer interaction unit; the data visualization unit is used to visualize the prediction model and the commuting bottleneck area data of the employed population; The human-computer interaction unit is used for managers to adjust model parameters, simulate the commuting behavior of employed people under different conditions, and immediately propose optimization plans.

3. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 2 is characterized in that: In the step 1, a multi-source data sandbox of the city center crowd is constructed. The cloud computing platform collects the latitude and longitude data and timestamp data of the crowd passing through a 250m×250m grid based on the communication between the cellular communication module and the base station, as well as the road data, land use data, building form data and business POI data of the city center collected by the high-resolution panoramic camera and environmental sensors. The above data are stored in the shp format to construct a multi-source spatiotemporal database of the city center. The data format is unified and input into the GIS geographic information platform, and spatially overlapped with the oblique photography data to construct a multi-source data sandbox of the city center crowd.

4. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 3 is characterized in that: The crowd attribute characteristics of 8 indicators in the three dimensions of natural attributes, social attributes, and regional attributes in step 2 are shown in Table 2: Table 2 Crowd attribute characteristics 5. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 4, characterized in that: The characteristic index set of the temporal and spatial activity preference of the employed population in step 2 is shown in Table 3: Table 3. Characteristics of the temporal and spatial activity preference indicators of employed people 6. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 5, characterized in that: The step three is to translate the rules of the behavioral decisions of the employed population. The state of the commuting individual is composed of four dimensions: commuting start time, commuting duration, current grid position, and current road traffic volume. The commuting state of the individual is obtained at a time interval of 10 minutes. One day can be divided into 144 time steps. For each commuting state, the commuting individual has two choices: stay in the current commuting activity or transfer to the next commuting activity. The state of the commuting individual changes dynamically with the decision of the commuting individual. The road congestion level is reflected in the reward function based on the commuting state, so that the intelligent agent can find the optimal commuting path in an unknown environment.

7. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 6, characterized in that: In step 3, the attraction utility function is used to translate the influence weights of built environment factors into rules, and the co-evolutionary logit model is used to deduce the influence of built environment factors on the commuting modes of commuting individuals in the city center. The commuting modes of commuting individuals are extracted into four categories: walking, non-motorized vehicles, motor vehicles, and public transportation. The built environment factors are used as influencing factors, and the multinomial logit (MNL) model is used to determine the probability of the commuting mode being selected. The formula is: Among them, P{I t (d|s)} is the choice of D at time t i Select limb d∈D i The probability of being selected; I(d|s) is the probability of selecting limb d in determining D i The probability of being selected under the choice state s.

8. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 7, characterized in that: The step 4 is to obtain the spatiotemporal data of the commuting trajectories of employed people, including user ID, commuting time period, commuting route, commuting timestamp and the longitude and latitude information of their corresponding locations.

9. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 8, characterized in that: The step six visualizes the commuting simulation results of various types of urban center employed people, collects surrounding LBS and built environment data using information collection devices, calculates the spatiotemporal data of the commuting trajectories of urban center employed people in combination with the cloud computing platform, and converts the commuting data of the predicted commuting bottleneck areas into visualized distribution cloud maps and vector maps, which are fed back to the smart display screen.

10. The multi-agent based method for identifying commuting bottlenecks of employed people in urban centers according to claim 9, characterized in that: The step six utilizes interactive functions to make rapid decisions and optimize bottleneck areas, that is, the information collection device can collect real-time monitoring data and synchronize it to the cloud computing platform, verify the results of the prediction of the collected previous data, and realize the detailed and instant information collection and interactive feedback of the commuting bottlenecks of the commuting population in the city center. The management personnel use the PCVR equipment to make decision responses to the commuting bottleneck areas through voice commands or touch screen control, realize the prevention of commuting bottlenecks and crowd management in the city center area, and improve the commuting efficiency of the employed population in the city center.

Images