Airport ride-hailing passenger waiting time prediction method, information guidance method and system

Abstract

The application relates to the technical field of airport land-side traffic management and control, and discloses an airport online car-hailing passenger waiting time prediction method, an information guiding method and a system. The information guiding method is cyclically operated according to a preset period, passenger and driver waiting times of two types of connection modes are calculated after a connection system basic data set is acquired, and a double-connection system joint operation state is determined; a user equilibrium shunting ratio under no information guiding is solved, a candidate guiding intensity set is generated in combination with operation state constraints, equilibrium shunting ratios corresponding to each candidate value are calculated through perception cost correction, and the guiding intensity is determined and a scheme is published with the optimal comprehensive operation cost as the target. Therefore, under the premise of improving the rationality of online car-hailing passenger waiting time prediction, the passenger flow distribution and collaborative operation effect of the double-connection system are optimized, which helps to relieve the congestion pressure of the online car-hailing connection area, and provides reference support for fine management and control of airport land-side connection.

Description

Technical Field

[0001] This invention relates to the field of airport landside traffic control technology, specifically a method for predicting passenger waiting time for airport ride-hailing services, an information guidance method for joint airport taxi and ride-hailing services, and an information guidance system for joint airport taxi and ride-hailing services. Background Technology

[0002] With the continuous expansion of civil aviation transportation and the gradual maturation of the ride-hailing industry, the combined connection of taxis and ride-hailing services is gradually becoming a common connection pattern for landside transportation at large airports. The operational efficiency of the connection system significantly affects the passenger travel experience and the overall operational order of the hub. Based on the two main directions of demand forecasting and operational management, existing technologies have roughly formed two technical routes.

[0003] In the field of demand forecasting, existing technologies have conducted considerable research on the quantitative forecasting of airport connection demand. For example, the patent application with application number 202311839945.8, entitled "A Method for Airport Taxi Demand Forecasting," employs a Vector Autoregression (VAR) model to forecast airport taxi demand. By jointly modeling the time series of airport taxi trips, ride-hailing trips, and arriving passenger numbers, it effectively incorporates the impact of ride-hailing services on traditional taxi demand, capturing the linear dependencies and dynamic feedback effects among multiple variables. Such solutions typically do not rely on passenger personal information, making data acquisition relatively easy, and can output taxi demand forecasting results under the premise of clearly understanding the competitive relationship between the two modes. However, most of these technologies focus on forecasting the total travel demand, with less emphasis on refined modeling of passenger waiting times and less consideration of the non-linear growth characteristics of congestion caused by vehicle accumulation in ride-hailing connection areas. This makes it difficult to output quantitative waiting time results that can directly support passenger travel decisions. Furthermore, their technological boundaries largely remain at the demand forecasting level, and a collaborative management and information guidance mechanism for dual-mode connection systems has not yet been formed.

[0004] In the field of ride-hailing operation and management, there are already many vehicle dispatch optimization schemes based on spatiotemporal prediction in existing technologies. For example, the patent application scheme with application number 202410684423.3 and titled "Global Supply and Demand Balance Dispatch Optimization Method for Ride-Hailing Based on Spatiotemporal Demand Prediction" divides the ride-hailing operation area into spatial grids, uses a two-layer ensemble learning model to predict the spatiotemporal order demand of each sub-area, and further constructs a multi-objective integer programming model that includes driving distance, driver income, supply and demand differences, and passenger waiting time. Combined with a path search algorithm, it generates a global vehicle dispatch scheme, which can optimize the regional supply and demand matching efficiency while taking into account the interests of the platform, drivers, and passengers. However, such solutions are mostly designed for urban wide-area operation scenarios. The core idea is usually to match supply and demand through cross-regional vehicle dispatch, and rarely carry out special optimization for the closed and concentrated scenario of airport landside connection. They usually do not distinguish the two-stage waiting structure of ride-hailing "online matching waiting + offline connection waiting", nor have they established a joint operation status judgment system for taxi and ride-hailing dual connection systems. They also find it difficult to achieve the global optimization of the overall system operation cost by guiding and controlling the diversion ratio through information guidance for passengers.

[0005] In summary, most existing technologies can only achieve single-dimensional total demand forecasting or are only applicable to wide-area ride-hailing vehicle dispatching scenarios. They often struggle to simultaneously address key issues such as accurate quantification of waiting times for dual-mode connections in airport scenarios, determination of joint operation status boundaries, and quantitative optimization of information guidance intensity based on passenger diversion. Consequently, they fail to fully meet the actual needs of refined management of joint connections between taxis and ride-hailing vehicles on the airport landside. Furthermore, most existing solutions do not start from the traffic flow operation mechanism to characterize the nonlinear relationship between vehicle accumulation, traffic speed, and waiting time in ride-hailing connection areas. They also lack a linkage mechanism between waiting time prediction results and physical control actions in connection areas. This makes it difficult to directly translate prediction results into technical means of congestion mitigation, and fails to improve the collaborative efficiency of dual-connection systems from the perspective of traffic flow operation. Summary of the Invention

[0006] To address the technical problem that existing airport shuttle forecasting methods only focus on total demand and fail to consider the non-linear growth characteristics of congestion caused by vehicle accumulation in ride-hailing transfer areas, resulting in insufficient accuracy in waiting time prediction and an inability to support pre-emptive congestion control in transfer areas, this invention provides a method for predicting passenger waiting times for airport ride-hailing services. To address the technical problem that existing technologies lack a dual-system traffic flow coordination and control mechanism for closed airport shuttle scenarios, and lack quantitative information guidance and physical control linkage measures, making it difficult to achieve optimal overall traffic efficiency in transfer areas, this invention provides an information guidance method and system for joint airport taxi and ride-hailing shuttle services.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A method for predicting passenger waiting time for airport ride-hailing services includes the following steps: By utilizing vehicle detection units in the ride-hailing pick-up area deployed on the airport's landside, data exchange interfaces with ride-hailing platforms, and arrival passenger flow monitoring units, the k-th execution time period I is obtained. k The demand rate λ for ride-hailing passengers within the country p,R (k,x), available ride-hailing vehicle supply rate λ d,R (k) and predicted vehicle accumulation in ride-hailing pick-up areas And use this to calculate I k Total waiting time W for ride-hailing passengers p,R (k,x): ; ; Where x represents the proportion of taxis diverted among passengers choosing the taxi-ride-hailing combined shuttle service; W m (k,x) represents the online matching waiting time for ride-hailing passengers; ΔT represents the waiting time for offline pick-up and drop-off of ride-hailing services; ΔT represents the matching time interval of the ride-hailing platform. The ratio of the ride-hailing supply gap formed per unit of time. For I k The accumulated backlog of passengers who have not been matched with ride-hailing services. The average virtual waiting time for ride-hailing passengers due to a shortage of ride-hailing supply; ; Where t0 is the average pick-up time for ride-hailing services; N C denoted as the critical vehicle accumulation volume in the free-flow area of ​​the ride-hailing pick-up zone; L is the effective operating length of the ride-hailing pick-up zone; v(·) represents the relationship function between the vehicle accumulation volume and the operating speed of the ride-hailing pick-up zone; θ represents the congestion time scale parameter; β represents the congestion sensitivity parameter. The additional travel delay caused by a decrease in driving speed after the number of vehicles in the ride-hailing pick-up area exceeds a critical value; The relative saturation rate of vehicles in the ride-hailing pick-up area; Additional congestion and delays caused by vehicle queuing and weaving in the ride-hailing pick-up area; W p,R (k,x) is output to the airport information release terminal and / or the ride-hailing platform dispatch terminal to generate connection prompt information or trigger the ride-hailing connection area congestion warning signal; the congestion warning signal is used to drive the ride-hailing connection area entrance signal light control unit to adjust the vehicle release interval, and / or link the ride-hailing platform dispatch terminal to limit the order dispatch rate in the airport area.

[0008] As a further improvement to the above scheme, the process for obtaining the predicted vehicle accumulation in the ride-hailing pick-up area is as follows: Select the ride-hailing passenger demand rate λ p,R (k,x) and the available ride-hailing vehicle supply rate λ d,R The smaller value in (k) is used as the vehicle flow rate q entering the ride-hailing pick-up area after completing online matching. R (k,x); Based on vehicle flow rate q R (k,x), construct a calculation model for predicting the accumulation of vehicles in the ride-hailing pick-up area: ; in, , These represent the predicted vehicle accumulation in the ride-hailing pick-up area at the r-th and r+1-th iterations, respectively. The predicted vehicle accumulation volume for ride-hailing pick-up areas is The waiting time for ride-hailing services to pick up passengers at the designated location; Based on the current actual number of vehicles accumulated in the ride-hailing pick-up area or q R (k,x) t0 represents the initial vehicle accumulation volume in the ride-hailing pick-up area. The computational model is iteratively calculated; when the difference between the predicted vehicle accumulation in the ride-hailing pick-up area obtained from two adjacent iterations is less than or equal to the preset convergence accuracy, the iteration stops, and the predicted vehicle accumulation in the ride-hailing pick-up area obtained from the last iteration is determined as the final predicted vehicle accumulation in the ride-hailing pick-up area. .

[0009] As a further improvement to the above scheme: the ride-hailing passenger demand rate λ p,R The formula for calculating (k,x) is: ; ; Among them, Q p (k) represents the execution period I. k Total demand of arriving passengers choosing a combined taxi-ride-hailing service; For execution period I k Total demand rate of arriving passengers choosing the taxi-ride-hailing combined shuttle service; Δt is the duration of a single execution period; Available ride-hailing supply rate λ d,R The process of obtaining (k) is as follows: Get execution period I k The number of valid ride-hailing vehicles arriving in the ride-hailing pick-up area within the previous U consecutive historical execution periods is denoted as N for the number of valid ride-hailing vehicles arriving in the u-th historical time window. R (ku); Calculate the execution period I according to the following formula. k Available ride-hailing supply rate λ d,R (k): ; in, , is the weighting coefficient for the effective ride-hailing arrivals corresponding to the u-th historical execution period.

[0010] An information guidance method for airport taxi and ride-hailing joint connections operates cyclically according to a preset update cycle, and performs the following steps in each execution period: The basic datasets of the taxi shuttle system and the ride-hailing shuttle system during the execution period were obtained by using traffic monitoring equipment deployed on the landside of the airport, taxi pick-up area counting units, and ride-hailing platform data interfaces. Using the basic dataset as input and combining the diversion ratio, we calculate the waiting time for taxi passengers, the waiting time for taxi drivers, the waiting time for ride-hailing passengers, and the waiting time for ride-hailing drivers. The waiting time for ride-hailing passengers is calculated using the airport ride-hailing passenger waiting time prediction method. Based on the basic dataset and the calculated waiting time parameters, taxi supply constraints and ride-hailing online matching and offline connection constraints are constructed to determine the joint operation status of the taxi-ride-hailing joint connection system. The generalized travel cost is calculated based on the passenger waiting time of the two types of connection methods. The user equilibrium diversion ratio under the condition of no information guidance is solved. Based on this, combined with the joint constraints of the joint operation state, a candidate information guidance set is generated within the preset information guidance intensity range. The passenger perceived equilibrium diversion ratio corresponding to each candidate information guidance value is calculated one by one. Based on the perceived balanced diversion ratio of each passenger, the actual waiting time is calculated back, the comprehensive operating cost of the joint connection system is obtained, and after the joint operation status is verified, the candidate information guidance value that minimizes the comprehensive operating cost is selected as the target information guidance intensity, and the connection information guidance scheme is generated and released. The shuttle information guidance plan will be released through airport display screens, mobile service terminals, on-site broadcasts and / or ride-hailing platforms to guide passengers to choose their own shuttle methods and to regulate the passenger flow allocation ratio between the two systems. When the next execution period arrives, update the base dataset and repeat the above steps.

[0011] As a further improvement to the above scheme: the basic dataset includes dynamic running parameters and static calibration parameters; Dynamic operating parameters include the total demand rate for arriving passenger transfers during the execution period. Taxi supply arrival rate λ d,T(k) Available ride-hailing vehicle supply rate λ d,R (k), and the initial vehicle accumulation in the ride-hailing pick-up area. ; Static calibration parameters include taxi pick-up area service parameters, ride-hailing pick-up area physical parameters, passenger selection parameters, and information guidance and management parameters.

[0012] As a further improvement to the above scheme: Execution Period I k The waiting time for ride-hailing drivers within the city (W) d,R The formula for calculating (k,x) is as follows: ; The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of ride-hailing services exceeded the demand, and ride-hailing drivers formed an online virtual queue. The waiting time for ride-hailing drivers consisted of the basic waiting time and the virtual queue time, with the total duration of the execution period as the upper limit. The average base waiting time for ride-hailing drivers under the fixed interval matching mechanism is the mathematical expectation of the waiting time for ride-hailing drivers within a single ride-hailing platform's matching time interval. For execution period I k The cumulative backlog of unmatched ride-hailing drivers within the period represents the total number of drivers available during the entire execution period. k Internal factors include an oversupply of ride-hailing vehicles and the total number of newly added ride-hailing drivers awaiting dispatch. For execution period I k The average number of ride-hailing drivers in the region, corresponding to the average value within a time period under the condition that the backlog increases linearly over time; The average virtual waiting time for ride-hailing drivers due to oversupply of ride-hailing vehicles; when At that time, the demand for ride-hailing services exceeded the supply, and ride-hailing drivers could obtain orders within a single matching interval, with only a basic waiting time and no virtual queue. when At that time, execution period I k If there are no ride-hailing passenger orders within the specified time period, the ride-hailing driver will not have a matching opportunity during this execution period, and the waiting time will be counted as the total execution period duration Δt. min{·} is the minimum value operation; Execution Period I k Taxi passenger waiting time W p,T The formula for calculating (k,x) is as follows: ; ; Where, λ p,T (k,x) represents the execution time period I. k Taxi passenger demand rate within the area; P wait (·) represents the Erlang-C probability function for passengers needing to queue; c represents the number of taxi pick-up lanes. The average service time for a single passenger who chooses a taxi to complete the process of placing luggage, boarding the vehicle, and the vehicle departing; The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of taxis was sufficient, and the waiting time for taxi passengers was only due to the multi-channel service queuing in the taxi pick-up area, which was calculated using the M / M / c queuing model; The remaining capacity of the taxi pick-up area, that is, the portion of the taxi pick-up area's service capacity that exceeds the demand from taxi passengers; At this time Let be the ratio of the probability of taxi passengers queuing to the remaining capacity of taxi service, and let be the average waiting time of taxi passengers when supply is sufficient. when At that time, there was a shortage of taxis. In addition to queuing for pick-up services, taxi passengers also had to wait for vehicles to be replenished. The total waiting time was the sum of the basic service queuing time and the queuing time due to the supply gap. At this time Queue time for basic services at taxi pick-up areas in scenarios where taxi supply is limited; For execution period I k The cumulative backlog of unserved taxi passengers within the period represents the total number of passengers served during the entire execution period. k The total number of passengers waiting for additional taxis due to insufficient taxi supply; For execution period I k The average number of taxi passengers in the area, corresponding to the average value within a time period under the condition that the backlog increases linearly with time; Additional waiting time for taxi passengers due to insufficient taxi supply; Execution Period I k Taxi driver waiting time W d,T The formula for calculating (k,x) is as follows: ; ; The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of taxis exceeds the release capacity, and a queue of vehicles forms in the vehicle storage pool. The waiting time for taxi drivers is the average waiting time in the vehicle storage pool. For execution period I k The accumulated taxi backlog in the internal parking pool represents the total number of taxis in the entire execution period I. k The total number of newly added taxis awaiting release due to internal limitations in release capacity; For execution period I k The average taxi backlog within the period, corresponding to the average value of the backlog over a period of time under the condition that the backlog increases linearly with time; The average waiting time for taxi drivers in the holding area; when When the supply of taxis is less than the release capacity, taxis can be released directly into the passenger pick-up area upon arrival, with no vehicles backing up in the holding pool and drivers waiting in line for 0 hours. q d,T (k,x) represents the execution time period I. k Effective release and processing rate of taxis within the area.

[0013] As a further improvement to the above scheme: Taxi supply constraint: If during execution period I k Inside, there is If the supply is sufficient, the taxi shuttle system is in a state of unrestricted supply; otherwise, the taxi shuttle system is in a state of restricted supply. Online ride-hailing matching and offline pick-up constraints: If during the execution period I k Inside, there is Then the ride-hailing shuttle system status will be recorded as "online matching is unrestricted and offline ride-hailing shuttle areas are not congested"; if there are Then the status of the ride-hailing shuttle system will be recorded as "online matching is unrestricted, but offline ride-hailing shuttle areas are congested"; if there are If so, the status of the ride-hailing shuttle system will be recorded as online matching restricted and offline ride-hailing shuttle area congested. Based on the constraint determination results, the taxi shuttle system is divided into two states: unrestricted supply and restricted supply. The ride-hailing shuttle system is divided into three states: unrestricted online matching and uncongested offline ride-hailing shuttle areas; unrestricted online matching and congested offline ride-hailing shuttle areas; and restricted online matching and congested offline ride-hailing shuttle areas. The two types of shuttle system states are combined in pairs to obtain a total of six joint operation states, and the corresponding single constraint combinations form joint constraints that correspond one-to-one with the joint operation states.

[0014] As a further improvement to the above scheme, the calculation process for the passenger perception balance diversion ratio is as follows: Based on passenger waiting times and corresponding fares for the two types of connecting modes, an execution period I is constructed. k The cost difference function G(k,x) for passenger travel without information guidance: ; Among them, F T For taxi fares; F R The fare is for ride-hailing services; α is the passenger time value coefficient. Find the zero point of the passenger travel cost difference function within the range of the diversion ratio values, and take the diversion ratio corresponding to the zero point as the user equilibrium diversion ratio in the case of no information guidance; if there is no zero point in the range, the user equilibrium diversion ratio is determined according to the comparison result of the generalized travel cost at the endpoint of the diversion ratio range: G(k,x) is positive when x takes the values ​​of 0 and 1, so the output user equilibrium diversion ratio is 1; otherwise it is 0. Taking the range of user traffic equalization ratios as a reference, combined with the execution period I k The joint constraints corresponding to the joint operation state within the set generate a candidate information guidance set within the preset range of information guidance intensity values. The splitting results corresponding to each candidate information guidance intensity within the set do not exceed the corresponding joint constraints. For each candidate information guidance intensity δ... m (k) is used as a correction term for passenger perceived travel cost to construct the execution period I. k Passenger perceived cost difference function : ; Find the zero point of the passenger perceived cost difference function within the range of diversion ratio values, and use the diversion ratio corresponding to the zero point as the passenger perceived equilibrium diversion ratio of the corresponding candidate information guiding value; if there is no zero point within the range of diversion ratio values, determine the passenger perceived equilibrium diversion ratio based on the comparison results of perceived generalized travel costs at the endpoints of the diversion ratio range. When x takes the value of 0 or 1, it is a positive output, and the output passenger perception equalization diversion ratio is 1; otherwise it is 0.

[0015] As a further improvement to the above scheme, the specific steps for determining the target information guidance intensity and generating and releasing the information guidance scheme are as follows: For each candidate information guidance intensity, based on its corresponding passenger perception balance diversion ratio, the actual values ​​of taxi passenger waiting time, taxi driver waiting time, ride-hailing passenger waiting time, and ride-hailing driver waiting time are recalculated. The comprehensive operating cost of the joint connection system corresponding to the guidance strength of the candidate information is calculated using the following formula. : ; Where, x m The guiding strength δ of candidate information m (k) corresponds to the passenger perception equilibrium diversion ratio; For execution period I k The internal split ratio is x m At that time, the total demand rate of arriving passengers choosing the taxi-ride-hailing combined shuttle service; W p,T (k,x m W p,R (k,x m W d,T (k,x m ) and W d,R (k,x m ) are respectively in the execution period I k The internal split ratio is x m Taxi passenger waiting time, total ride-hailing passenger waiting time, taxi driver waiting time, and ride-hailing driver waiting time; ω p ω T ω R These are the weights for passenger waiting costs, taxi driver waiting costs, and ride-hailing driver dispatch costs, respectively. Verification is performed based on the joint operation status, and candidate information guidance intensity that would cause the taxi shuttle system to enter a state of supply constraint and fail to reduce the overall operating cost is eliminated; when the difference in the overall operating cost corresponding to multiple candidate information guidance intensities is less than a preset threshold, the candidate value with the smaller information guidance intensity value is selected. Among the remaining feasible candidate values, the candidate information guidance intensity that minimizes the overall operating cost is selected as the target information guidance intensity, and a connection information guidance scheme is generated and released based on the target information guidance intensity.

[0016] An information guidance system for airport taxi and ride-hailing combined services includes: The data acquisition module is used to acquire the basic dataset of taxi and ride-hailing connection services during each execution period; The calculation and processing module is configured to execute the information guidance method for the joint connection of airport taxis and ride-hailing services; when calculating the waiting time of ride-hailing passengers, the airport ride-hailing passenger waiting time prediction method is used for the solution. The information publishing module is used to generate and publish connection information guidance schemes based on the output results of the calculation and processing module; The joint control and execution module includes a traffic light control unit at the entrance of the ride-hailing pick-up area and a taxi holding pool release control unit, which is used to adjust the operating parameters of the pick-up system according to the instructions of the calculation and processing module; The information guidance system runs in a cycle according to a preset update period, and updates the basic dataset through the data acquisition module when the next execution period arrives.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. In the airport ride-hailing passenger waiting time prediction method, the total waiting time for ride-hailing passengers is divided into two parts: online matching waiting time and offline connection waiting time. The online matching waiting time is calculated in segments based on the relative supply and demand relationship, while the offline connection waiting time introduces the correlation between the vehicle accumulation and operating speed in the ride-hailing connection area. A free-flow critical vehicle accumulation volume is set as the state boundary. The predicted vehicle accumulation volume in the ride-hailing connection area under steady state is obtained through iterative solution. This illustrates the nonlinear relationship between vehicle accumulation and waiting time in the ride-hailing connection area, improving the accuracy of waiting time prediction and enabling early identification of congestion risks in the connection area. Through the synergistic effect of information guidance and physical control, the peak vehicle accumulation in the ride-hailing connection area can be effectively reduced, the average passage speed in the connection area can be increased, and the overall passenger evacuation time can be shortened. This improves the coordinated operation efficiency of the airport landside dual connection system from the perspective of traffic flow operation.

[0018] 2. An information guidance method for the joint connection of taxis and ride-hailing services on the airport landside is proposed. Based on the waiting time calculation results of the two types of connection systems, two types of constraints are constructed: taxi supply and online matching and offline connection of ride-hailing services. The joint operation state boundary of the two systems is delineated. On this basis, the user equilibrium diversion ratio under no information guidance is solved as a benchmark. Within the capacity constraint range, a set of candidate guidance intensities is generated. The equilibrium diversion ratio corresponding to each candidate value is calculated by correcting the perceived cost. Then, the actual waiting time and the overall system operation cost are calculated back. The optimal target guidance intensity is selected to generate a guidance scheme. This helps to coordinate the passenger flow distribution of the two connection methods in the closed airport connection scenario, improve the collaborative operation effect of the two systems, and help reduce the overall system operation cost. Attached Figure Description

[0019] Figure 1 This is a flowchart of the information guidance method.

[0020] Figure 2 This is a schematic diagram of the boundary of the joint operation status of taxis and ride-hailing services.

[0021] Figure 3 This is a comparison chart of the overall system operating costs before and after information guidance under different connection requirement levels in simulation verification.

[0022] Figure 4 This is a graph showing the change in the overall operating cost of the system before and after information guidance during 48 consecutive execution periods in the simulation verification.

[0023] Figure 5 This is a comparison chart showing the number of times the ride-hailing pick-up area's operating status appeared before and after information guidance during 48 execution periods in the simulation verification. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] This embodiment addresses the technical problems in airport landside connection scenarios, where inaccurate perception of the dual-mode connection operation status and failure to consider the nonlinear characteristics of congestion in ride-hailing connection areas during waiting time prediction lead to inaccurate information guidance, exacerbated congestion in ride-hailing connection areas, and low overall passenger evacuation efficiency. It leverages physical traffic sensing equipment deployed on the airport landside, standardized data interaction interfaces, and control execution terminals. By constructing a segmented waiting time quantification model that conforms to the natural laws of traffic flow and an equilibrium diversion optimization algorithm, it outputs implementable connection information guidance schemes and collaborative control instructions. This achieves collaborative optimization of the taxi and ride-hailing dual-connection systems, ultimately alleviating congestion in ride-hailing connection areas, improving passenger evacuation efficiency, and reducing overall system operating costs.

[0026] like Figure 1 As shown, this embodiment provides an information guidance method for the joint airport taxi and ride-hailing service, wherein the information guidance system for the joint airport taxi and ride-hailing service operates cyclically according to a preset update cycle. During the k-th run, the current decision time corresponds to the k-th execution period I. k The information guidance system sequentially executes eight core steps: acquiring the basic dataset, calculating the waiting time for the two types of connection systems, identifying the operational status of the joint connection system, solving for the balanced diversion ratio of users without information guidance, generating the candidate information guidance set and calculating the perceived balanced diversion ratio, determining the target information guidance intensity, generating and publishing the information guidance scheme, and periodically updating the operational data. It then generates and publishes the execution period I. k The system implements an internal connection information guidance scheme. When the information guidance system reaches the next decision time, it updates the operational data and enters the next cycle.

[0027] I. Obtaining the basic dataset for the execution period

[0028] Information guidance system obtains execution time period I k The basic datasets for the taxi shuttle system and the ride-hailing shuttle system include dynamic operating parameters updated over time and pre-set or calibrated static calibration parameters.

[0029] (a) Dynamic operating parameters

[0030] Dynamic operating parameters include execution time period I k Total demand rate of arriving passengers within the port Taxi supply arrival rate λ d,T (k) Available ride-hailing vehicle supply rate λ d,R (k), and the initial value of the accumulated vehicle volume in the ride-hailing pick-up area. .

[0031] 1. Total demand rate for connecting arriving passengers

[0032] Data is collected by the arrival passenger flow monitoring unit, which includes high-definition video passenger flow statistics equipment and gate counting devices deployed at the terminal arrival level exit and the entrance of the ride-hailing pick-up area. It can count the passenger flow entering the landside ride-hailing pick-up area in real time. At the same time, it is linked with the airport flight information management system to predict the total demand for arrival passenger pick-up during the operation period by combining flight arrival time, passenger capacity and passenger evacuation time window.

[0033] Execution Period I k The total demand Q of arriving passengers choosing the taxi-ride-hailing combined shuttle service p (k) can be obtained through flight arrival information, passenger flow monitoring data, airport arrival passenger transfer statistics, or historical operational data. In this embodiment, the information guidance system will execute time period I. k As a forecast time window, the execution period I is estimated based on historical statistical data of scheduled flight arrival times, actual flight arrival times, flight passenger capacity, and arrival passenger evacuation times. k The total number of passengers entering the inland side ride-hailing transfer area is as follows: Record execution period I k The starting time is t k The execution period is Δt, and the termination time is t. k +Δt.

[0034] The information guidance system determines the shortest evacuation time for passengers on the i-th arriving flight from their flight to their arrival at the airport's landside ride-hailing shuttle area, based on parameters such as the airport terminal layout, the walking distance from the arrival gate to the ride-hailing shuttle area, average walking speed, and baggage claim time. and longest evacuation time Let T be the actual arrival time of the i-th arriving flight. i The evacuation time interval for passengers of this flight entering the landside ride-hailing pick-up area is as follows: .

[0035] For the i-th arriving flight, if its evacuation time interval and execution time period I k The time intervals overlap, that is, they satisfy: , Then the flight will be designated as the operation period I. k Related flights arriving in Hong Kong.

[0036] For each associated arriving flight, the information guidance system estimates the total number of arriving passengers A for that flight based on the flight's passenger capacity or aircraft type, number of seats, and load factor. i In this embodiment, it is assumed that the passengers on this flight are in the execution period I. k The proportion of ride-hailing services arriving at the airport landside pick-up area is equal to the proportion r of the overlap between the evacuation time interval and the execution time interval to the total evacuation time. i,k The calculation formula is: ; Here, max{·} represents the maximum value operation; min{·} represents the minimum value operation.

[0037] Therefore, the i-th associated flight is assigned to the execution time slot I. k The number of passengers inside is A i r i,k .

[0038] In other embodiments, the evacuation time distribution function can be further used to calculate the arrival ratio of passengers on a single flight within the execution period. The evacuation time distribution function can be determined by airport preset, simulation setting, on-site observation data, historical operation data or parameter calibration methods. For example, continuous probability distributions such as normal distribution and log-normal distribution can be used to describe the distribution law of passenger evacuation time, and the passenger arrival ratio within the corresponding period can be calculated by integration.

[0039] The information guidance system sums the passenger volume of all related arriving flights that meet the time conditions and allocates them to the execution time slot to obtain execution time slot I. k Total number of arriving passengers at the inland side ride-hailing transfer area The summation range covers all related arriving flights.

[0040] Based on historical travel mode selection data, the estimated percentage ρ of passengers entering the taxi-ride-hailing joint connection system is then calculated. k Finally, the execution period I is obtained. k Total demand of arriving passengers choosing taxi-ride-hailing combined shuttle service Execution period I k The total demand rate for inbound passenger transfers within the port is .

[0041] 2. Taxi supply arrival rate

[0042] The data is collected by the taxi pick-up area counting unit, which includes inductive loop detectors and vehicle release counting equipment deployed at the exit of the holding pool and the entrance of the pick-up area. It can count the number of empty taxis released from the holding pool to the pick-up area in real time per unit time. The weighted moving average method is used to process the historical data to obtain the taxi supply arrival rate during the execution period.

[0043] Information guidance system obtains execution time period I k The number of empty taxis released from the holding pool to the taxi pick-up area during the previous U consecutive historical execution periods is denoted as N for the number of empty taxis in the u-th historical execution period. T (ku). The execution period I is calculated using the weighted moving average method. k Taxi supply arrival rate λ d,T (k): ; in, Let be the weighting coefficient for the number of empty taxis corresponding to the u-th historical execution period, and the sum of all weighting coefficients is 1. Distance from execution period I k The more recent the historical execution period, the larger the corresponding weight coefficient value.

[0044] 3. Available ride-hailing supply rate

[0045] Data is collected through the data interaction interface of the ride-hailing platform. This interface is a standardized API data interaction interface between the airport landside traffic control system and the ride-hailing platform server. It follows a unified data communication protocol and synchronizes operational data such as the number of empty ride-hailing vehicles available for orders around the airport and the number of vehicles entering the airport at fixed intervals. At the same time, data is verified by high-definition checkpoint equipment deployed at the entrance of the ride-hailing pick-up area to remove invalid data of vehicles entering the airport.

[0046] Information guidance system obtains execution time period I k The number of valid ride-hailing vehicles arriving in the ride-hailing pick-up area within the previous U consecutive historical execution periods, denoted as N in the u-th historical execution period. R (ku). The execution period I is calculated using the weighted moving average method. k Available ride-hailing supply rate λ d,R (k): ; in, Let be the weighting coefficient of the effective ride-hailing arrival volume corresponding to the u-th historical execution period, and the sum of all weighting coefficients is 1.

[0047] Similarly, the distance from execution period I k The more recent the historical execution period, the larger the corresponding weight coefficient value.

[0048] 4. Initial vehicle accumulation in the ride-hailing pick-up area

[0049] The data is collected by the vehicle detection unit in the ride-hailing pick-up area. This detection unit includes high-definition checkpoint cameras and RFID vehicle identification equipment deployed at the entrance and exit of the ride-hailing pick-up area. By statistically analyzing the difference between the total number of vehicles entering and leaving the area within a certain period, the actual number of vehicles accumulated in the ride-hailing pick-up area is obtained in real time and used as the initial value for iterative calculation.

[0050] The information guidance system obtains the execution time period I through vehicle detection equipment in the ride-hailing pick-up area, video recognition results, or entry and exit records. k The actual number of vehicles accumulated in the ride-hailing pick-up area at the initial moment is recorded as the initial vehicle accumulation in the ride-hailing pick-up area. .

[0051] All the above dynamic operating parameters are collected through physical traffic sensing equipment: the arrival passenger flow monitoring unit uses high-definition video passenger flow statistics equipment and gate counting devices deployed at the terminal arrival exit and transfer area entrance, with a sampling frequency of 1 time / minute; the taxi pick-up area counting unit uses ground loop detectors and vehicle release counting equipment deployed at the holding pool exit and pick-up area entrance; the ride-hailing transfer area vehicle detection unit uses high-definition checkpoint cameras and RFID radio frequency identification devices deployed at the transfer area entrance and exit, with a sampling frequency of no less than 1 time / 10 seconds; all collected data are transmitted in real time to the industrial computing server of the control center through the airport industrial Ethernet to ensure that the data source is traceable and verifiable.

[0052] (ii) Static calibration parameters

[0053] The static calibration parameters are all obtained based on on-site airport surveys and historical operational data, which is a routine technical calibration task in the field of transportation engineering. The service parameters of the taxi pick-up area and the physical parameters of the ride-hailing pick-up area were obtained by fitting the data through on-site measurements and continuous operation data observation for more than 7 days. Passenger selection parameters and information guidance management parameters are obtained through historical travel behavior statistics and iterative optimization and calibration of control effects.

[0054] All static parameters can be determined by those skilled in the art using conventional technical means in conjunction with the actual scenario of the target airport, without requiring creative effort.

[0055] Static calibration parameters include taxi pick-up area service parameters, ride-hailing pick-up area physical parameters, passenger selection parameters, and information guidance and management parameters.

[0056] Taxi pick-up area service parameters: number of taxi pick-up lanes (c), average service time for a single passenger to complete luggage placement, boarding, and vehicle departure. .

[0057] Physical parameters of ride-hailing pick-up areas: matching time interval ΔT between ride-hailing platforms, and critical vehicle accumulation N in the free flow of the ride-hailing pick-up area. C The following parameters are defined: average ride-hailing pick-up time t0 under low load conditions, effective operating length L of ride-hailing pick-up area, relationship function v(·) between vehicle accumulation in ride-hailing pick-up area and operating speed in ride-hailing pick-up area, congestion time scale parameter θ, and congestion sensitivity parameter β.

[0058] Passenger selection parameters: Taxi fare F T Online ride-hailing fare F R Passenger time value coefficient α.

[0059] Information guidance management parameters: Passenger waiting cost weight ω p Taxi driver waiting cost weight ω T Weighting of ride-hailing driver waiting-for-dispatch costs ω R The upper limit of the information guidance intensity value.

[0060] The information guidance system integrates the above dynamic operating parameters with static calibration parameters to form execution period I. k The basic dataset is used as input for subsequent steps of waiting time calculation, running status identification, and guidance strength calculation.

[0061] II. Waiting times for the two types of connection systems

[0062] All waiting time calculations in this chapter are executed on the industrial computing server (computation processing module) of the airport control center. All quantitative models are based on the classic queuing theory and natural laws of fluid dynamics in the field of traffic engineering. They are quantitative descriptions of the real traffic operation mechanism in the ride-hailing pick-up area, aiming to solve the technical problem that waiting time cannot be accurately predicted manually in real scenarios, rather than abstract mathematical calculations.

[0063] The information guidance system takes the basic dataset as input and, combined with the given passenger taxi selection ratio x, calculates the execution time period I. k The data includes taxi passenger waiting time, taxi driver waiting time, ride-hailing passenger waiting time, and ride-hailing driver waiting time. The proportion of passengers choosing ride-hailing services is 1-x.

[0064] 1. Passenger demand rate

[0065] Execution Period I k Taxi passenger demand rate λ p,TThe formula for calculating (k,x) is: ; Execution Period I k The demand rate λ for ride-hailing passengers within the country p,R The formula for calculating (k,x) is: ; The demand rates for the two types of passengers are used for subsequent waiting time calculations.

[0066] 2. Ride-hailing passenger waiting time

[0067] To accurately depict the two-stage real waiting process of ride-hailing passengers—"initiating order matching online → getting into the ride-hailing pick-up area offline"—this study combines the traffic operation characteristics of the ride-hailing pick-up area to break down the total waiting time of ride-hailing passengers into two parts: online matching waiting time and offline pick-up waiting time, and constructs quantitative calculation models for each.

[0068] Total waiting time W for ride-hailing passengers p,R (k,x) is the waiting time W determined by online matching of ride-hailing passengers. m (k,x) represents the waiting time for offline pick-up of ride-hailing services. It consists of two parts, and the calculation formula is: ; in, For execution period I k Predict the vehicle accumulation volume in the ride-hailing pick-up area under a given passenger diversion ratio x.

[0069] (1) Online matching waiting time for ride-hailing passengers

[0070] Online matching waiting time for ride-hailing passengers W m (k,x) is calculated piecewise based on the relative relationship between the demand rate of ride-hailing passengers and the supply rate of available ride-hailing vehicles. The formula is as follows: ; The physical meanings of each parameter, segment, and combination term in the formula are as follows: ΔT: The time interval for matching rides on ride-hailing platforms.

[0071] ΔT / 2: Average basic matching waiting time for ride-hailing passengers under the fixed interval matching mechanism, in minutes; since the time when passengers initiate requests is evenly distributed between two matching cycles, the mathematical expectation of the waiting time is 1 / 2 of the matching interval, which is the inherent basic waiting time brought about by the matching mechanism.

[0072] The ride-hailing supply gap rate formed per unit of time, expressed in people / min, represents the rate at which demand exceeds capacity.

[0073] Execution Period I k The cumulative backlog of unmatched passengers, expressed in person, represents the total number of newly added passengers awaiting matching due to insufficient supply during the entire execution period.

[0074] The average number of ride-hailing passengers during the execution period, expressed in person. Since the backlog increases linearly from 0 to the maximum value at the end of the period, the average backlog during the period is half of the value at the end of the period, which is a standard processing method for approximating fluid queues.

[0075] The average virtual waiting time for ride-hailing passengers due to supply gaps, in minutes; the average backlog divided by the capacity processing rate gives the average waiting time for a single ride-hailing passenger.

[0076] Online matching waiting time for ride-hailing passengers W m The calculation principle of (k,x) is as follows: Real ride-hailing platforms do not use an instant matching model. Instead, they set a fixed matching interval, collect all passenger calls and nearby available vehicles within that interval, and then perform the matching. Therefore, after a passenger sends a ride request, there will be a base waiting time with an average / expected value of 0.5ΔT.

[0077] When demand is lower than supply, i.e. Matching can be completed within a single matching interval, at which point the online waiting time for ride-hailing passengers is the base waiting time: 0.5ΔT.

[0078] Conversely, in addition to the basic waiting time, unmatched passengers will form a virtual queue online, increasing the matching rate. The waiting time in this virtual queue is characterized using a fluid queue. Specifically, the ride-hailing supply gap rate formed per unit time is... During the execution period In this situation, the backlog of passengers who have not been matched with a car is Since the backlog increases approximately linearly from zero to the ending backlog during the execution period, the average backlog during the execution period is taken as half of the ending backlog. The flow rate for handling the backlog of vehicles is currently at [value missing]. Get the virtual queue waiting time Furthermore, when passenger demand for transportation surges, leading to insufficient capacity, the online waiting time for ride-hailing passengers is expressed as... .

[0079] (2) Predicted vehicle accumulation in ride-hailing pick-up areas

[0080] There is an implicit correlation between the accumulation of vehicles in the ride-hailing pick-up area and the pick-up waiting time, which conforms to the natural law of Little's law in queuing theory. In order to solve the vehicle accumulation in steady state, the fixed point iteration method is used for calculation. This method is a conventional technique in the field of traffic engineering for solving nonlinear steady-state equations.

[0081] The information guidance system first determines the flow rate q of vehicles that have completed online matching and entered the ride-hailing pick-up area. R (k,x) represents the smaller value between the demand rate of ride-hailing passengers and the supply rate of available ride-hailing vehicles. ; Under steady-state average conditions (within the execution period, without considering instantaneous changes in the number of vehicles, and estimating vehicle accumulation using the average vehicle flow rate and offline connection time within the period), the predicted vehicle accumulation in the ride-hailing connection area satisfies the following constraint (computation model): the predicted vehicle accumulation is equal to the product of the vehicle flow rate and the offline connection waiting function, i.e.: ; in, , These represent the predicted vehicle accumulation in the ride-hailing pick-up area at the r-th and r+1-th iterations, respectively.

[0082] The information guidance system uses any one of the following methods—fixed-point iteration, bisection method, or Newton's iteration method—to solve the above calculation model and obtain the predicted vehicle accumulation in the ride-hailing pick-up area. The specific process is as follows: Select an initial value for the iteration; the initial value can be the initial number of vehicles accumulated in the current ride-hailing pick-up area. Or the product of vehicle flow rate and average connection time under low load. Iterative calculations are performed according to the computational model.

[0083] When the difference between the predicted vehicle accumulation in the ride-hailing connection area obtained from two consecutive iterations is less than or equal to the preset convergence accuracy When, that is, satisfied Stop the iteration and determine the vehicle accumulation amount obtained in the last iteration as the final predicted vehicle accumulation amount for the ride-hailing pick-up area. .

[0084] (3) Waiting time for ride-hailing offline pick-up

[0085] Online ride-hailing offline pick-up waiting time The calculation is performed piecewise based on the relative relationship between the predicted vehicle accumulation volume in the ride-hailing pick-up area and the critical vehicle accumulation volume in free flow. The formula is as follows: ; The physical meanings of each parameter, segment, and combination term in the formula are as follows: t0: Baseline connection time in free-flow state, in minutes, including the base time for the entire process of ride-hailing vehicle entering, stopping to pick up passengers, and leaving, excluding any additional delays due to congestion.

[0086] The additional travel delay caused by the decrease in driving speed after the accumulated number of ride-hailing vehicles exceeds the critical value, in minutes; the travel time corresponding to the current accumulated number of ride-hailing vehicles is subtracted from the free-flow travel time under the critical state, and the free-flow travel portion already included in the base time is deducted to avoid double counting.

[0087] The relative saturation rate of vehicles in the ride-hailing pick-up area is a dimensionless value that represents the proportion of ride-hailing vehicles exceeding the critical capacity, reflecting the degree of congestion in the ride-hailing pick-up area.

[0088] : Additional congestion delay caused by congestion interplay, in minutes; The nonlinear characteristics of congestion deterioration are characterized by a power function. The higher the saturation rate, the faster the delay increases. Essentially, it is a scenario-based transformation of the BPR delay function in traffic engineering. The coefficient θ transforms the dimensionless relative saturation rate into a delay value with a time dimension, representing the magnitude of the congestion intensity.

[0089] Online ride-hailing offline pick-up waiting time The calculation principle is as follows: The waiting time at the ride-hailing pick-up area includes two parts: the base waiting time and the additional congestion waiting time. ; The baseline waiting time is the free-flow connection time t0. Additional waiting time includes the decrease in operating speed due to vehicle accumulation and delays caused by congestion.

[0090] When the predicted vehicle accumulation in the ride-hailing pick-up area does not exceed the free-flow critical vehicle accumulation, that is... Ride-hailing vehicles can maintain free flow within the ride-hailing pick-up and drop-off area, and the average time required for a vehicle to go from entering to picking up a passenger and leaving remains at t0, without additional congestion waiting time, i.e., the corresponding time increment. At this time, the waiting time for ride-hailing offline pick-up is expressed as... .

[0091] When the predicted vehicle accumulation in the ride-hailing pick-up area exceeds the free-flow critical vehicle accumulation, that is... When vehicles slow down in the ride-hailing pick-up area, additional congestion and delays occur due to factors such as vehicles weaving together and temporary stops. In this case, in addition to the base waiting time, there is also an additional congestion waiting time. The additional time caused by the decrease in vehicle speed is: To avoid double-counting the base runtime, this extra time needs to be subtracted. Additional delays caused by vehicle queuing and weaving are expressed as follows: ,in This study characterizes the relative saturation level of the current predicted vehicle accumulation exceeding the capacity of the ride-hailing pick-up area. Considering that the additional delay increases non-linearly with the saturation level of the ride-hailing pick-up area, a congestion sensitivity parameter β is introduced to characterize the sensitivity of congestion delay to vehicle accumulation. This dimensionless saturation level is transformed into time using a congestion time parameter θ, thus characterizing the intensity of the impact of congestion on delay.

[0092] Using piecewise functions allows for calculations to be performed according to different actual situations, resulting in higher efficiency.

[0093] 3. Waiting time for taxis

[0094] (1) Taxi passenger waiting time

[0095] First, calculate the taxi pick-up area service capacity μ. T This refers to the number of passengers that the boarding area can serve per unit of time. ; Taxi passenger waiting time W p,T (k,x) is calculated based on the M / M / c queuing model, combined with the taxi supply arrival rate λ. d,T (k) and taxi passenger demand rate λ p,T The relative relationship between (k,x) is solved piecewise, and the formula is: ; The physical meanings of each parameter, each segment, and the combination term in the formula are as follows: λ p,T (k,x) represents the execution time period I. k Taxi passenger demand flow rate when the internal and diversion ratio is x, in person / min; P wait (·) represents the Erlang-C probability function for passengers needing to queue; c represents the number of taxi pick-up lanes. The average service time for a single passenger who chooses a taxi to complete the process of placing luggage, boarding the vehicle, and the vehicle departing, in minutes per person; The maximum service capacity of the taxi pick-up area is expressed in people / min, representing the maximum number of passengers that the pick-up area can serve per unit of time. when At that time, the supply of taxis was sufficient, and the waiting time for taxi passengers was only due to the multi-channel service queuing in the taxi pick-up area, which was calculated using the M / M / c queuing model; The remaining capacity of the taxi pick-up area, measured in people / min, which is the portion of the service capacity that exceeds the demand of taxi passengers; At this time is the ratio of queuing probability to remaining taxi service capacity, and is the average queuing time for taxi passengers when supply is sufficient, in minutes. At this time, the passenger pick-up area input flow rate is determined by the taxi supply rate. when At that time, there was a shortage of taxis. In addition to queuing for pick-up services, taxi passengers also had to wait for vehicles to be replenished. The total waiting time was the sum of the basic service queuing time and the queuing time due to the supply gap. At this time The queuing time for basic services in the taxi pick-up area under the scenario of limited taxi supply is expressed in minutes. At this time, the input flow rate of the taxi pick-up area is determined by the taxi supply availability rate. For execution period I k The cumulative backlog of unserved taxi passengers within the period, expressed in person, representing the total number of passengers served during the entire execution period. k The total number of passengers waiting for additional taxis due to insufficient taxi supply; For execution period I k The average number of taxi passengers in the area, in person, is the average value within a time period under the condition that the backlog increases linearly with time. The additional waiting time for taxi passengers due to insufficient taxi supply is expressed in minutes.

[0096] Taxi passenger waiting time W p,T The calculation principle of (k,x) is as follows: Since the number of taxi pick-up lanes is usually limited, passengers experience a basic waiting time during the pick-up process. This mainly involves the multi-lane service process of queuing to board, placing luggage, and the vehicle departing. This part can be characterized using the M / M / C queuing model. This is used to study situations where the taxi pick-up area is in a controllable operating state during a given time period.

[0097] When the taxi supply rate can meet the passenger demand rate, that is... At this point, the number of taxis available can meet the passenger demand, and passengers do not need to wait for a vehicle to arrive; the waiting time is the basic waiting time. .

[0098] According to queuing theory, we know that: ; In the formula, n represents the channel number, and n! represents the factorial of n.

[0099] It should be noted that the above queuing model calculation applies to situations where the taxi pick-up area is in a controllable operational state during the execution period. A controllable operational state means that there is no continuous queue overflow, lane congestion, or vehicle backlog in the taxi pick-up area, and the on-site dispatcher can reflect factors such as luggage placement, boarding, vehicle departure, and minor traffic friction in the average service time through pick-up lane organization, vehicle release control, and passenger queue management. In this scenario, the service capacity of the taxi pick-up area can be represented by [the capacity of the taxi pick-up area], and the service queue waiting time for passengers in the pick-up area is calculated using the M / M / c queuing model.

[0100] When supply cannot meet passenger demand, that is Passenger waiting is further affected by the insufficient supply of empty taxis, which is characterized using a fluid queue model. Specifically, the taxi supply gap formed per unit time is... During the execution period In this situation, the backlog of taxi passengers who haven't waited for a taxi is Since the backlog increases approximately linearly from zero to the final backlog during the execution period, the system takes the average backlog during the execution period as half of the final backlog. With supply rate λ d,T (k) represents the flow rate for handling backlogged passengers, resulting in the virtual queue waiting time. .

[0101] (2) Taxi driver waiting time

[0102] First, calculate the effective taxi release rate q. d,T (k,x) represents the effective release capacity of the taxi shuttle system for vehicles in the parking lot, which is the smaller value between the taxi passenger demand rate and the service capacity of the pick-up area. ; in, This indicates the maximum passenger pick-up rate that the taxi pick-up area can achieve under sufficient passenger demand. When the passenger pick-up area is open, the effective taxi release rate is determined by the service capacity of the passenger pick-up area; conversely, when the passenger pick-up area is open, the effective taxi release rate is determined by passenger demand.

[0103] λ p,T (k,x) and The comparison is essentially a comparison between vehicle supply arrival rate and vehicle release service rate. Parameters Although described as "average service time per passenger," its actual service scenario in the airport taxi pick-up area is the total time for a single taxi to complete a single passenger pick-up service (including the complete cycle of luggage placement, boarding, and vehicle departure for that group of passengers). Therefore Essentially, it represents the vehicle service rate (vehicles / unit of time) in the passenger pick-up area, which is numerically equivalent to the passenger flow rate under the assumption of "one vehicle, one batch of service". Taxi supply arrival rate λ d,T (k) represents the vehicle flow rate (vehicles / unit time). Under the assumption of single-vehicle, single-batch service, the number of empty taxis arriving per unit time is equivalent to the number of passenger service batches that can be handled per unit time, and is related to the effective release processing rate q. d,T The numerical magnitude and physical dimensions of (k,x) are completely unified.

[0104] Therefore λ d,T (k) and q d,T The comparison of (k,x) essentially compares the arrival rate of empty taxis with the vehicle release rate at the exit of the parking pool to determine whether there is a vehicle backlog in the parking pool.

[0105] Taxi driver waiting time W d,T (k,x) is calculated in segments based on the relative relationship between taxi supply arrival rate and effective release processing rate, using the following formula: ; The physical meanings of each parameter, segment, and combination term in the formula are as follows: when At that time, the supply of taxis exceeds the release capacity, and a queue of vehicles forms in the vehicle storage pool. The waiting time for taxi drivers is the average waiting time in the vehicle storage pool. For execution period I k The accumulated number of taxis in the internal parking pool, expressed in vehicles, represents the total number of taxis in the entire execution period. k The total number of newly added taxis awaiting release due to internal limitations in release capacity; For execution period I k The average number of taxis in the area, in units of vehicles, is the average value over a period of time under the condition that the backlog increases linearly with time. This is the average queuing time for taxi drivers in the holding area, expressed in minutes. This time only refers to the time spent queuing for vehicles to be released and does not include the time spent picking up passengers.

[0106] when When the supply of taxis is less than the release capacity, taxis can be released directly into the passenger pick-up area upon arrival, with no vehicles backing up in the holding pool and drivers waiting in line for 0 hours. q d,T (k,x) represents the execution time period I. k Effective release and processing rate of taxis within the area.

[0107] Taxi driver waiting time W d,TThe calculation principle of (k,x) is as follows: Similarly, using piecewise functions allows for calculations to be performed according to different actual situations, resulting in higher timeliness.

[0108] when At that time, the taxi supply rate should not be lower than the effective release rate; taxis exceeding the effective release rate will result in a vehicle backlog in the holding pool. During the execution period length In this situation, the vehicle backlog is Since the backlog increases approximately linearly from zero to the final backlog during the execution period, the system takes the average backlog during the execution period as half of the final backlog. . with q d,T (k,x) represents the flow rate for handling backlogged passengers, and the driver waiting time at this time is expressed as: .

[0109] when When the supply of taxis is less than the release capacity, vehicles can be released directly into the pick-up area upon arrival, with no vehicles accumulating in the storage pool and drivers waiting in line for 0 hours.

[0110] 4. Ride-hailing driver waiting time

[0111] Ride-hailing driver waiting time W d,R (k,x) is calculated piecewise based on the relative relationship between the demand rate of ride-hailing passengers and the supply rate of available ride-hailing vehicles. The formula is as follows: ; The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of ride-hailing services exceeded the demand, and ride-hailing drivers formed an online virtual queue. The waiting time for ride-hailing drivers consisted of the basic waiting time and the virtual queue time, with the total duration of the execution period as the upper limit. The average base waiting time for ride-hailing drivers under the fixed interval matching mechanism is expressed in minutes, which corresponds to the mathematical expectation of the waiting time for ride-hailing drivers within a single ride-hailing platform matching time interval. For execution period I k The cumulative number of unmatched ride-hailing drivers within the period, expressed in vehicles, represents the total number of unmatched drivers during the entire execution period. k The internal factor of oversupply is the total number of newly added ride-hailing drivers awaiting dispatch. For execution period I k The average number of ride-hailing drivers in the region, in units of vehicles, is the average value within a time period under the condition that the backlog increases linearly over time. The average virtual waiting time for ride-hailing drivers due to oversupply is expressed in minutes. when At that time, the demand for ride-hailing services exceeded the supply, and ride-hailing drivers could obtain orders within a single matching interval, with only a basic waiting time and no virtual queue. when At that time, execution period I k If there are no ride-hailing passenger orders within the specified time period, the ride-hailing driver will not have a matching opportunity during this execution period, and the waiting time will be counted as the total execution period duration Δt.

[0112] Ride-hailing driver waiting time W d,R The calculation principle of (k,x) is as follows: The time cost for ride-hailing drivers providing airport shuttle services corresponds to the waiting time from when the driver enters the dispatchable state until an order is received. Entering the ride-hailing shuttle area is considered part of the driver's service process and is not included in the driver's waiting time. Therefore, the driver's waiting time is symmetrical to the online matching waiting time for ride-hailing passengers. Ride-hailing drivers also have a base waiting time with an average / expected value of 0.5ΔT within the platform matching interval.

[0113] Specifically, when passenger demand for transportation exceeds the supply of ride-hailing services, that is... For ride-hailing drivers, matching can be completed within a single matching interval, at which point the driver's waiting time is... .

[0114] Conversely, when In addition to the basic waiting time, unmatched drivers will also form a virtual queue online, with a matching rate of λ. p,R (k,x) is used to characterize the waiting time of this virtual queue using a fluid queue. Specifically, the passenger demand gap formed per unit time is... During the execution period In this case, the backlog of unmatched drivers is Since the backlog increases approximately linearly from zero to the ending backlog during the execution period, the average backlog during the execution period is taken as half of the ending backlog. , with λ p,R (k,x) is used as the flow rate for handling backlogged drivers, and the virtual queue waiting time is obtained. Therefore, the waiting time for ride-hailing drivers at this time is expressed as .

[0115] when If there are no ride-hailing passenger orders in the current execution period, the waiting time for ride-hailing drivers in that execution period is counted as the execution period length Δt, and the length is recalculated after the demand rate is updated in the next execution period.

[0116] The information guidance system saves the calculated taxi passenger waiting time, taxi driver waiting time, ride-hailing passenger waiting time, and ride-hailing driver waiting time as execution period I. k The waiting time parameter is input into the subsequent steps of joint operation status identification, balanced diversion ratio calculation and guidance intensity optimization.

[0117] III. Identification of the Operational Status of the Joint Connection System

[0118] Based on the basic dataset and the calculated waiting time parameters, the system constructs taxi supply constraints, ride-hailing online matching constraints, and ride-hailing offline connection constraints, and determines the joint operation status of the taxi-ride-hailing joint connection system accordingly.

[0119] 1. Two types of constraint boundary determination

[0120] (1) Taxi supply constraints

[0121] If during execution period I k Internal satisfaction: ; If the supply is not limited, the taxi shuttle system is determined to be in a state of unrestricted supply; otherwise, the taxi shuttle system is determined to be in a state of limited supply.

[0122] (2) Constraints on online matching and offline connection for ride-hailing services

[0123] If during execution period I k Internal satisfaction: ; The status of the ride-hailing connection system will be recorded as "online matching is unrestricted and offline ride-hailing connection areas are not congested".

[0124] If during execution period I k Internal satisfaction: ; The status of the ride-hailing shuttle system will then be recorded as "unrestricted online matching but congested offline ride-hailing shuttle areas"; If during execution period I k Internal satisfaction: ; The ride-hailing system status will then be recorded as online matching restricted and offline ride-hailing pick-up areas congested.

[0125] The physical essence of the online matching and offline connection constraints for ride-hailing services is to determine whether the vehicle flow rate entering the ride-hailing connection area exceeds the free-flow critical vehicle flow rate. Using passenger demand flow rate directly in the comparison is a common simplified assumption based on the airport ride-hailing scenario: The default assumption of the scenario is that a single ride-hailing vehicle provides a single pick-up service for one passenger (or one batch of passengers). After online matching is completed, each passenger's demand corresponds to one ride-hailing vehicle being dispatched to the pick-up area. Therefore, the passenger flow rate that is successfully matched is completely equivalent to the vehicle flow rate that enters the pick-up area. The unification of dimensions can be achieved through "one-to-one correspondence between people and vehicles".

[0126] Logical correspondence: When the supply of ride-hailing vehicles is sufficient (online matching is unrestricted), the vehicle flow rate entering the ride-hailing pick-up area is the ride-hailing passenger demand rate λ. p,R (k,x) and the available ride-hailing vehicle supply rate λ d,R The smaller value in (k) indicates that the passenger demand flow rate directly determines the vehicle inflow rate, therefore λ can be directly used. p,R (k,x) and critical vehicle flow rate N C The / t0 comparison is a simplified expression of engineering modeling; when the supply is insufficient, it corresponds to the "online matching is limited" state, and the vehicle flow rate is determined by the supply side, which is consistent with the subsequent judgment logic.

[0127] 2. Combined Operational Status

[0128] Based on the determination results of the above two types of constraints, the taxi shuttle system is divided into two states: unrestricted supply and restricted supply. The ride-hailing shuttle system is divided into three states: unrestricted online matching and uncongested offline ride-hailing shuttle areas; unrestricted online matching and congested offline ride-hailing shuttle areas; and restricted online matching and congested offline ride-hailing shuttle areas. For example... Figure 2 As shown, with the proportion of passengers choosing taxis (x) as the horizontal axis and the execution period I as the vertical axis, k Total demand rate of domestic arriving passengers Within the plane with the vertical axis, combining the constraints of taxi supply, ride-hailing offline connection, and ride-hailing online matching, the taxi-ride-hailing joint connection system can be divided into six joint operation states, providing boundary basis for the generation and verification of information guidance intensity.

[0129] The six joint operation states are as follows: Status 1: Taxi supply is unrestricted, online ride-hailing matching is unrestricted, and offline ride-hailing pick-up areas are not congested; Status 2: Taxi supply is unrestricted, online ride-hailing matching is unrestricted, and offline ride-hailing pick-up areas are congested; Status 3: Taxi supply is not limited, online matching for ride-hailing services is limited, and offline ride-hailing pick-up areas are congested; Status 4: Taxi supply is limited, online ride-hailing matching is unrestricted, and offline ride-hailing pick-up areas are not congested; Status 5: Taxi supply is limited, online ride-hailing matching is unrestricted, and offline ride-hailing pick-up areas are congested; Status 6: Taxi supply is limited, online ride-hailing matching is limited, and offline ride-hailing pick-up areas are congested.

[0130] Each joint operation state corresponds to a set of constraint conditions, forming a joint constraint that corresponds one-to-one with the joint operation state, providing boundary basis for the generation and verification of subsequent candidate information guidance sets. When the passenger diversion ratio x=0 for taxis, the taxi shuttle system is directly determined to be in an empty operation state; when the passenger diversion ratio x=1 for taxis, the ride-hailing shuttle system is directly determined to be in an empty operation state and included in the joint operation state combination result.

[0131] IV. User Distribution Ratio in the Absence of Information Guidance

[0132] The information guidance system calculates the generalized travel cost based on passenger waiting times for the two types of connecting modes, and solves for the user equilibrium diversion ratio in the absence of information guidance, which serves as the benchmark for subsequent information guidance calculations.

[0133] 1. Passenger Generalized Travel Cost Difference Function

[0134] Based on passenger waiting times and corresponding fares for the two types of connecting modes, an execution period I is constructed. k The cost difference function G(k,x) for passenger travel without information guidance: ; Among them, F T For taxi fares; F R α represents the fare for ride-hailing services; α represents the passenger's time value coefficient.

[0135] Specifically, the generalized travel cost for passengers choosing taxis is expressed as: ; The generalized cost of travel for ride-hailing passengers is expressed as: ; Subtracting the generalized travel costs of ride-hailing passengers from those of taxi passengers yields the passenger travel cost difference function under no-information guidance.

[0136] 2. User traffic distribution ratio

[0137] The information guidance system identifies the zero point of the passenger travel cost difference function within the diversion ratio range x∈[0,1]. The diversion ratio satisfying G(k,x)=0 is taken as the user equilibrium diversion ratio in the case of no information guidance. The solution can be obtained using any of the following methods: bisection method, iterative search method, or Newton's iteration method. Iteration stops and the user equilibrium diversion ratio is output when the difference between two adjacent iterations is less than a preset precision, or the absolute value of the cost difference function is less than a preset threshold. If no zero point exists within the diversion ratio range, the boundary equilibrium diversion ratio is determined based on the comparison of generalized travel costs at the endpoints of the range. If G(k,x) takes positive values ​​at x=0 and x=1, then the user distribution ratio under the condition of no information guidance is 1. If G(k,x) takes negative values ​​at x=0 and x=1, then the user distribution ratio under the condition of no information guidance is 0.

[0138] V. Candidate Information Guidance Set and Perceptual Balance Diversion Ratio

[0139] The information guidance system combines the joint constraints corresponding to the joint operation status to generate a set of candidate information guidance within the preset information guidance intensity range, and calculates the passenger perception balance diversion ratio corresponding to each candidate information guidance value.

[0140] 1. Candidate Information Guidance Set

[0141] Using the user traffic distribution ratio without guidance as a baseline state reference (as a baseline state, combined with state analysis, identify problems in the absence of guidance information and determine the necessity of guidance (whether it reduces system costs)), combined with the execution period I k The joint constraints corresponding to the internal joint operation state generate a candidate information guidance set containing several candidate information guidance intensities within a preset range of information guidance intensity values. The passenger diversion results corresponding to each candidate information guidance intensity within the set should avoid significantly exacerbating the limited taxi supply, limited online matching of ride-hailing services, or congestion in offline ride-hailing pick-up areas. Let the m-th candidate information guidance intensity be δ. m (k), the guiding strength of all candidate information satisfies , where δ max This is the preset upper limit for information guidance intensity.

[0142] 2. Passenger perceived cost difference function

[0143] Guiding strength δ for each candidate information m (k) is used as a correction term for passenger perceived travel cost to construct the execution period I. k Passenger perceived cost difference function : ; The intensity of information guidance affects passengers' perceived waiting time for ride-hailing services, but does not change the actual waiting time within the system. This is because, in airport landside shuttle scenarios, the queuing status of taxis in the pick-up area is usually easier for arriving passengers to observe directly. In contrast, due to factors such as ride-hailing vehicles entering the pick-up area, congestion within the pick-up area, merging, and uncertainty regarding pick-up locations, passengers find it difficult to accurately and promptly perceive the actual waiting time of the ride-hailing system upon exiting the station. Therefore, under information guidance, the perceived travel cost for ride-hailing passengers is: ; The difference function of perceived travel cost under information guidance is obtained by subtracting the perceived travel cost of ride-hailing passengers from that of taxi passengers.

[0144] The information guidance system will find the zero point of the passenger perceived cost difference function in the interval x∈[0,1] of the diversion ratio range, which will satisfy... The diversion ratio is used as the passenger perceived equilibrium diversion ratio corresponding to the candidate information guiding value. If there is no zero point within the value range of the diversion ratio, the boundary perceived equilibrium diversion ratio is determined based on the comparison results of perceived generalized travel costs at the endpoints of the range. like If the values ​​at x=0 and x=1 are both positive, then the passenger perception balance diversion ratio corresponding to this candidate information guidance value is 1. like If the values ​​at x=0 and x=1 are both negative, then the passenger perception balance diversion ratio corresponding to this candidate information guidance value is 0.

[0145] The information guidance system repeats the above calculation for all candidate values ​​within the candidate information guidance set to obtain the passenger perception equilibrium diversion ratio corresponding to the guidance intensity of each candidate information, denoted as x. m .

[0146] VI. Intensity of Target Information Guidance

[0147] The information guidance system calculates the actual waiting time based on the perceived balanced diversion ratio of each passenger, obtains the comprehensive operating cost of the joint connection system, and selects the candidate information guidance value that minimizes the comprehensive operating cost as the target information guidance intensity after combining the joint operation status verification.

[0148] 1. Actual waiting time

[0149] Guiding strength δ for each candidate information m (k), based on its corresponding passenger perception equilibrium diversion ratio x m According to the calculation method in Part II of this embodiment, the execution period I is recalculated. kThe actual values ​​of taxi passenger waiting time, taxi driver waiting time, ride-hailing passenger waiting time, and ride-hailing driver waiting time are denoted as W. p,T (k,x m W p,R (k,x m W d,T (k,x m ) and W d,R (k,x m ).

[0150] 2. Overall system operating cost

[0151] The comprehensive operating cost of the joint connection system corresponding to the guidance strength of the candidate information is calculated using the following formula. : ; in, For execution period I k The internal split ratio is x m At that time, the total demand rate of arriving passengers choosing the taxi-ride-hailing combined shuttle service.

[0152] The operational pressure on the system is not limited to passenger waiting areas. Using only passenger waiting time as the optimization metric might lead to the system directing too many passengers to one mode of transport, thus increasing the waiting time for drivers on the other side. Conversely, using only vehicle waiting time as the metric ignores the efficiency of arriving passenger departures. Therefore, from the airport management perspective, passenger waiting costs, taxi driver waiting costs, and ride-hailing driver waiting costs are all included in the system's evaluation metrics. Since passenger waiting, taxi driver waiting, and ride-hailing driver waiting have different levels of importance to airport operational support, weights are assigned to each. , , .

[0153] 3. Candidate Information Guidance Strength Verification

[0154] The information guidance system verifies the guidance strength of all candidate information based on the joint operation status: For candidate information guidance strengths that lead to a supply-constrained state in the taxi shuttle system and do not reduce the overall operating cost of the system, they are removed from the feasible candidate set; When the difference in the overall system operating cost corresponding to the guidance strength of multiple candidate information is less than a preset threshold, the candidate with the smaller guidance strength value is selected to reduce unnecessary demand shifts and maintain the stability of information release content between adjacent time periods.

[0155] 4. Select target information guidance intensity

[0156] Among the remaining feasible candidate values, the candidate information guidance strength that minimizes the overall system operating cost is selected as the execution period I. k The target information guidance intensity. The information guidance system synchronously outputs the balanced diversion ratio corresponding to the target information guidance intensity, the actual waiting time of the two connection methods, and the joint operation status, and inputs the subsequent information guidance scheme generation steps.

[0157] VII. Information Guidance Plan Generation and Release

[0158] The information guidance system generates execution period I based on the target information guidance intensity, the corresponding balanced traffic distribution ratio, the waiting time of the two connection methods, and the joint operation status. k The system will provide and publish a shuttle service information guidance plan. This plan includes estimated taxi waiting times, estimated ride-hailing waiting times, congestion status of ride-hailing pick-up areas, recommended shuttle methods, and one or more of the following: information release time and effective information period. The effective information period and execution period are defined in section I. k Correspondingly, or pre-set by airport managers according to airport information display rules.

[0159] The shuttle information guidance plan is disseminated through one or more channels, including airport display screens, mobile web pages, electronic directional displays, voice announcements, on-site staff prompts, or platform-coordinated prompts. The information guidance system simultaneously records the execution period. k The information guidance plan that has been released includes the target information guidance intensity, the corresponding balanced diversion ratio, the recommended connection method, the information release content, the release channel, and the corresponding joint operation status.

[0160] In addition to information dissemination to passengers, the calculation results of this method can also be directly output to airport landside traffic control execution equipment, forming a coordinated control action at the physical level and realizing a complete technical closed loop from quantitative calculation to operational optimization. Ride-hailing pick-up area congestion control: When the predicted vehicle accumulation in the ride-hailing pick-up area exceeds the preset level 1 congestion threshold, the system sends a flow restriction instruction to the traffic light control unit at the entrance of the ride-hailing pick-up area to dynamically adjust the entry interval of ride-hailing vehicles and reduce the inflow rate of vehicles in the ride-hailing pick-up area; when it exceeds the level 2 congestion threshold, the system will coordinate with the ride-hailing platform dispatch terminal to temporarily restrict the dispatch of orders in the airport area to prevent further deterioration of congestion.

[0161] Taxi capacity coordinated dispatch: When it is determined that the taxi connection system is in a state of supply constraint, and there are dispatchable taxis in the storage pool, the system sends an instruction to open more channels to the taxi storage pool release control system to increase the taxi release rate and supplement the connection capacity; when the taxi storage pool is severely backlogged, the linkage information release terminal guides passengers to prioritize taxis, digests the backlog of vehicles in the storage pool, and balances the operating pressure on both sides.

[0162] Real-time triggering of early warning signals: When the predicted passenger waiting time or driver waiting time exceeds the preset early warning threshold, the system automatically triggers on-site broadcasts and guidance screens to issue early warning prompts, and at the same time sends alarm information to the airport operation and maintenance management terminal to prompt management personnel to intervene in advance to guide traffic.

[0163] Specifically, for example, when the predicted vehicle accumulation in the ride-hailing pick-up area exceeds the first-level congestion threshold, the system automatically issues a flow restriction instruction to the entrance traffic light control unit, increasing the entry release interval by 20% from the baseline value to reduce the vehicle inflow rate. When the level 2 congestion threshold is exceeded, the ride-hailing platform's dispatch interface will be activated simultaneously to temporarily limit the number of orders dispatched in the core area of ​​the airport to prevent vehicles from continuing to flood in and exacerbating congestion. When the taxi shuttle system is determined to be in a state of supply constraint for more than one execution period, the system will automatically trigger the command to open more lanes in the storage pool, increasing the number of release lanes from the baseline to the maximum available number, thereby speeding up the replenishment of transport capacity.

[0164] All of the above instructions and signals directly affect the actual traffic flow on the airport landside through physical hardware devices, which can actually change the vehicle operation status and passenger diversion ratio in the ride-hailing transfer area, ultimately achieving the technical effect of alleviating transfer congestion and improving evacuation efficiency.

[0165] 8. Periodic Update of Operational Data

[0166] When the information guidance system reaches the next decision time, the information guidance system stops using execution period I. k The running data generates new guidance results, and the system begins to read the running status quantities corresponding to the next execution period, updates the basic dataset, and repeats the above steps in a loop.

[0167] When updating operational data, the information guidance system obtains the updated total demand for arriving passenger transfers, taxi supply arrival rate, available ride-hailing vehicle supply rate, and vehicle accumulation in the ride-hailing transfer area. It also reads the actual diversion results and published information guidance plans from the previous execution period. If some dynamic operational parameters are missing, they are supplemented using the most recent valid values ​​from the same data source, historical data from similar time periods, or preset values ​​from the information guidance system. If the data exceeds the preset reasonable range, it is corrected or removed according to the preset data verification rules.

[0168] The information guidance system updates the execution time period number k to k+1, uses the updated basic dataset as the new input, returns to the basic dataset acquisition step, and generates the information guidance scheme for the next execution time period.

[0169] IX. Simulation Verification

[0170] To verify the actual effect of the airport taxi and ride-hailing joint connection information guidance method in this embodiment, a numerical simulation environment was constructed to conduct comparative tests. By analyzing the system operation indicators under two scenarios—no information guidance and information guidance—the optimization effect and scenario adaptability of the method were verified.

[0171] (a) Simulation settings and baseline parameters

[0172] This simulation sets the information guidance update cycle to Δt=15min and constructs 48 consecutive execution periods to simulate the rolling information guidance process of the airport landside taxi-ride-hailing joint connection system during a typical continuous operation period. Within each execution period, the system regenerates the basic dataset, calculates the waiting time and joint operation status of the two connection methods, solves the balanced diversion ratio under no guidance and information guidance scenarios, and selects the target information guidance intensity with the best overall system operating cost.

[0173] The baseline parameters for the simulation experiments were constructed with reference to publicly available statistical data. They are mainly used for constructing typical scenarios and verifying methodological mechanisms, and do not represent the precise operational status of a specific airport. The baseline parameter settings are shown in the table below: Table 1 Basic parameters for numerical simulation

[0174] Table 1 shows the free-flow velocity v in the ride-hailing pick-up area. f The speed of the ride-hailing shuttle area when the vehicle accumulation is 0 is the calibration parameter of the speed function v.

[0175] (II) Overall cost optimization effect under different demand levels

[0176] To verify the applicability of the method under different demand intensities, based on the baseline supply conditions of the first execution period, three connection demand scenarios were set up: low, medium, and high, corresponding to total passenger connection demand rates of 34 people / min, 42 people / min, and 50 people / min, respectively. Under the condition that the taxi supply rate, ride-hailing supply rate, physical parameters of the ride-hailing connection area, and passenger selection parameters remain unchanged, the overall system operating costs with and without information guidance were compared. The results are as follows: Figure 3 As shown.

[0177] Simulation results show that, under all three demand scenarios, the overall system operating cost after information guidance is lower than that without information guidance. When the demand level is 34 people / min, the overall system operating cost decreases from 300.31 to 201.87, a reduction of approximately 32.8%; when the demand level is 42 people / min, the overall system operating cost decreases from 246.53 to 145.19, a reduction of approximately 41.1%; and when the demand level is 50 people / min, the overall system operating cost decreases from 236.34 to 125.86, a reduction of approximately 46.7%.

[0178] The overall system operating cost of this invention is a weighted average of three parts: passenger waiting cost, taxi driver waiting cost, and ride-hailing driver waiting cost, simultaneously covering two dimensions: passenger travel efficiency and capacity utilization efficiency. Under fixed baseline supply parameters, when total demand for connections is low, both taxi and ride-hailing capacity are significantly surplus, resulting in long waiting times for drivers and a high proportion of driver waiting costs in the overall cost. As total demand for connections gradually increases, passenger waiting time increases accordingly, leading to a corresponding rise in passenger waiting costs. However, the increased demand simultaneously improves the capacity utilization rate of both types of connection systems, significantly reducing driver waiting times and resulting in a significant decrease in driver waiting costs. Under the weighting and baseline parameters set in this simulation, the decrease in driver waiting costs is greater than the increase in passenger waiting costs. Therefore, the overall system operating cost shows a downward trend as connection demand increases, a pattern consistent with the actual operating characteristics of airport connection systems: "high capacity idleness and losses during low-demand periods, and improved capacity utilization efficiency after demand increases."

[0179] The results show that, under different passenger demand intensities, the information guidance method proposed in this embodiment can optimize passenger flow allocation in the dual-connection system by adjusting the passenger diversion ratio, thereby improving the overall system operation. Furthermore, the cost optimization effect of the method becomes more significant as demand levels increase, effectively alleviating the pressure on the connection system in high-demand scenarios. It should be noted that the horizontal axis of this test represents different connection demand levels set within a single time period, used to verify the method's scenario adaptability, and not different time periods in a continuous rolling simulation.

[0180] (III) Overall cost performance under multi-period rolling operation

[0181] To further verify the applicability of the method in dynamic operating scenarios, a rolling simulation environment with 48 consecutive execution periods was constructed, each period lasting 15 minutes, corresponding to a total continuous operation of 12 hours. Within each execution period, the system re-executes the basic dataset update process: predicting arrival and transfer demand based on the flight arrival and evacuation time windows of the current period, estimating taxi and ride-hailing supply rates based on historical window vehicle arrival volumes, and then re-completing waiting time calculation, joint operation status identification, balanced diversion ratio solution, and target information guidance intensity determination.

[0182] The changes in the overall system operating cost before and after information guidance within 48 time periods are as follows: Figure 4 As shown in the figure. The results show that during continuous rolling operation, the overall system operating cost after information guidance is lower than that without information guidance, indicating that this method can adapt to dynamically changing supply and demand scenarios and continuously play an optimization role during multi-period rolling updates.

[0183] (iv) Improvement effect on congestion in ride-hailing pick-up areas

[0184] The number of times different operating states of the ride-hailing pick-up area occurred before and after information guidance was statistically analyzed over 48 rolling time periods. The results are as follows: Figure 5 As shown in the figure. Statistical results show that without information guidance, ride-hailing services experienced 39 instances of offline congestion and 9 instances of no congestion; after information guidance was implemented, the number of offline congestion instances decreased to 37, and the number of no-congestion instances increased to 11.

[0185] The reason for the above changes is that, without information guidance, passengers spontaneously choose their connecting methods based solely on their own experience, which easily leads to a higher proportion of connecting demand on the ride-hailing side. This results in an increase in the number of vehicles accumulating in the ride-hailing connecting area, causing it to be in a state of congestion for a long time. After implementing the information guidance method in this embodiment, the system corrects the perceived travel cost of passengers through the intensity of information guidance, objectively presents the impact of waiting costs caused by congestion in the ride-hailing connecting area, guides some passengers to choose taxi connecting methods, reduces the vehicle inflow rate on the ride-hailing side, thereby reducing the frequency of congestion in the ride-hailing connecting area and improving the overall operation order of the dual connecting system.

[0186] In summary, the simulation results show that the information guidance method proposed in this embodiment can dynamically adjust the balanced diversion ratio between taxis and ride-hailing vehicles according to different passenger demand levels and the joint system operation status. This helps to reduce the overall operating cost of the system, alleviate congestion in the ride-hailing transfer area, and improve the overall operation status of the airport landside connection system.

[0187] 10. Practical Application

[0188] This method, through calculations of target information guidance intensity, passenger perception balance diversion ratio, predicted waiting times for the two types of connection systems, and joint operation status determination results, does not simply output numerical values. Instead, it directly serves as the decision-making basis for the airport landside connection intelligent management and control system, and is applied to multi-dimensional actual management and control aspects, achieving a complete technical closed loop from quantitative calculation to operational optimization. In practical applications, this includes the following two aspects: On the one hand, based on the above calculation results, the management body can release guidance information to arriving passengers through multiple channels such as terminal electronic guidance screens, passenger travel service terminals, on-site broadcasts, and collaborative push from ride-hailing platforms. This information includes the estimated waiting time for the two types of connecting methods, the congestion status of the ride-hailing connecting area, and recommended connecting methods. This quantitative information can correct passengers' perception of travel costs, guide passengers to divert themselves, and avoid overload of passengers due to a single connecting method caused by information asymmetry. On the other hand, airport operators can combine waiting time predictions with joint operational status to coordinate the taxi waiting pool release system and the ride-hailing capacity dispatch platform to dynamically adjust the taxi release rate and the scale of ride-hailing vehicles entering the airport. This allows them to supplement capacity during periods of insufficient supply and control vehicle inflow during periods of congestion, achieving coordinated dispatch of the two types of connecting transport capacity. At the same time, ride-hailing transfer area managers can identify congestion risks in advance based on the predicted vehicle accumulation and take proactive control measures such as opening temporary lanes and adjusting passenger flow to suppress the non-linear deterioration of congestion in the ride-hailing transfer area.

[0189] In addition, the multi-scenario operational data accumulated over a long period of time by this method can provide quantitative support for expanding the capacity of ride-hailing pick-up areas, optimizing the layout of passenger pick-up channels, and formulating capacity guarantee mechanisms, thus serving the medium- and long-term planning and upgrading of airport landside pick-up systems.

[0190] Through the above-mentioned practical applications, this method can effectively reduce the overall travel costs for passengers, reduce the ineffective consumption of transportation resources, alleviate traffic congestion in ride-hailing transfer areas, and effectively solve the practical problem of insufficient collaborative management of dual-mode transfers at large airports.

[0191] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for predicting passenger waiting time for airport ride-hailing services, characterized in that, Includes the following steps: By utilizing vehicle detection units in the ride-hailing pick-up area deployed on the airport's landside, data exchange interfaces with ride-hailing platforms, and arrival passenger flow monitoring units, the k-th execution time period I is obtained. k The demand rate λ for ride-hailing passengers within the country p,R (k,x), available ride-hailing vehicle supply rate λ d,R (k) and predicted vehicle accumulation in ride-hailing pick-up areas And use this to calculate I k Total waiting time W for ride-hailing passengers p,R (k,x): ； ； Where x represents the proportion of taxis diverted among passengers choosing the taxi-ride-hailing combined shuttle service; W m (k,x) represents the online matching waiting time for ride-hailing passengers; ΔT represents the waiting time for offline pick-up and drop-off of ride-hailing services; ΔT represents the matching time interval of the ride-hailing platform. The ratio of the ride-hailing supply gap formed per unit of time. For I k The accumulated backlog of passengers who have not been matched with ride-hailing services. The average virtual waiting time for ride-hailing passengers due to a shortage of ride-hailing supply; ； Where t0 is the average pick-up time for ride-hailing services; N C denoted as the critical vehicle accumulation volume in the free-flow area of ​​the ride-hailing pick-up zone; L is the effective operating length of the ride-hailing pick-up zone; v(·) represents the relationship function between the vehicle accumulation volume and the operating speed of the ride-hailing pick-up zone; θ represents the congestion time scale parameter; β represents the congestion sensitivity parameter. The additional travel delay caused by a decrease in driving speed after the number of vehicles in the ride-hailing pick-up area exceeds a critical value; The relative saturation rate of vehicles in the ride-hailing pick-up area; Additional congestion and delays caused by vehicle queuing and weaving in the ride-hailing pick-up area; W p,R (k,x) is output to the airport information release terminal and / or the ride-hailing platform dispatch terminal to generate connection prompt information or trigger the ride-hailing connection area congestion warning signal; the congestion warning signal is used to drive the ride-hailing connection area entrance signal light control unit to adjust the vehicle release interval, and / or link the ride-hailing platform dispatch terminal to limit the order dispatch rate in the airport area.

2. The method for predicting passenger waiting time for airport ride-hailing services according to claim 1, characterized in that, The process for obtaining the predicted vehicle accumulation in the ride-hailing pick-up area is as follows: Select the ride-hailing passenger demand rate λ p,R (k,x) and the available ride-hailing vehicle supply rate λ d,R The smaller value in (k) is used as the vehicle flow rate q entering the ride-hailing pick-up area after completing online matching. R (k,x); Based on vehicle flow rate q R (k,x), construct a calculation model for predicting the accumulation of vehicles in the ride-hailing pick-up area: ； in, , These represent the predicted vehicle accumulation in the ride-hailing pick-up area at the r-th and r+1-th iterations, respectively. The predicted vehicle accumulation volume for ride-hailing pick-up areas is The waiting time for ride-hailing services to pick up passengers at the designated location; Based on the current actual number of vehicles accumulated in the ride-hailing pick-up area or q R (k,x) t0 represents the initial vehicle accumulation volume in the ride-hailing pick-up area. The computational model is iteratively calculated; when the difference between the predicted vehicle accumulation in the ride-hailing pick-up area obtained from two adjacent iterations is less than or equal to the preset convergence accuracy, the iteration stops, and the predicted vehicle accumulation in the ride-hailing pick-up area obtained from the last iteration is determined as the final predicted vehicle accumulation in the ride-hailing pick-up area. .

3. The method for predicting passenger waiting time for airport ride-hailing services according to claim 1, characterized in that, Ride-hailing passenger demand rate λ p,R The formula for calculating (k,x) is: ； ； Among them, Q p (k) represents the execution period I. k Total demand of arriving passengers choosing a combined taxi-ride-hailing service; For execution period I k Total demand rate of arriving passengers choosing the taxi-ride-hailing combined shuttle service; Δt is the duration of a single execution period; Available ride-hailing supply rate λ d,R The process of obtaining (k) is as follows: Get execution period I k The number of valid ride-hailing vehicles arriving in the ride-hailing pick-up area within the previous U consecutive historical execution periods is denoted as N for the number of valid ride-hailing vehicles arriving in the u-th historical time window. R (ku); Calculate the execution period I according to the following formula. k Available ride-hailing supply rate λ d,R (k): ； in, , is the weighting coefficient for the effective ride-hailing arrivals corresponding to the u-th historical execution period.

4. A method for providing information guidance for the combined airport taxi and ride-hailing service, characterized in that, It runs in a loop according to a preset update cycle, and performs the following steps in each execution period: The basic datasets of the taxi shuttle system and the ride-hailing shuttle system during the execution period were obtained by using traffic monitoring equipment deployed on the landside of the airport, taxi pick-up area counting units, and ride-hailing platform data interfaces. Using the basic dataset as input and combining the diversion ratio, the waiting time for taxi passengers, the waiting time for taxi drivers to be dispatched, the waiting time for ride-hailing passengers, and the waiting time for ride-hailing drivers to be dispatched are calculated respectively; wherein the waiting time for ride-hailing passengers is calculated using the airport ride-hailing passenger waiting time prediction method according to any one of claims 1-3. Based on the basic dataset and the calculated waiting time parameters, taxi supply constraints and ride-hailing online matching and offline connection constraints are constructed to determine the joint operation status of the taxi-ride-hailing joint connection system. The generalized travel cost is calculated based on the passenger waiting time of the two types of connection methods. The user equilibrium diversion ratio under the condition of no information guidance is solved. Based on this, combined with the joint constraints of the joint operation state, a candidate information guidance set is generated within the preset information guidance intensity range. The passenger perceived equilibrium diversion ratio corresponding to each candidate information guidance value is calculated one by one. Based on the perceived balanced diversion ratio of each passenger, the actual waiting time is calculated back, the comprehensive operating cost of the joint connection system is obtained, and after the joint operation status is verified, the candidate information guidance value that minimizes the comprehensive operating cost is selected as the target information guidance intensity, and the connection information guidance scheme is generated and released. The shuttle information guidance plan will be released through airport display screens, mobile service terminals, on-site broadcasts and / or ride-hailing platforms to guide passengers to choose their own shuttle methods and to regulate the passenger flow allocation ratio between the two systems. When the next execution period arrives, update the base dataset and repeat the above steps.

5. The information guidance method for airport taxi and ride-hailing combined transportation according to claim 4, characterized in that, The basic dataset includes dynamic runtime parameters and static calibration parameters; Dynamic operating parameters include the total demand rate for arriving passenger transfers during the execution period. Taxi supply arrival rate λ d,T (k) Available ride-hailing vehicle supply rate λ d,R (k), and the initial vehicle accumulation in the ride-hailing pick-up area. ; Static calibration parameters include taxi pick-up area service parameters, ride-hailing pick-up area physical parameters, passenger selection parameters, and information guidance and management parameters.

6. The information guidance method for airport taxi and ride-hailing combined transportation according to claim 4, characterized in that, Execution Period I k The waiting time for ride-hailing drivers within the city (W) d,R The formula for calculating (k,x) is as follows: ； The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of ride-hailing services exceeded the demand, and ride-hailing drivers formed an online virtual queue. The waiting time for ride-hailing drivers consisted of the basic waiting time and the virtual queue time, with the total duration of the execution period as the upper limit. The average base waiting time for ride-hailing drivers under the fixed interval matching mechanism is the mathematical expectation of the waiting time for ride-hailing drivers within a single ride-hailing platform's matching time interval. For execution period I k The cumulative backlog of unmatched ride-hailing drivers within the period represents the total number of drivers available during the entire execution period. k Internal factors include an oversupply of ride-hailing vehicles and the total number of newly added ride-hailing drivers awaiting dispatch. For execution period I k The average number of ride-hailing drivers in the region, corresponding to the average value within a time period under the condition that the backlog increases linearly over time; The average virtual waiting time for ride-hailing drivers due to oversupply of ride-hailing vehicles; when At that time, the demand for ride-hailing services exceeded the supply, and ride-hailing drivers could obtain orders within a single matching interval, with only a basic waiting time and no virtual queue. when At that time, execution period I k If there are no ride-hailing passenger orders within the specified time period, the ride-hailing driver will not have a matching opportunity during this execution period, and the waiting time will be counted as the total execution period duration Δt. min{·} is the minimum value operation; Execution Period I k Taxi passenger waiting time W p,T The formula for calculating (k,x) is as follows: ； ； Where, λ p,T (k,x) represents the execution time period I. k Taxi passenger demand rate within the area; P wait (·) represents the Erlang-C probability function for passengers needing to queue; c represents the number of taxi pick-up lanes. The average service time for a single passenger who chooses a taxi to complete the process of placing luggage, boarding the vehicle, and the vehicle departing; The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of taxis was sufficient, and the waiting time for taxi passengers was only due to the multi-channel service queuing in the taxi pick-up area, which was calculated using the M / M / c queuing model; The remaining capacity of the taxi pick-up area, that is, the portion of the taxi pick-up area's service capacity that exceeds the demand from taxi passengers; At this time Let be the ratio of the probability of taxi passengers queuing to the remaining capacity of taxi service, and let be the average waiting time of taxi passengers when supply is sufficient. when At that time, there was a shortage of taxis. In addition to queuing for pick-up services, taxi passengers also had to wait for vehicles to be replenished. The total waiting time was the sum of the basic service queuing time and the queuing time due to the supply gap. At this time Queue time for basic services at taxi pick-up areas in scenarios where taxi supply is limited; For execution period I k The cumulative backlog of unserved taxi passengers within the period represents the total number of passengers served during the entire execution period. k The total number of passengers waiting for additional taxis is due to insufficient taxi supply. For execution period I k The average number of taxi passengers in the area, corresponding to the average value within a time period under the condition that the backlog increases linearly with time; Additional waiting time for taxi passengers due to insufficient taxi supply; Execution Period I k Taxi driver waiting time W d,T The formula for calculating (k,x) is as follows: ； ； The physical meaning of each segment and the combination term in the formula is as follows: when At that time, the supply of taxis exceeds the release capacity, and a queue of vehicles forms in the vehicle storage pool. The waiting time for taxi drivers is the average waiting time in the vehicle storage pool. For execution period I k The accumulated taxi backlog in the internal parking pool represents the total number of taxis in the entire execution period I. k The total number of newly added taxis awaiting release due to internal limitations in release capacity; For execution period I k The average taxi backlog within the period, corresponding to the average value of the backlog over a period of time under the condition that the backlog increases linearly with time; The average waiting time for taxi drivers in the holding area; when When the supply of taxis is less than the release capacity, taxis can be released directly into the passenger pick-up area upon arrival, with no vehicles backing up in the holding pool and drivers waiting in line for 0 hours. q d,T (k,x) represents the execution time period I. k Effective release and processing rate of taxis within the area.

7. The information guidance method for airport taxi and ride-hailing combined connection according to claim 4, characterized in that, Taxi supply constraints: If during execution period I k Inside, there is If the supply is sufficient, the taxi shuttle system is in a state of unrestricted supply; otherwise, the taxi shuttle system is in a state of restricted supply. Online ride-hailing matching and offline pick-up constraints: If during the execution period I k Inside, there is Then the ride-hailing shuttle system status will be recorded as "online matching is unrestricted and offline ride-hailing shuttle areas are not congested"; if there are Then the status of the ride-hailing shuttle system will be recorded as "online matching is unrestricted, but offline ride-hailing shuttle areas are congested"; if there are If so, the status of the ride-hailing shuttle system will be recorded as online matching restricted and offline ride-hailing shuttle area congested. Based on the constraint determination results, the taxi shuttle system is divided into two states: unrestricted supply and restricted supply. The ride-hailing shuttle system is divided into three states: unrestricted online matching and uncongested offline ride-hailing shuttle areas; unrestricted online matching and congested offline ride-hailing shuttle areas; and restricted online matching and congested offline ride-hailing shuttle areas. The two types of shuttle system states are combined in pairs to obtain a total of six joint operation states, and the corresponding single constraint combinations form joint constraints that correspond one-to-one with the joint operation states.

8. The information guidance method for airport taxi and ride-hailing combined connection according to claim 4, characterized in that, The calculation process for the passenger perceived equalization diversion ratio is as follows: Based on passenger waiting times and corresponding fares for the two types of connecting modes, an execution period I is constructed. k The cost difference function G(k,x) for passenger travel without information guidance: ； Among them, F T For taxi fares; F R The fare is for ride-hailing services; α is the passenger time value coefficient. Find the zero point of the passenger travel cost difference function within the range of the diversion ratio values, and take the diversion ratio corresponding to the zero point as the user equilibrium diversion ratio in the case of no information guidance; if there is no zero point in the range, then determine the user equilibrium diversion ratio based on the comparison results of the generalized travel costs at the endpoints of the diversion ratio range: G(k,x) is positive when x takes the values ​​of 0 and 1, then the output user equilibrium diversion ratio is 1; otherwise it is 0. Taking the range of user traffic equalization ratios as a reference, combined with the execution period I k The joint constraints corresponding to the joint operation state within the set generate a candidate information guidance set within the preset range of information guidance intensity values. The splitting results corresponding to each candidate information guidance intensity within the set do not exceed the corresponding joint constraints. For each candidate information guidance intensity δ... m (k) is used as a correction term for passenger perceived travel cost to construct the execution period I. k Passenger perceived cost difference function : ； Find the zero point of the passenger perceived cost difference function within the range of diversion ratio values, and use the diversion ratio corresponding to the zero point as the passenger perceived equilibrium diversion ratio of the corresponding candidate information guiding value; if there is no zero point within the range of diversion ratio values, determine the passenger perceived equilibrium diversion ratio based on the comparison results of perceived generalized travel costs at the endpoints of the diversion ratio range. When x takes the value of 0 or 1, it is a positive output, and the output passenger perception equalization diversion ratio is 1; otherwise it is 0.

9. The information guidance method for airport taxi and ride-hailing combined connection according to claim 4, characterized in that, The specific steps for determining the target information guidance intensity and generating and releasing the information guidance plan are as follows: For each candidate information guidance intensity, based on its corresponding passenger perception balance diversion ratio, the actual values ​​of taxi passenger waiting time, taxi driver waiting time, ride-hailing passenger waiting time, and ride-hailing driver waiting time are recalculated. The comprehensive operating cost of the joint connection system corresponding to the guidance strength of the candidate information is calculated using the following formula. : ； Where, x m The guiding strength δ of candidate information m (k) corresponds to the passenger perception equilibrium diversion ratio; For execution period I k The internal split ratio is x m At that time, the total demand rate of arriving passengers choosing the taxi-ride-hailing combined shuttle service; W p,T (k,x m W p,R (k,x m W d,T (k,x m ) and W d,R (k,x m ) are respectively in the execution period I k The internal split ratio is x m Taxi passenger waiting time, total ride-hailing passenger waiting time, taxi driver waiting time, and ride-hailing driver waiting time; ω p ω T ω R These are the weights for passenger waiting costs, taxi driver waiting costs, and ride-hailing driver dispatch costs, respectively. Verification is performed based on the joint operation status, and candidate information guidance intensity that would cause the taxi shuttle system to enter a state of supply constraint and fail to reduce the overall operating cost is eliminated; when the difference in the overall operating cost corresponding to multiple candidate information guidance intensities is less than a preset threshold, the candidate value with the smaller information guidance intensity value is selected. Among the remaining feasible candidate values, the candidate information guidance intensity that minimizes the overall operating cost is selected as the target information guidance intensity, and a connection information guidance scheme is generated and released based on the target information guidance intensity.

10. An information guidance system for the combined connection of airport taxis and ride-hailing services, characterized in that, include: The data acquisition module is used to acquire the basic dataset of taxi and ride-hailing connection services during each execution period; The calculation and processing module is configured to execute the information guidance method for airport taxi and ride-hailing joint connection as described in any one of claims 4-9; wherein when calculating the waiting time of ride-hailing passengers, the method for predicting the waiting time of airport ride-hailing passengers as described in any one of claims 1-3 is used for the calculation. The information publishing module is used to generate and publish connection information guidance schemes based on the output results of the calculation and processing module; The joint control and execution module includes a traffic light control unit at the entrance of the ride-hailing pick-up area and a taxi holding pool release control unit, which is used to adjust the operating parameters of the pick-up system according to the instructions of the calculation and processing module; The information guidance system runs in a cycle according to a preset update period, and updates the basic dataset through the data acquisition module when the next execution period arrives.

Images