High-speed rail passenger transport hub connection transport capacity configuration optimization method

Abstract

The invention provides a high-speed rail passenger transport hub connection transport capacity configuration optimization method, and belongs to the field of traffic management. The method comprises the following steps: proposing a simplified calculation formula of waiting time for connecting a subway, a bus, a taxi and an online hailed taxi in a high-speed rail passenger transport hub; a high-speed rail passenger transport hub connection transport capacity configuration optimization model based on double-layer programming is constructed, an upper layer model is a departure interval decision-making model based on multi-objective optimization, and a lower layer model is a traveler travel mode selection model based on random user balance; and a hybrid intelligent optimization algorithm is provided for solving the bilayer planning. According to the method provided by the invention, optimization of connection transport capacity configuration of the high-speed rail passenger transport hub can be realized, and a basis is provided for formulating a passenger flow evacuation scheme of the high-speed rail passenger transport hub.

Description

Technical Field

[0001] The present invention belongs to the field of traffic planning and management, and relates to a method for optimizing the connection capacity configuration of a high-speed rail passenger transport hub. Background Art

[0002] High-speed rail passenger hubs are key connecting points between high-speed rail and urban transportation networks. Common connections within hubs include subways, buses, taxis, ride-hailing services, and private cars. With the increasing number of high-speed rail passengers, the issue of passenger evacuation within hubs has attracted increasing attention. Research on optimizing capacity allocation for connecting connections within high-speed rail passenger hubs is beneficial for accurately dispatching connecting capacity and achieving rapid passenger evacuation.

[0003] High-speed rail passenger hub capacity allocation requires scientific and effective allocation within the context of limited transport resources, integrating and coordinating various connecting transport modes to maximize overall efficiency. However, generalized per capita waiting time and operating costs are two conflicting objectives. Therefore, it is necessary to comprehensively consider the generalized per capita waiting time and operating costs of connecting modes at high-speed rail passenger hubs, establish a method for calculating waiting times for different connecting transport modes, and construct a comprehensive capacity allocation optimization method for high-speed rail passenger hubs using genetic algorithms. This method optimizes and coordinates the average departure intervals of all transport modes connecting to the high-speed rail passenger hub, ultimately achieving a capacity allocation for different connecting transport modes that comprehensively considers waiting times and operating costs. Summary of the Invention

[0004] Purpose of the invention: To propose a method for optimizing the configuration of connecting transport capacity of high-speed rail passenger transport hubs, and to provide a solution for the configuration of connecting transport capacity of high-speed rail passenger transport hubs.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is a method for optimizing the connection capacity configuration of a high-speed rail passenger transport hub. The specific implementation process is as follows:

[0006] 1. A method for optimizing the connection capacity configuration of a high-speed rail passenger transport hub, characterized in that the steps of this technology are as follows:

[0007] S1: Propose a simplified calculation formula for the total waiting time of each connecting transportation mode at the high-speed rail passenger transport hub;

[0008] S2: Constructing a high-speed rail passenger transport hub connection capacity configuration optimization model based on double-layer planning;

[0009] S3: Propose a hybrid intelligent optimization algorithm to solve bi-level programming.

[0010] Said S1 comprises:

[0011] S11: When the passenger flow arrives at the same time at the beginning of the study period, the calculation formula is as follows:

[0012] S111: Propose a simplified calculation formula for the total waiting time for subways within high-speed rail passenger transport hubs

[0013] As attached Figure 1 As shown, attached Figure 1 The figure depicts the changes in the number of passengers waiting for the subway at different times. The total waiting time of the passengers who choose to take the subway is the product of the number of passengers and the waiting time of each passenger, which is the attached Figure 1 By calculating the area of ​​the graph, we can get the total waiting time T1 for taking the subway:

[0014]

[0015] Among them, t w1 It represents the average travel time of passengers from the exit to the subway station; q1 represents the average number of people carried by each subway during the study period, which is generally determined through traffic surveys; t1 represents the average departure interval of the subway during the study period, which is used to indicate the capacity allocation of the subway; Q1 represents the passenger flow that chooses to take the subway.

[0016] S112: Propose a simplified calculation formula for the total waiting time of buses in high-speed rail passenger transport hubs

[0017] As attached Figure 2 As shown, attached Figure 2 The figure depicts the changes in the number of passengers waiting for the bus at different times. The total waiting time of the passengers who choose to take the bus is the product of the number of passengers and the waiting time of each passenger, which is the additional Figure 2 By calculating the area of ​​the graph, we can get the total waiting time T2 for taking the bus:

[0018]

[0019] Among them, t w2 represents the average travel time of passengers from the exit to the bus stop; q2 represents the average number of people carried by each bus during the study period, which is generally determined through traffic surveys; t2 represents the average departure interval of buses during the study period, which is used to represent the bus capacity allocation; Q2 represents the passenger flow that chooses to take the bus.

[0020] S113: Propose a simplified calculation formula for the total taxi waiting time within high-speed rail passenger transport hubs

[0021] As attached Figure 3 As shown, attached Figure 3 The figure depicts the changes in the number of passengers waiting for taxi service at different times. Therefore, the total waiting time of passengers who choose to take a taxi is the product of the number of passengers and the waiting time of each passenger, which is the additional Figure 3By calculating the area of ​​the graph, we can get the total waiting time T3 for taking a taxi:

[0022]

[0023] Among them, t s represents the average service time of a taxi; t w3 It represents the average travel time of passengers from the exit to the taxi dispatch station; n1 represents the number of people who can immediately receive taxi service. Its value is usually determined through traffic surveys. It is necessary to count the number of people who can immediately receive taxi service when the passenger flow arrives multiple times during the study period and take the average value; Q3 represents the passenger flow that chooses to take a taxi; t3 represents the average departure interval of taxis, which is actually the inverse of the taxi arrival rate.

[0024] S114: Propose a simplified calculation formula for the total waiting time for ride-hailing services within high-speed rail passenger transport hubs

[0025] As attached Figure 4 As shown, attached Figure 4 The figure depicts the changes in the number of passengers waiting for ride-hailing at different times. The total waiting time of passengers who choose to take ride-hailing is the product of the number of passengers and the waiting time of each passenger, which is the product of the number of passengers and the waiting time of each passenger. Figure 4 By calculating the area of ​​the graph, we can get the total waiting time T4 for taking a ride-hailing car:

[0026]

[0027] Among them, t w4 It represents the average travel time of passengers from the exit to the online car-hailing dispatch station; n2 represents the number of people who can immediately receive online car-hailing services. Its value is usually determined through traffic surveys. It is necessary to count the number of people who can immediately receive online car-hailing services when passengers arrive multiple times within the study period and take the average value; Q4 represents the passenger flow that chooses to take online car-hailing; t4 represents the average departure interval of online car-hailing, which is actually the inverse of the arrival rate of online car-hailing.

[0028] S12: When the passenger flow arrives multiple times during the study period, the arrival time and Taking the arrival time as an example, the calculation method is as follows:

[0029] S121: Propose a simplified calculation formula for the total waiting time for subways within high-speed rail passenger transport hubs

[0030] As attached Figure 5 As shown, attached Figure 5 Describes the changes in the number of passengers waiting for the subway at different times. At time t The total waiting time of passengers who choose to take the subway is the product of the number of passengers and the waiting time of each passenger. Figure 5 The total waiting time T1 for taking the subway is obtained from the graph area of ​​:

[0031]

[0032] Where h represents the length of the survey period; Q1 represents the passenger flow at the beginning; Indicates passenger flow arriving at any given moment; Indicates The flow of passengers waiting at the platform at all times.

[0033] S122: Propose a simplified calculation formula for the total waiting time of buses in high-speed rail passenger transport hubs

[0034] As attached Figure 6 As shown, attached Figure 6 Depicts the changes in the number of passengers waiting for the bus at different times. At time t The total waiting time of the passengers who choose to take the bus is the product of the number of passengers and the waiting time of each passenger. Figure 6 The total waiting time T2 for taking the bus is obtained from the area of ​​the graph:

[0035]

[0036] Where Q2 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The flow of passengers waiting at the platform at all times.

[0037] S123: Propose a simplified calculation formula for the total taxi waiting time within high-speed rail passenger transport hubs

[0038] As attached Figure 7 As shown, attached Figure 7 Describes the changes in the number of passengers waiting for taxi service at different times. At time t The total waiting time of the passengers who choose to take a taxi is the product of the number of passengers and the waiting time of each passenger. Figure 7 The total waiting time T3 for taking a taxi is obtained from the area of ​​the graph:

[0039]

[0040] Where Q3 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The flow of passengers waiting at the taxi dispatch station at all times.

[0041] S124: Propose a simplified calculation formula for the total waiting time for ride-hailing services within high-speed rail passenger transport hubs

[0042] As attached Figure 8 As shown, attached Figure 8 Depicts the changes in the number of passengers waiting for ride-hailing vehicles at different times. At time t The total waiting time of passengers who choose to take the online car-hailing service is the product of the number of passengers and the waiting time of each passenger. Figure 8 The total waiting time T4 for choosing to take a ride-hailing car is obtained from the graph area:

[0043]

[0044] Where, is The number of people who can immediately accept online ride-hailing services at the time. Q4 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The flow of passengers waiting at the online car-hailing dispatch station at all times.

[0045] The S2 includes:

[0046] S21: Establish the upper-level planning model as a departure interval decision model based on multi-objective optimization

[0047] S211: The decision variable is the average departure interval t of different transportation modes.

[0048] S212: Goal 1 is to minimize the generalized average waiting time T average , the waiting time T through multiple transportation modes i The sum of the two divided by the total passenger flow Qz is:

[0049]

[0050] in,

[0051]

[0052] Where M is the set of all transportation modes connected to the high-speed rail passenger transport hub, i represents a certain transportation mode, and Q i It represents the passenger flow of the i-th mode of transportation, Ti represents the total waiting time of the i-th mode of transportation.

[0053] Goal 2 is to minimize the dispatch cost C, the unit cost of dispatching an additional vehicle through the i-th mode of transportation is c i Multiplying by the number of vehicles dispatched. Determining the unit cost of different modes of transportation requires consulting official websites for publicly available information and relevant research papers. Specifically, this information may include operating costs, fuel consumption, maintenance costs, equipment depreciation, labor costs, and other relevant factors for each mode of transportation. Unit costs for different modes of transportation can also be calculated using cost components.

[0054]

[0055] Where, T y is the length of the study period; c i represents the additional operating cost of adding one vehicle to the i-th mode of transportation; t i represents the average departure interval of the i-th mode of transportation.

[0056] S213: Constraints include

[0057] Constraint 1: The number of passengers carried by each mode of transportation must be greater than or equal to the number of passengers allocated; that is, the passenger evacuation capacity of each mode of transportation must be greater than or equal to the number of travelers who choose this mode of transportation. The number of passengers allocated to each mode of transportation is determined by the hybrid intelligent algorithm introduced later.

[0058]

[0059] In the formula, o1 represents the average number of passengers carried by taxis, and its value is usually obtained through traffic surveys, which count the number of passengers carried by multiple taxis in the taxi dispatch station of the passenger hub within a certain period of time and take the average value; o2 represents the average number of passengers carried by online-hailing taxis, and its value is also obtained through traffic surveys, which count the number of passengers carried by multiple online-hailing taxis in the online-hailing taxi dispatch station of the passenger hub within a certain period of time and take the average value.

[0060] Constraint 2: The average departure interval should be between the upper and lower limits of the departure interval determined by each mode of transportation.

[0061] t min <t<t max

[0062] In this formula, the average intervals between different modes of transportation need to be determined through inquiries or surveys based on actual conditions. For subways, the upper limit of the interval depends on the maximum safety limit determined under current operating rules, while the lower limit depends on the minimum safe interval that can be maintained for connecting subways. For buses, the upper limit of the interval depends on the maximum safety limit determined under current operating rules, while the lower limit depends on the inverse of the number of buses available during the study period. For taxis, the upper limit of the interval depends on the inverse of the number of vehicles serving the high-speed rail hub under normal circumstances, while the lower limit depends on the average taxi service time. For ride-hailing services, the upper limit of the interval depends on the inverse of the number of vehicles serving the high-speed rail hub under normal circumstances, while the lower limit depends on the parking lot entrance and exit capacity restrictions.

[0063] S22: Establish the lower-level planning model as a travel mode selection model based on stochastic user equilibrium

[0064] S221: The optimization objective is based on the individual choice of the traveler: the traveler influences the results of the upper model by choosing the transportation method with the shortest corresponding time. The objective formula for the lower model is as follows:

[0065]

[0066] Wherein, parameter δ is a parameter that measures the error between the perceived walking time and the actual walking time.

[0067] To determine parameter δ, we use a high-speed train arriving at a passenger hub as an example and, through traffic surveys, determine the proportion of passengers choosing different modes of transportation. The number of passengers on a specific high-speed train is input into the algorithm, with parameter δ set to an initial value of 0.01. The algorithm is then run to determine the algorithm's assigned passenger flow distribution ratios for each mode of transportation. If the algorithm-assigned passenger flow ratios for each mode of transportation are similar or even zero, parameter δ is too small. If the algorithm-assigned passenger flow ratios for each mode of transportation differ significantly, or if the passenger flow distributions for some modes are identical, parameter δ is too large. The parameter is repeatedly adjusted in steps of 0.001 until the algorithm-assigned passenger flow ratios for each mode of transportation do not differ by more than 10% from the actual surveyed ratios, while also satisfying the constraints.

[0068] S222: Constraints include

[0069] Constraint 1: The passenger flow carried by each mode of transportation should be a number greater than zero.

[0070] Q i >0,i∈M

[0071] The S3 includes:

[0072] Step 0: Initial data determination. Determine the objective function f(t), the maximum number of iterations maxit, the population size npc, the range of the decision variable t, the step size len, the crossover probability pc, and the mutation probability pm. Also, give the waiting time for each mode of transportation under the initial zero flow and the current average departure interval.

[0073] The objective function f(t) is the two objective functions determined by the upper model; the maximum number of iterations maxit determines the number of generations at which the iterations are stopped. If there are obvious problems with the output results and the output image does not reach Pareto optimality, the maximum number of iterations maxit can be increased. If the image basically does not change in the subsequent iterations, the maximum number of iterations maxit can be reduced to ensure that the output results tend to be stable; the population size npc will affect the convergence result of the Pareto optimality. The larger the scale, the clearer the trend of the output results will be, but the running time will increase. It needs to be determined by comprehensively considering the output effect and the running time; the step size len is the minimum unit of change of the decision variable t, which needs to be determined according to the minimum time unit of the study period; the current average departure interval is obtained by investigating the departure intervals of various transportation modes in the transportable hub during the study period and taking the average value. For taxis and online car-hailing, the inverse of the average arrival rate during the study period is taken; the crossover probability pc and the mutation probability pm will affect the population diversity and convergence speed. The crossover probability pc is generally 0.6-0.9, and the mutation probability pm is generally the inverse of the number of decision variables. When the algorithm converges slowly or the output image is not good, the crossover probability and mutation probability can be appropriately increased to increase population diversity and speed up convergence. When the algorithm results are of poor quality or cannot converge, the crossover probability and mutation probability can be appropriately reduced to make the algorithm more focused on searching for the optimal solution space.

[0074] The calculation formula for the waiting time of each mode of transportation under initial zero flow is as follows:

[0075]

[0076]

[0077] Where, T1 0 represents the subway waiting time under the initial zero flow; T2 0 represents the bus waiting time under initial zero flow; T3 0 represents the taxi waiting time under the initial zero flow; T4 0 represents the waiting time of online ride-hailing vehicles under initial zero flow; Indicates the average departure interval of the current subway; Indicates the average departure interval of buses at present; Indicates the average departure interval of taxis at present; Indicates the average departure interval of the current online ride-hailing service;

[0078] Step 1: Allocate traffic volume. Through the continuous average algorithm, obtain the passenger flow distribution Q of each mode of transportation at this time i First, set the passenger flow of all modes to zero and perform a passenger flow allocation. Then, using the zero flow allocation result as the first generation, iterate the passenger flow, additional flow, process flow, and accuracy parameters according to the following formula. When the accuracy parameter d < 1, stop the loop.

[0079] The algorithm formula is as follows:

[0080]

[0081] Where n represents the number of iterations; f i n represents the additional flow of the nth generation of the i-th mode of transportation; P i n represents the probability of the i-th mode of transportation being selected in the n-th generation; x i n represents the process flow of the nth generation of the i-th mode of transportation; represents the assigned passenger flow of the nth generation of the i-th mode of transportation; Represents the accuracy parameter of the nth generation of the i-th mode of transportation.

[0082] Step 2: Initial population generation. Randomly generate the average departure interval t of different modes of transportation between the upper and lower limits of the determined departure interval i The cost C and the generalized average waiting time T under different departure intervals are calculated by the objective function formula. average , until the population size is reached.

[0083] Step 3: Fast non-dominated sorting. Record the number of times all individuals are dominated by loop comparison and assign non-dominated ranks. Assume that there are multiple objective functions denoted as f j (x), when j is any integer in 1,2,…,m, k is also any integer in 1,2,…,m, but j≠k. If there are individuals x1 and x2 with f for any objective function j (x1) <f k (x2), then individual x1 is said to dominate x2; if there exists an objective function such that f j (x1)≤f k (x2) holds, and there exists an objective function that satisfies f j (x1)>f k(x2), in this case individuals x1 and x2 are said to be mutually non-dominated. Assign a non-dominated rank of 1 to the individual whose number of dominations is 0. Then, subtract 1 from the number of dominations of the remaining individuals. If the number of dominations is 0, assign a non-dominated rank of 2. Repeat this process until all individuals have been assigned non-dominated ranks.

[0084] Step 4: Determine the crowding degree. Calculate the crowding degree of all individuals separately; crowding degree is the density of individuals around a given point in the population, which is used as a reference value for screening individuals. The crowding degree calculation formula is as follows:

[0085]

[0086] Where, I[j+1].m and I[j-1].m are the previous function value and the next function value of j in function m respectively; and are the maximum and minimum values ​​in function m respectively.

[0087] Step 5: Selection. Using the binary tournament selection algorithm, two individuals are randomly selected and their non-dominated ranks and crowding degrees are compared. The individual with the lower non-dominated rank is prioritized. If two individuals have the same non-dominated rank, the individual with the higher crowding degree is selected and placed in the next generation of the population. This process is repeated until the population is restored. Suppose individual j is compared with another individual k. If individual j's non-dominated rank is lower than that of individual k, individual j is selected for the next generation of the population. If the two individuals have the same non-dominated rank, their crowding degrees are compared. If individual j's crowding degree is higher than that of individual k, individual j is selected for the next generation of the population.

[0088] Step 6: Crossover and mutation. Perform crossover and mutation operations on the new offspring individuals based on the set crossover probability pc and mutation probability pm. Crossover involves swapping the decision variables t of two individuals according to the set probability, generating two new individuals. Mutation involves randomly changing the decision variable t of an individual to generate a new individual.

[0089] Step 7: Elite Retention Strategy. The parent and offspring populations are merged to form a new population. Fast non-dominated sorting and selection operations are performed on this new population. The goal is to maintain population diversity while retaining the optimal solutions in each generation, preventing these optimal solutions from being destroyed during crossover and mutation. Finally, the best individuals are selected from the new population and added to the new parent set.

[0090] Step 8: Select the optimal solution. The loop stops after reaching the maximum number of iterations (maxit). The top 10 individuals in the latest generation, ranked from highest to lowest congestion, are selected and the utility Z of each solution is calculated. Utility Z serves as a reference value for selecting solutions. Changes in the two objective functions affect the magnitude of utility Z. When utility Z does not change much, the loop stabilizes.

[0091] The formula for calculating utility Z is as follows:

[0092]

[0093] In the formula, l represents the number of the selected solution, tt l represents the generalized average waiting time per person for solution l, C l represents the operating cost value of plan l; Z l represents the utility of plan l; tt max is the largest generalized average waiting time among the 10 selected options; C max It is the maximum operating cost value among the 10 selected options.

[0094] Step 9: Loop judgment. Calculate the change in utility Z between the two iterations, and record it as the utility change VZ. If VZ is less than 0.1, stop the loop; otherwise, repeat Step 1.

[0095] Step 10: Design of transport capacity allocation scheme. Use the selected optimal scheme to design a transport capacity allocation scheme. The formula for calculating the number of vehicles for different modes of transportation is as follows:

[0096]

[0097] Where S i is the number of vehicles in the i-th mode of transportation, is the average departure interval of the i-th mode of transportation selected as the optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] The specific implementation of this method will be further described in detail below with reference to the accompanying drawings.

[0099] Figure 1 A schematic diagram showing the calculation of waiting time and area for subways with simultaneous passenger arrivals according to the present invention is shown.

[0100] Figure 2 A schematic diagram showing the calculation of the waiting time area for buses with simultaneous passenger arrivals according to the present invention is shown.

[0101] Figure 3 A schematic diagram showing the calculation of taxi waiting time area for simultaneous passenger arrivals according to the present invention is shown.

[0102] Figure 4A schematic diagram showing the area calculation for waiting time of online ride-hailing vehicles with simultaneous passenger arrival according to the present invention is shown.

[0103] Figure 5 A schematic diagram showing the calculation of waiting time and area for a subway with passengers arriving in two batches according to the present invention is shown.

[0104] Figure 6 A schematic diagram showing the calculation of waiting time and area for a bus with passenger flow arriving in two batches according to the present invention is shown.

[0105] Figure 7 A schematic diagram showing the calculation of taxi waiting time and area when passengers arrive in two batches according to the present invention is shown.

[0106] Figure 8 A schematic diagram showing the calculation of waiting time and area for online ride-hailing vehicles with passenger flow arriving in two batches according to the present invention is shown.

[0107] Figure 9 Iteration diagram showing the algorithmic utility of an analysis example of the present invention.

[0108] Figure 10 Iterative graph showing the algorithm utility change of the analysis example of the present invention.

[0109] Figure 11 A Pareto graph showing the output of an example algorithm for analyzing the present invention.

[0110] Figure 12 A flow chart showing a method for optimizing the configuration of connecting transport capacity at a high-speed rail passenger transport hub according to the present invention is shown. DETAILED DESCRIPTION

[0111] To more clearly illustrate the present invention, the present invention is further described below with reference to preferred examples and accompanying drawings. The following detailed description is illustrative and non-restrictive and should not be used to limit the scope of protection of the present invention. In this example, the time unit is minutes.

[0112] 1. A simplified calculation formula for the total waiting time of each connecting transportation mode at the high-speed rail passenger hub is proposed, taking the simultaneous arrival of passengers during the study period as an example.

[0113] S11: Propose a simplified calculation formula for the total waiting time for subways within high-speed rail passenger transport hubs

[0114] In this example, q1 = 2640, t w1 =2.77, substitute into the formula as follows:

[0115]

[0116] S12: Propose a simplified calculation formula for the total waiting time of buses in high-speed rail passenger transport hubs

[0117] In this example, q2=48, t w2=2.88, substitute into the formula as follows:

[0118]

[0119] S13: Propose a simplified calculation formula for the total taxi waiting time within high-speed rail passenger transport hubs

[0120] In this example, n1=0, t s =0.13, t w3 =2.60, substitute into the formula as follows:

[0121]

[0122] S14: Propose a simplified calculation formula for the total waiting time for ride-hailing services within high-speed rail passenger transport hubs

[0123] In this example, n2=12, t w4 =2.60, substitute into the formula as follows:

[0124]

[0125] 2. Construct a high-speed rail passenger hub connecting capacity configuration optimization model based on double-layer planning;

[0126] S21: Establishing the upper-level planning model as a departure interval decision model based on multi-objective optimization;

[0127] S211: The decision variable is the average departure interval t of different transportation modes i .

[0128] S212: Goal 1 is to minimize the generalized average waiting time T average , obtained by dividing the sum of the waiting time T of multiple transportation modes by the total passenger flow Qz. In this example, the total passenger flow Qz is 9600.

[0129]

[0130] Objective 2 is to minimize the dispatch cost, which is obtained by multiplying the unit cost of adding a vehicle by the number of vehicles dispatched for different modes of transportation. In this example, the unit cost of the subway is 2978.84, the unit cost of the bus is 129.85, and the unit cost of taxis and online ride-hailing vehicles is 48.95. The research period T y Take 60.

[0131]

[0132] S213: Constraints include

[0133] Constraint 1: The number of passengers carried by each mode of transportation must be greater than or equal to the number of passengers assigned to it; that is, the passenger evacuation capacity of each mode of transportation must be greater than or equal to the number of travelers choosing that mode of transportation. In this example, traffic surveys determined that o1 = 1.34 and o2 = 1.75.

[0134]

[0135] Constraint 2: The average departure interval should be between the upper and lower limits of the departure interval determined by each mode of transportation.

[0136] t min <t<t max

[0137] In this example, the subway departure interval t1∈[3,9]; the bus departure interval t2∈[1,2.85]; the taxi departure interval t3∈[0.06,0.1]; and the online car-hailing departure interval t4∈[0.05,0.069].

[0138] S22: Establish the lower-level planning model as a travel mode selection model based on stochastic user equilibrium

[0139] The optimization objective is based on individual traveler choices: travelers influence the results of the upper-level model by choosing the mode of transportation that minimizes travel time. In this example, surveys revealed that the ratios of subway, bus, taxi, and ride-hailing are approximately 45:18:12:23. The algorithm was run to adjust the parameter δ so that the algorithm's passenger flow allocation closely matched the actual passenger flow distribution. The final parameter δ was set to 0.02, and the objective formula for the lower-level model was established as follows:

[0140]

[0141] Constraints include

[0142] Constraint 1: The passenger flow carried by each mode of transportation should be a number greater than zero.

[0143] Q i >0,i∈M

[0144] 3. Propose a hybrid intelligent optimization algorithm to solve bi-level programming

[0145] Step 0: Initial data determination. In this example, set the maximum number of iterations maxit = 100, the population size npc = 100, the step length len = 0.01, the crossover probability pc = 0.8, the mutation probability pm = 0.25, and the total passenger flow Qz = 9600. By calculation, the subway waiting time T1 under the initial zero flow 0 =12.8, the bus waiting time T2 under the initial zero flow0 =5, taxi waiting time T3 under initial zero flow 0 =2.1, the waiting time for online ride-hailing vehicles under initial zero flow is T4 0 =2.6; According to traffic survey, the current average departure interval of the subway in this example is t1 0 =9, the current average departure interval of buses is t2 0 =8, the current average taxi departure interval t3 0 =0.5, the current average departure interval of online ride-hailing services is t4 0 =0.5.

[0146] Step 1: Allocate traffic volume. Through the continuous average algorithm, obtain the passenger flow distribution Q of each mode of transportation at this time i First, set the passenger flow of all modes to zero and perform a passenger flow allocation. Then, using the zero flow allocation result as the first generation, iterate the passenger flow, additional flow, process flow, and accuracy parameters according to the following formula. When the accuracy parameter d < 1, stop the loop.

[0147] The algorithm formula is as follows:

[0148]

[0149] Where n represents the number of iterations; f i n represents the additional flow of the nth generation of the i-th mode of transportation; P i n represents the probability of the i-th mode of transportation being selected in the n-th generation; x i n represents the process flow of the nth generation of the i-th mode of transportation; represents the assigned passenger flow of the nth generation of the i-th mode of transportation; Represents the accuracy parameter of the nth generation of the i-th mode of transportation.

[0150] Step 2: Initial population generation. Randomly generate the average departure interval t of different modes of transportation between the upper and lower limits of the determined departure interval i The cost C and the generalized average waiting time T under different departure intervals are calculated by the objective function formula. average , until the population size is reached.

[0151] Step 3: Fast non-dominated sorting. Record the number of times all individuals are dominated by loop comparison and assign non-dominated ranks. Assume that there are multiple objective functions denoted as f j (x), when j is any integer in 1,2,…,m, k is also any integer in 1,2,…,m, but j≠k. If there are individuals x1 and x2 with f for any objective functionj (x1) <f k (x2), then individual x1 is said to dominate x2; if there exists an objective function such that f j (x1)≤f k (x2) holds, and there exists an objective function that satisfies f j (x1)>f k (x2), in this case individuals x1 and x2 are said to be mutually non-dominated. Assign a non-dominated rank of 1 to the individual whose number of dominations is 0. Then, subtract 1 from the number of dominations of the remaining individuals. If the number of dominations is 0, assign a non-dominated rank of 2. Repeat this process until all individuals have been assigned non-dominated ranks.

[0152] Step 4: Determine the crowding degree. Calculate the crowding degree of all individuals separately; crowding degree is the density of individuals around a given point in the population, which is used as a reference value for screening individuals. The crowding degree calculation formula is as follows:

[0153]

[0154] Where, I[j+1].m and I[j-1].m are the previous function value and the next function value of j in function m respectively; and are the maximum and minimum values ​​in function m respectively.

[0155] Step 5: Selection. Using the binary tournament selection algorithm, two individuals are randomly selected and their non-dominated ranks and crowding levels are compared. The individual with the lower non-dominated rank is selected first. If the two individuals have the same non-dominated rank, the individual with the higher crowding level is selected and placed in the next generation of the population. This process is repeated until the population recovers.

[0156] Step 6: Crossover and mutation. Perform crossover and mutation operations on the new offspring individuals according to the set crossover probability pc = 0.8 and mutation probability pm = 0.25.

[0157] Step 7: Elite retention strategy. The parent population and the child population are merged to form a new population. The new population is subjected to a fast non-dominated sorting operation and a selection operation. Then, the best individuals are selected from the new population and placed in the new parent population.

[0158] Step 8: Select the optimal solution. The loop stops after reaching the maximum number of iterations (maxit = 100). The top 10 individuals from the latest generation, ranked from highest to lowest crowding, are selected and the utility Z of each solution is calculated. Utility Z serves as a reference value for selecting solutions. Changes in the two objective functions affect the magnitude of utility Z. When utility Z does not change much, the loop stabilizes.

[0159] The formula for calculating utility Z is as follows:

[0160]

[0161] In the formula, l represents the number of the selected solution, tt l represents the generalized average waiting time per person for solution l, C l represents the operating cost value of plan l; tt max is the maximum generalized average waiting time among the 10 selected options; C max It is the maximum operating cost value among the 10 selected options.

[0162] Record the utility of each generation and draw the following Figure 9 The algorithm utility iteration graph shows the number of iterations on the horizontal axis and the utility of each generation on the vertical axis. This graph provides a complete record of the algorithm's utility changes during the iteration process, providing a basis for further data analysis and algorithm optimization.

[0163] Record the utility change value of each adjacent iteration and plot it as shown in the attached figure. Figure 10 The algorithm utility change iteration graph is shown in the figure, with the horizontal axis representing the number of iterations and the vertical axis representing the utility change between each adjacent iteration. This graph clearly shows the dynamic changes in the algorithm utility during the iteration process.

[0164] Step 9: Loop judgment. Calculate the change in utility Z between the two iterations, and record it as the utility change VZ. If VZ is less than 0.1, stop the loop, otherwise repeat Step 1. In this example, 14 loops were performed, and the final selected solution is shown in the table below. The output Pareto optimal curve is shown in the attached figure. Figure 11 shown.

[0165]

[0166] Step 10: Design a transport capacity allocation plan. Use the selected plan to design a transport capacity allocation plan. The formula for calculating the number of vehicles for different modes of transportation is as follows:

[0167]

[0168] Where S i is the number of vehicles in the i-th mode of transportation, is the average departure interval of the i-th mode of transportation selected as the optimal solution. The capacity allocation solution designed based on the selected solution is shown in the following table.

[0169] S1 S2 S3 S4 7 22 600 870

Claims

1. A method for optimizing the connection capacity configuration of a high-speed rail passenger transport hub, characterized in that Here are the steps: S1: Propose a simplified calculation formula for the total waiting time of each connecting transportation mode at the high-speed rail passenger transport hub; S2: Constructing a high-speed rail passenger transport hub connection capacity configuration optimization model based on double-layer planning; S3: Propose a hybrid intelligent optimization algorithm to solve bi-level programming; Said S1 comprises: S11: When the passenger flow arrives at the same time at the beginning of the study period, the calculation formula is as follows: S111: Propose a simplified calculation formula for the total waiting time for subways within high-speed rail passenger transport hubs The total waiting time T1 for taking the subway is: Among them, t w1 represents the average travel time from the exit to the subway station; q1 represents the average number of people carried by each subway train during the study period, determined through traffic surveys; t1 represents the average departure interval of the subway during the study period, used to indicate the subway's capacity allocation; Q1 represents the passenger volume that chooses to take the subway; S112: Propose a simplified calculation formula for the total waiting time of buses in high-speed rail passenger transport hubs The total waiting time T2 for taking the bus is: Among them, t w2 represents the average travel time from the exit to the bus stop; q2 represents the average number of people carried by each bus during the study period, determined through traffic surveys; t2 represents the average departure interval of buses during the study period, used to represent the bus capacity allocation; Q2 represents the passenger flow that chooses to take the bus; S113: Propose a simplified calculation formula for the total taxi waiting time within high-speed rail passenger transport hubs The total waiting time T3 for taking a taxi is: Among them, t s represents the average service time of a taxi; t w3 represents the average travel time for passengers from the exit to the taxi dispatch station; n1 represents the number of people who can immediately receive taxi service. Its value is determined through traffic surveys. It requires counting the number of people who can immediately receive taxi service when passengers arrive multiple times during the study period and taking the average value; Q3 represents the number of passengers who choose to take taxis; t3 represents the average taxi departure interval, which is actually the inverse of the taxi arrival rate. S114: Propose a simplified calculation formula for the total waiting time for ride-hailing services within high-speed rail passenger transport hubs The total waiting time T4 for taking an online taxi is: Among them, t w4 represents the average travel time for passengers from the exit to the online car-hailing dispatch station; n2 represents the number of people who can immediately accept online car-hailing services. Its value is determined through traffic surveys. It is necessary to count the number of people who can immediately accept online car-hailing services when passengers arrive multiple times during the study period and take the average value; Q4 represents the number of passengers who choose to take online car-hailing services; t4 represents the average departure interval of online car-hailing services, which is actually the inverse of the online car-hailing arrival rate. S12: When the passenger flow arrives multiple times during the study period, the arrival time and Taking the arrival time as an example, the calculation method is as follows: S121: Propose a simplified calculation formula for the total waiting time for subways within high-speed rail passenger transport hubs The total waiting time T1 for taking the subway is: Where h represents the length of the survey period; Q1 represents the passenger flow at the beginning; Indicates passenger flow arriving at any given moment; Indicates The number of passengers waiting at the platform at all times; S122: Propose a simplified calculation formula for the total waiting time of buses in high-speed rail passenger transport hubs The total waiting time T2 for taking the bus is: Where Q2 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The number of passengers waiting at the platform at all times; S123: Propose a simplified calculation formula for the total taxi waiting time within high-speed rail passenger transport hubs The total waiting time T3 for taking a taxi is: Where Q3 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The number of passengers waiting at the taxi dispatch station at all times; S124: Propose a simplified calculation formula for the total waiting time for ride-hailing services within high-speed rail passenger transport hubs The total waiting time T4 for taking an online taxi is: Where, is The number of people who can immediately accept online ride-hailing services at the time; Q4 represents the passenger flow at the start time; Indicates Increased passenger flow at all times; express The number of passengers constantly waiting at the ride-hailing dispatch station; The S2 includes: S21: Establish the upper-level planning model as a departure interval decision model based on multi-objective optimization S211: The decision variable is the average departure interval t of different transportation modes; S212: Goal 1 is to minimize the generalized average waiting time T average , the waiting time T through multiple transportation modes i The sum of the two divided by the total passenger flow Qz is: in, Where M is the set of all transportation modes connected to the high-speed rail passenger transport hub, i represents a certain transportation mode, and Q i It represents the passenger flow of the i-th mode of transportation, T i represents the total waiting time of the i-th mode of transportation; Goal 2 is to minimize the dispatch cost C, the unit cost of dispatching an additional vehicle through the i-th mode of transportation is c i Multiply by the number of vehicles dispatched; Where, T y is the length of the study period; c i represents the additional operating cost of adding one vehicle to the i-th mode of transportation; t i represents the average departure interval of the i-th mode of transportation; S213: Constraints include Constraint 1: The number of passengers carried by each mode of transportation must be greater than or equal to the number of passengers assigned to it. This means that the passenger evacuation capacity of each mode of transportation must be greater than or equal to the number of travelers who choose that mode of transportation. The number of passengers assigned to each mode of transportation is determined using the hybrid intelligent algorithm described later. Where o1 represents the average number of passengers carried by taxis, which is obtained by taking the average value of the number of passengers carried by multiple taxis at the taxi dispatch station of the passenger transport hub within a certain period of time through traffic surveys. o2 represents the average number of passengers carried by online-hailing taxis, which is also obtained by taking the average value of the number of passengers carried by multiple online-hailing taxis at the taxi dispatch station of the passenger transport hub within a certain period of time through traffic surveys. Constraint 2: The average departure interval should be between the upper and lower limits of the departure interval determined by each mode of transportation; t min <t<t max In the formula, the average departure interval of different modes of transportation is determined through inquiry or survey; for the subway, the upper limit of the departure interval depends on the maximum safety value determined under the current operating rules, and the lower limit of the departure interval depends on the minimum safe departure interval that can be maintained by the connecting subway; for buses, the upper limit of the departure interval depends on the maximum safety value determined under the current operating rules, and the lower limit of the departure interval depends on the inverse of the number of available buses during the study period; for taxis, the upper limit of the departure interval depends on the inverse of the number of vehicles serving the high-speed rail hub under normal circumstances, and the lower limit of the departure interval depends on the average service time of taxis; for online ride-hailing services, the upper limit of the departure interval depends on the inverse of the number of vehicles serving the high-speed rail hub under normal circumstances, and the lower limit of the departure interval depends on the traffic capacity restrictions of parking lot entrances and exits; S22: Establish the lower-level planning model as a travel mode selection model based on stochastic user equilibrium S221: The optimization objective is based on the individual choices of travelers: travelers influence the results of the upper model by choosing the transportation method with the shortest corresponding time. The objective formula for the lower model is as follows: Where, parameter δ is a parameter that measures the error between the perceived walking time and the actual walking time; The initial value of parameter δ is 0.01, and the parameter size is repeatedly adjusted in steps of 0.001 to ensure that the difference between the passenger flow distribution ratio of different transportation modes assigned by the algorithm and the passenger flow distribution ratio of different transportation modes in the actual survey does not exceed 10%, and the constraint conditions are met; S222: Constraints include Constraint 1: The passenger flow carried by each mode of transportation should be a number greater than zero; Q i >0,i∈M The S3 includes: Step 0: Determine the initial data; determine the objective function f(t), the maximum number of iterations maxit, the population size npc, the range of the decision variable t, the step size len, the crossover probability pc, and the mutation probability pm, and give the waiting time of each mode of transportation under the initial zero flow and the current average departure interval; The objective function f(t) is the two objective functions determined by the upper model; the maximum number of iterations, maxit, determines the number of generations at which the iterations stop; the step size, len, is the minimum unit of change in the decision variable t; the current average departure interval is obtained by surveying the departure intervals of various transportation modes in the transportable hub during the study period and taking the average value; for taxis and online ride-hailing services, the inverse of the average arrival rate during the study period is taken; the crossover probability, pc, is set to 0.6-0.9, and the mutation probability, pm, is taken to be the inverse of the number of decision variables; The calculation formula for the waiting time of each mode of transportation under initial zero flow is as follows: Where, represents the subway waiting time under initial zero flow; represents the bus waiting time under initial zero flow; represents the taxi waiting time under the initial zero flow; represents the waiting time of online ride-hailing vehicles under initial zero flow; Indicates the average departure interval of the current subway; Indicates the average departure interval of buses at present; Indicates the average departure interval of taxis at present; Indicates the average departure interval of the current online ride-hailing service; Step 1: Allocate traffic volume; obtain the passenger flow distribution Q of each mode of transportation at this time through the continuous average algorithm i First, set the passenger flow of all modes of transportation to zero and perform a passenger flow allocation. Then, using the zero flow allocation result as the first generation, iterate the passenger flow, additional flow, process flow, and accuracy parameters according to the following formula. When the accuracy parameter d < 1, stop the loop. The algorithm formula is as follows: Where n represents the number of iterations; represents the additional flow of the nth generation of the i-th mode of transportation; represents the probability of the i-th mode of transportation being selected in the n-th generation; represents the process flow of the nth generation of the i-th mode of transportation; represents the assigned passenger flow of the nth generation of the i-th mode of transportation; represents the accuracy parameter of the nth generation of the i-th mode of transportation; Step 2: Initial population generation; randomly generate the average departure interval t of different modes of transportation between the upper and lower limits of the determined departure interval i The cost C and the generalized average waiting time T under different departure intervals are calculated by the objective function formula. average , until the population size is reached; Step 3: Fast non-dominated sorting; record the number of times all individuals are dominated by loop comparison and assign non-dominated levels; assuming there are multiple objective functions denoted as f j (x), when j is any integer in 1, 2, ..., m, k is also any integer in 1, 2, ..., m, but j ≠ k; if there are individuals x1 and x2 with f for any objective function j (x1)<f k (x2), then individual x1 is said to dominate x2; if there exists an objective function such that f j (x1)≤f k (x2) holds, and there exists an objective function that satisfies f j (x1)>f k (x2), at this time, individuals x1 and x2 are said to be non-dominated by each other; the individual whose number of domination is 0 is assigned a non-domination level of 1; then the number of dominations of other individuals is reduced by 1, and if it is 0, the non-domination level is assigned to 2; and so on, until all individuals have been assigned non-domination levels; Step 4: Determine the crowding degree; calculate the crowding degree of all individuals separately; the crowding degree is the density of individuals around a given point in the population, which is used as a reference value for screening individuals; the crowding degree calculation formula is as follows: Where, I[j+1].m and I[j-1].m are the previous function value and the next function value of j in function m respectively; and are the maximum and minimum values ​​in function m respectively; Step 5: Selection; Using the binary tournament selection algorithm, randomly select two individuals to compare their non-dominated ranks and crowding degrees; give priority to individuals with lower non-dominated ranks. If the non-dominated ranks of the two individuals are the same, select the individual with higher crowding degrees and place the better individual in the next generation population. Repeat this process until the population size is restored. Suppose that individual j is compared with another individual k. If the non-dominated rank of individual j is lower than that of individual k, then individual j is selected to be placed in the next generation population. If the non-dominated ranks of the two individuals are the same, compare their crowding degrees. If the crowding degree of individual j is higher than that of individual k, then individual j is selected to be placed in the next generation population. Step 6: Crossover and mutation: Perform crossover and mutation operations on the new offspring individuals according to the set crossover probability pc and mutation probability pm. Crossover is to exchange the decision variables t of two individuals according to the set probability to produce two new individuals; mutation is to randomly change the decision variable t of an individual to generate a new individual. Step 7: Elite retention strategy: merge the parent population and the child population to form a new population, and perform fast non-dominated sorting and selection operations on the new population; Step 8: Select the optimal solution; stop the iteration after the maximum number of iterations (maxit), select the top 10 individuals in the latest generation ranked from high to low in terms of crowding, and calculate the utility Z of each solution; utility Z serves as a reference value for screening solutions. Changes in the two objective functions will affect the size of utility Z. When the change in utility Z is not large, the cycle tends to be stable. The utility Z calculation formula is as follows: In the formula, l represents the number of the selected solution, tt l represents the generalized average waiting time per person for solution l, C l represents the operating cost value of plan l; Z l represents the utility of plan l; tt max is the maximum generalized average waiting time among the 10 selected options; C max It is the maximum operating cost value among the 10 selected options; Step 9: Loop judgment; calculate the change in utility Z between the two iterative calculations, and record it as the utility change ΔZ. If ΔZ is less than 0.1, stop the loop, otherwise repeat Step 1. Step 10: Design a transport capacity allocation plan; the formula for calculating the number of vehicles for different modes of transportation is as follows: Where S i is the number of vehicles in the i-th mode of transportation, is the average departure interval of the i-th mode of transportation selected as the optimal solution.

Images