Variable route bus service optimization method based on selection of origin and destination of passenger

By optimizing passenger start and ending points selection and travel strategies in variable line bus services, a social welfare maximization model was established, and the problem of unified fare affecting operational efficiency was solved, the social welfare per unit time was maximized, and differentiation was demonstrated. The superiority of fare.

CN119027151APending Publication Date: 2024-11-26CHINA ACAD OF TRANSPORTATION SCI
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410996795.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing technology ignores the need for off-site pickup and drop-offs caused by uniform fares when optimizing variable line bus services, affects operational efficiency, and is difficult to effectively solve the fare and time issues in the selection of 'first mile/last mile' travel method. .

Method used

It proposes a variable line bus service optimization method based on passenger starting and ending point selection. Through service area modeling, travel strategy selection and social welfare maximization model, considering flexible needs and traveler selection preferences, and optimizing fare policy to realize social welfare per unit time maximize.

Benefits of technology

It provides bus planners with reference to designing fixed stations, departure intervals and fares, improves operational efficiency, optimizes social welfare per unit time, and reveals the impact of key parameters, proving that differentiated fares are better than uniform fares.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027151A_ABST
    Figure CN119027151A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of traffic service, and discloses a passenger origin-destination selection-based variable route bus service optimization method, which comprises the following steps of: firstly, modeling a service area which is a rectangular area with length and width; then a travel strategy is selected, and travelers of the bus with the variable route can get on and off the bus at the fixed station and can also select to get on and off the bus at the demand response station; establishing a model, and maximizing unit time and social welfare of the variable route public transportation system; and finally, solving a model numerical value, specifically, solving the model numerical value by adopting a BONMIN solver. According to the invention, optimization is carried out when a ticket price policy is made so as to realize social welfare maximization in unit time. According to the variable route bus pricing scheme, elastic demands and traveler selection preferences are considered, and references for designing fixed stations, departure intervals and ticket prices in a given service area are provided for bus planners.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of traffic services, and in particular to a method for optimizing variable route public transportation services based on passenger starting and ending point selections. Background Art

[0002] Variable route buses have the characteristics of traditional buses, i.e. fixed routes, stops and timetables, as well as the characteristics of demand-responsive buses, i.e. vehicles are allowed to deviate from the basic routes within a certain range to pick up and drop off passengers. When there is no demand for pick-up and drop-off, the vehicles provide services according to fixed routes and timetables. These characteristics make them more suitable for areas with low bus demand. More and more studies have shown that the applicability of variable route bus systems has been widely recognized and verified in actual road networks. According to the research of Potts et al., among the many types of flexible bus services, variable route buses are considered to be the most popular form.

[0003] Variable route buses have both the high efficiency of conventional buses and the flexibility of demand-responsive buses, so they have attracted the attention of more and more researchers. In terms of operation optimization, Qiu et al. proposed a dynamic station strategy to improve the performance of variable route buses in an operating environment with uncertain travel demand. The study showed that the proposed strategy can reduce user costs by 30% without any additional operating costs. Nourbakhsh et al. proposed an alternative variable route bus system. The study found that compared with comparable systems (such as fixed route transportation networks and taxi services), the proposed bus system often has the lowest system cost at medium and low demand levels. Zheng et al. proposed a junction strategy, that is, passengers can get on and off at a predetermined stop or at a junction within an acceptable walking distance. Experimental results show that the junction strategy can significantly reduce the rejection rate by 24% at the cost of walking an acceptable distance. Qiu et al. identified a new type of bus policy, called semi-flexible operation policy, to fill the gap between flexible bus services and traditional fixed route systems. Li et al. and Zhang et al. proposed different junction strategies, respectively. The results show that the proposed strategies can improve the performance of variable route bus services under unexpected high travel demand levels. Zheng et al. proposed a delayed arrival strategy to reduce the number of rejected passengers and the idle time at fixed stations due to the uncertainty of travel demand. These studies only consider the travel time (walking time, waiting time and time in the car) of passengers in the variable route bus system, ignoring that the uniform fare will lead to additional off-station pick-up and drop-off demand, thus affecting operational efficiency.

[0004] Regarding travel mode choice, most of the literature focuses on different types of bus services or even different modes of transportation. Chavis et al. developed a mode choice model that describes how bus users choose emerging competitive bus options. The results show that monetary cost, expected walking time, expected waiting time in the car, and expected waiting time are significant predictors of choosing flexible transportation mode types, which is consistent with the utility terms that we consider to affect travelers' travel strategy choices. Yu Xiaohua et al. introduced the travel utility indifference threshold into the decision-making process of travelers, and the results showed that the randomness in the travel choice process can be accurately reflected. Qiu et al. studied the choice model between fixed route and variable route strategies in bus systems. Zheng et al. quantified how passengers choose among three competing travel modes (fixed route bus, variable route bus, and private car). Li et al. constructed an analytical and simulation model to help planners choose between demand response operation strategies and fixed route operation strategies, while deciding whether and when to switch from one operation strategy to another during the day. Chen

[19] Chen F, Yin Z, Ye Y, et al. Taxi hailing choice behavior and economic benefit analysis of emission reduction based on multi-mode travel big data[J]. Transport Policy, 2020, 97: 73-84. et al. proposed a binomial logit model to analyze the choices between subway and taxi, bicycle and taxi, and identify the corresponding influencing factors. For the "first mile / last mile", these travel mode options either have an average travel fare that is too high or an average travel time that is too long, and cannot effectively solve the "first mile / last mile" problem.

[0005] To address the above-mentioned shortcomings, a variable route bus service optimization method based on passengers' starting and ending point selection is needed. Summary of the invention

[0006] The purpose of the present invention is to provide a method for optimizing variable route bus service based on passenger starting and ending point selection. The present invention optimizes how to maximize social welfare per unit time when formulating fare policies. The proposed variable route bus pricing scheme takes into account elastic demand and traveler selection preferences, and provides a reference for bus planners to design fixed stops, departure intervals and fares in a given service area.

[0007] The present invention is implemented in this way. The present invention provides a variable route bus service optimization method based on passenger starting and ending point selection, which is specifically performed in the following steps:

[0008] S1: Model the service area, which is a rectangular area with a length of L and a width of W. N fixed stations are set in the service area, and the fixed stations are evenly distributed along the reference route according to the labels 1, 2, ..., N, and the maximum allowable width W / 2 of the reference route and the two sides from which the service vehicle can deviate is set.

[0009] S2: Choose a travel strategy. Travelers on the variable route bus can get on and off at fixed stops or at demand response stops.

[0010] Travel strategies include four types of travel strategies: Figure 2-Figure 4 ;

[0011] I: No demand response service is accepted, that is, the traveler first walks from the starting point to the fixed station closest to the starting point, then takes a bus from the fixed station to the fixed station closest to the destination, and finally walks from the fixed station to the destination;

[0012] II: Only receiving demand response services at the destination, that is, travelers first walk from the starting point to the fixed station closest to the starting point, and then take a bus to the destination;

[0013] III: Only receiving demand response services at the starting point, that is, the traveler first takes a bus from the starting point to the fixed station closest to the destination, and then walks from the fixed station to the destination;

[0014] IV: Receive demand response service at the starting and ending points, that is, travelers take the bus directly from the starting point to the destination.

[0015] Further, the practical variables of the present invention are defined as shown in Table 1;

[0016] Table 1 Symbols

[0017]

[0018]

[0019]

[0020] S3: Build a model to maximize the unit time and social welfare of a bus system with variable routes;

[0021] The model includes travel demand that is uniformly distributed spatially and over the service period in the service area; that is, demand response stations are uniformly distributed in the service area, and all ride requests are accepted. The scheduled departure time of fixed stations is predetermined based on the expected demand level, the distance between two points is represented by the Manhattan distance, and the linear motion calculation is first horizontally along the X axis and then vertically along the Y axis when the traveler walks and the variable route bus travels. The model includes elastic demand function, probability selection function, travel disutility term, fare, walking time, waiting time, time in the car, social welfare maximization without fiscal constraints, operator profit per unit time, and consumer surplus per unit time.

[0022] The model building is carried out according to the following steps:

[0023] S 3.1 : The actual travel demand Q of the service area is characterized by the elastic demand function, as shown in formula (1);

[0024] Q(p)=q(1-e p p), formula (1)

[0025] Where q is the potential travel demand in the service area, e p is a coefficient to measure the sensitivity of demand to travel impedance, p is the generalized travel cost of the traveler, and the generalized travel cost p of the traveler is as follows:

[0026] p=F+e K K+e A A+e R R, formula (2)

[0027] Among them, F, K, A, and R are the average fare, average walking time, average waiting time, and average time in the car of travelers respectively; K ,e A ,e R They are the value of unit walking time, unit waiting time and unit time in the car respectively.

[0028] The average fare F, average walking time K, average waiting time A and average in-vehicle time R of travelers are as shown in formulas (3) to (6);

[0029]

[0030] Among them, P i ; i=1,2,3,4 are the probabilities that travelers choose four travel strategies a, b, c, d, F i ; i=1,2,3,4 are the expected fares of travelers who choose four travel strategies a, b, c, d, K i; i = 1, 2, 3, 4 is the expected walking time for travelers to choose four travel strategies a, b, c, d, A i ; i=1,2,3,4 are the expected waiting times for travelers to choose four travel strategies a, b, c, d, R i ; i=1,2,3,4 is the expected in-vehicle time for travelers who choose four travel strategies a, b, c, d;

[0031] S 3.2 : The probability of travelers choosing four travel strategies, P1, P2, P3, P4, is calculated through the probability selection function, as shown in formula (7);

[0032]

[0033] Among them, X1, X2, X3, and X4 are the travel disutilities of travelers who choose four travel strategies: a, b, c, and d;

[0034] The travel disutility X1, X2, X3, X4 of the four travel strategies chosen by travelers a, b, c, d is as shown in formula (8);

[0035] X i =β0+β F F i +β K K i +β A A i +β R R i ,i=1,2,3,4, formula (8)

[0036] Among them, β0,β F ,β K ,β A ,β R is the travel disutility coefficient;

[0037] S 3.3 : Calculate the negative utility term of travel, as shown in formula (9)-formula (10);

[0038] For type I travelers who choose travel strategy a, their starting point falls evenly within the service area with a probability of 1, while the probability of their destination is approximately Therefore, the probability of a type I traveler choosing travel strategy a is As formula (9);

[0039]

[0040] The sum of the probabilities of type I and type II travelers choosing travel strategy a is 1, so the probability of type II travelers choosing travel strategy a is As formula (10);

[0041]

[0042] S 3.4 : Calculate the fare. Type I travelers who choose travel strategy a walk directly from the starting point to the end point without taking a car, that is, they do not need to buy a ticket. Travelers who choose travel strategies b and c receive one demand response service, while travelers who choose travel strategy d receive two demand response services. The expected fares F1, F2, F3, and F4 of travelers who choose travel strategies a, b, c, and d are as shown in formulas (11) to (13);

[0043]

[0044] F2=F3=(α+1)f, formula (12)

[0045] F4=(2α+1)f, formula (13)

[0046] Among them, f is the basic fare of the variable route bus, α is the fare adjustment coefficient for the traveler to receive an additional demand response service under the variable route bus service;

[0047] S 3.5 : Calculate the traveler's walking time; as shown in formula (14)-formula (17);

[0048] The expected longitudinal distance d1 between the fixed station in the service area and the adjacent demand response station is as follows:

[0049]

[0050] The expected longitudinal distance d2 between any two adjacent demand response sites in the service area is as follows:

[0051]

[0052] Travelers who choose travel strategy d do not need to walk (K4 = 0). The expected walking time K1, K2, K3 of travelers who choose travel strategies a, b, and c are as follows:

[0053]

[0054] Among them, V wk is the average walking speed of travelers;

[0055] S 3.6 : Calculate the waiting time of travelers;

[0056] The one-way travel distance D of a bus is as follows:

[0057]

[0058] Where δ is the number of passengers served by a bus in a single trip in the variable route bus service, and the single trip time T of the bus service is as follows:

[0059]

[0060] Among them, V rd is the average speed of the service vehicles, T p is the fixed station dwell time, T q Dwell time at a demand response site;

[0061] Since all ride demands are accepted, the relationship between T and δ can be expressed as follows:

[0062]

[0063] Where M is the fleet size of the bus service,

[0064] Combining equations (14) to (20), we can obtain T and δ as shown in equations (21) to (22):

[0065]

[0066] The departure interval h is equal to the quotient of the single-trip time and the fleet size, as shown in formula (23):

[0067]

[0068] Passengers who walk to a fixed bus stop can usually be divided into two groups: passengers who are unfamiliar with the bus schedule and arrive at the station randomly (proportion μ), and passengers who tend to follow the bus schedule to eliminate their waiting time (proportion 1-μ). According to the research of Qiu et al. [4], the waiting time w of passengers who arrive at the station randomly is as follows:

[0069]

[0070] Among them, E(·) is the expectation, Var(·) is the variance;

[0071] In particular, type I travelers who choose travel strategy a walk directly to the destination and can choose to depart at any time without waiting. The expected waiting time A1 and A2 of travelers who choose travel strategies a and b are as follows: (25)-(26);

[0072]

[0073] A2=μw. Formula (26)

[0074] Travelers who choose travel strategies c and d depart from the demand response station. The system will provide them with the estimated arrival time of the vehicle in advance. The waiting time is defined as the difference between the actual arrival time and the estimated arrival time of the vehicle. For any traveler who chooses travel strategies c and d, waiting will only occur when the travelers who successfully make reservations after them are arranged to board the vehicle before them. Referring to the research of Zheng et al.

[10] , the expected waiting time A3 and A4 of travelers who choose travel strategies c and d are as follows:

[0075]

[0076] Among them, j is the number of people who need to wait.

[0077] S 3.7 : Calculate the time in the car; among the travelers who choose travel strategy a, type I travelers walk directly from the starting point to the end point, and they do not need to take a car; type II travelers will rely on variable route buses in the middle of the journey; the essence of the problem is to randomly select two points from N fixed stations and calculate the expected distance; referring to the research of Zheng et al.

[10] , the expected time in the car R1 of travelers who choose travel strategy a is as follows:

[0078]

[0079] The boarding (disembarking) point of travelers who choose travel strategies b and c is located at a fixed station, and the disembarking (disembarking) point is located within the fixed station interval. The essence of the problem is to randomly select two points, one at a fixed station and the other at any position within the fixed station interval, and calculate the expected distance. Referring to the research of Zheng et al.

[10] , the expected in-vehicle time R2 and R3 of travelers who choose travel strategies b and c are as follows:

[0080]

[0081] The boarding and alighting times of travelers who choose d travel strategies are randomly distributed in [0, T]; the essence of the problem is to randomly select two points on the line segment and calculate the expected distance. Referring to the research of Zheng et al.

[10] , the expected in-vehicle time R4 of travelers who choose d travel strategies is as follows:

[0082]

[0083] S 3.8 : To calculate the maximum social welfare without fiscal constraints, first calculate the unit time

[0084] Operator profits;

[0085] The operating income per unit time V is as shown in formula (31);

[0086] V=QF. Formula (31)

[0087] The linear relationship between vehicle hourly cost and capacity has been widely used in the literature, such as Jara-Díaz et al.

[26] of research

[0088] The unit time operating cost C is as shown in formula (32):

[0089] C=(c0+c θ θ)M, formula (32)

[0090] Among them, c0 is the fixed cost coefficient, c θ is the variable cost coefficient, θ is the capacity of the vehicle;

[0091] The operator profit per unit time Z is the sum of the operating income per unit time V and the operating

[0092] The difference between the cost C, as shown in formula (33);

[0093] Z=VC. Formula (33)

[0094] S 3.8 : Calculate the consumer surplus per unit time. The following method is used to calculate the consumer surplus per unit time U, as shown in formula (34);

[0095]

[0096] The calculation principle of this formula is as follows Figure 4 As shown, the integral upper limit in formula (34) is the generalized travel cost corresponding to Q = 0, that is Figure 4 Point A in the equation, the lower limit of the integral is any generalized travel cost value p, that is, Figure 4 Point B in the formula, where the integral element Q(p') corresponds to Figure 4 The horizontal length of the area element in the equation, dp' is the vertical length of the area element, so in The definite integral on the above gives the area of ​​the shaded region, which is the consumer surplus.

[0097] S 3.9 : Calculate the maximum social welfare per unit time; as shown in formula (35)-formula (41);

[0098] S=Z+U. Formula (35)

[0099] Max S(N,h,f), formula (36)

[0100]

[0101] N≥4, formula (39)

[0102] M≥1, formula (40)

[0103] h,f≥0, formula (41)

[0104] Wherein, N and M are integers;

[0105] Formula (37) ensures that the actual travel demand is less than or equal to the potential travel demand. Formula (38) defines the vehicle capacity constraint, ensuring that the vehicle capacity is sufficient to meet the accumulated passenger demand in a departure interval. Formula (39) indicates that there are at least two intermediate fixed stations in the setting scenario, Formula (40) indicates that there are vehicle pick-ups, Formula (41) is the non-negative constraint of the departure interval and the basic fare, Formula (42) is the integer constraint of the number of fixed stations and the fleet size, and Formulas (39)-(42) ensure feasibility;

[0106] S 3.10 : Calculate the fiscal constraint welfare maximization, as shown in formula (42);

[0107] Z+Z f ≥0,

[0108] Z f is the fiscal subsidy per unit time, when the subsidy limit Z f The fiscal constraint (43) may or may not be binding for different values ​​of Z. f When Z is less than the deficit generated when social welfare per unit time is maximized without fiscal constraints, the fiscal constraint (43) becomes binding, forcing the bus company to ensure that it does not lose money in operation under the condition of fiscal subsidies, which will directly affect the pricing of variable route buses. f When the deficit is greater than the deficit that would result from maximizing social welfare per unit time without fiscal constraints, fiscal feasibility is guaranteed and fiscal constraints (43) are redundant.

[0109] S4: Solve the model numerically, specifically using the BONMIN solver to solve the model numerically.

[0110] The BONMIN solver is used to solve the model. The specific steps are as follows:

[0111] S 4.1 : Parameter setting, specifically setting the values ​​of unit walking time value, unit waiting time value, unit in-car time value, travel negative utility coefficient, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly, specifically set to 25$ / h, 15$ / h, 20$ / h, and then set the travel negative utility coefficient to 1.24, -0.1, -0.01, -0.13, -0.07, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly is set to 0.46; as shown in Table 2;

[0112] Table 2 Model parameter values

[0113]

[0114]

[0115] S 4.2 :Based on the parameter values ​​of the model, the optimal values ​​of the decision variables are solved by the BONMIN solver. The number of fixed stations is 9, the departure interval is 0.18h, and the basic fare is 2.26$. In addition, the operator profit per unit time, consumer surplus per unit time, and social welfare per unit time under the optimal conditions are also solved;

[0116] The parameter results of the BONMIN solver are shown in Table 4. The branch and bound method was used to solve the problem, combining the IPOPT and CBC solvers. After 25 iterations and 2 node branches, the optimal solution was successfully obtained. The absolute gap is 0.0460, which is much smaller than the objective function value of 1772.46, and the relative gap is 2.5956e-05, which is very different from the objective function value. Therefore, the convergence is very good and the solution is reliable. The solver takes 4.3106 seconds to solve, which is relatively fast.

[0117] S 4.3 : Analyze potential travel demand and parameter sensitivity;

[0118] S 4.4 : The fare adjustment coefficient is obtained, and it is better to implement differentiated fares than to implement a unified fare.

[0119] Furthermore, the present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, the method described in any one of the above is implemented.

[0120] Compared with the prior art, the present invention has the following beneficial effects:

[0121] 1. The present invention discusses how bus planners can optimize when formulating fare policies to maximize social welfare per unit time. The study considers the operator scenario under vehicle capacity constraints, introduces the concept of elastic demand, and optimizes variable-route buses, taking into account factors such as fares, walking time, in-vehicle time, and waiting time. When establishing a social welfare maximization model, a probabilistic selection function is used to deal with the proportional relationship between travelers' choice of different travel strategies. At the same time, the impact of fiscal constraints on the system is considered. The variable-route bus pricing scheme proposed in the study takes into account elastic demand and travelers' choice preferences, and provides bus planners with a reference for designing fixed stops, departure intervals, and fares in a given service area.

[0122] 2. This paper reveals the influence of some key parameters through numerical research. The results show that there is a threshold of potential travel demand when optimizing the number of fixed stations. The sensitivity of the length-width ratio of the service area is discussed, and the optimal vehicle capacity is found. When studying the effect of fiscal restrictions, it is found that when the fiscal subsidy per unit time is greater than the deficit generated when the social welfare per unit time is maximized without fiscal restrictions, fiscal restrictions are redundant, which is in line with expectations. BRIEF DESCRIPTION OF THE DRAWINGS

[0123] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It is understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0124] Figure 1 is a flow chart of the method of the present invention;

[0125] Figure 2 is a variable route bus service map of the present invention;

[0126] Figure 3 It is a diagram of four travel strategies of travelers under the variable route bus service of the present invention;

[0127] Figure 4 is a graph of two types of travelers of a travel strategy of the present invention;

[0128] Figure 5 is a schematic diagram of consumer surplus calculation of the present invention;

[0129] Figure 6 is the influence diagram of the potential travel demand on the optimal number of fixed stations and the optimal departure interval of the present invention;

[0130] Figure 7 is the impact diagram of the potential travel demand of the present invention on the optimal fare and social welfare per unit time;

[0131] Figure 8 is a diagram showing the influence of the aspect ratio of the rectangular service area on the optimal departure interval and the actual travel demand of the present invention;

[0132] Fig. 9 is the influence diagram of the aspect ratio of the rectangular service area of ​​the present invention on the optimal ticket price and social welfare per unit time;

[0133] Fig.10 is a diagram showing the impact of the fare adjustment coefficient of the present invention on the optimal fare and travel strategy selection;

[0134] Fig.11is a diagram showing the effect of the fare adjustment coefficient of the present invention on social welfare per unit time;

[0135] Fig.12 is the impact diagram of the vehicle capacity of the present invention on the social welfare per unit time;

[0136] Fig.13 is the impact diagram of the fiscal constraint of the present invention on the optimal fare and social welfare per unit time;

[0137] Fig.14 is a graph showing the impact of the financial constraints of the present invention on the financial results;

[0138] Fig.15 is the impact diagram of the financial constraints of the present invention on the actual travel demand and travel strategy selection;

[0139] Fig.16 It is the MTA 646 night bus route map of the present invention;

[0140] Fig.17 It is the degree of optimization of social welfare per unit time by implementing differentiated fares compared to implementing uniform fares. DETAILED DESCRIPTION

[0141] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but is only for selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0142] See also Figure 1-Figure 17 The present invention provides a variable route bus service optimization method based on passenger starting and ending point selection, which is specifically performed in the following steps:

[0143] Please follow the steps below:

[0144] S1: Model the service area, which is a rectangular area with a length of L and a width of W. N fixed stations are set in the service area, and the fixed stations are evenly distributed along the reference route according to the labels 1, 2, ..., N, and the maximum allowable width W / 2 of the reference route and the two sides from which the service vehicle can deviate is set.

[0145] S2: Choose a travel strategy. Travelers on the variable route bus can get on and off at fixed stops or at demand response stops.

[0146] Travel strategies include four types of travel strategies: Figure 2-Figure 4 ;

[0147] The service area is modeled as a rectangle with a length of L and a width of W. There are N fixed stations in the service area, which are equidistantly distributed along the reference route according to the labels 1, 2, ..., N (see Figure 1 ), demand response stations are stops set up according to passenger demand. In the variable route bus service mode, service vehicles travel back and forth between the first and last stations to transport passengers. W / 2 is the maximum allowable width of the two sides and the reference route that the service vehicle can deviate from. The service vehicle adopts the principle of no retrograde during driving.

[0148] When the starting and ending points of travelers who choose travel strategy a are close, they can choose to walk directly from the starting point to the end point without relying on public transportation in the middle. For these two options, the difference in their walking distance in the Y-axis direction is negligible, while the walking distance in the X-axis direction will be very different. Generally, when the walking distance in the X-axis direction is less than the distance between two adjacent fixed stations, travelers will choose to walk directly to the destination. Therefore, we divide travelers who choose travel strategy a into two categories based on whether the walking distance in the X-axis direction is less than the distance between two adjacent fixed stations, and their proportions are respectively The details are as follows:

[0149] I: No demand response service is accepted, that is, the traveler first walks from the starting point to the fixed station closest to the starting point, then takes a bus from the fixed station to the fixed station closest to the destination, and finally walks from the fixed station to the destination;

[0150] II: Only receiving demand response services at the destination, that is, travelers first walk from the starting point to the fixed station closest to the starting point, and then take a bus to the destination;

[0151] III: Only receiving demand response services at the starting point, that is, the traveler first takes a bus from the starting point to the fixed station closest to the destination, and then walks from the fixed station to the destination;

[0152] IV: Receive demand response service at the starting and ending points, that is, travelers take the bus directly from the starting point to the destination.

[0153] Further, the practical variables of the present invention are defined as shown in Table 1;

[0154] Table 1 Symbols

[0155]

[0156]

[0157] S3: Build a model to maximize the unit time and social welfare of a bus system with variable routes;

[0158] The model includes travel demand that is uniformly distributed spatially and over the service period in the service area; that is, demand response stations are uniformly distributed in the service area, and all ride requests are accepted. The scheduled departure time of fixed stations is predetermined based on the expected demand level, the distance between two points is represented by the Manhattan distance, and the linear motion calculation is first horizontally along the X axis and then vertically along the Y axis when the traveler walks and the variable route bus travels. The model includes elastic demand function, probability selection function, travel disutility term, fare, walking time, waiting time, time in the car, social welfare maximization without fiscal constraints, operator profit per unit time, and consumer surplus per unit time.

[0159] The model building is carried out according to the following steps:

[0160] S 3.1 : The actual travel demand Q of the service area is characterized by the elastic demand function, as shown in formula (1);

[0161] Q(p)=q(1-e p p), formula (1)

[0162] Where q is the potential travel demand in the service area, e p is a coefficient to measure the sensitivity of demand to travel impedance, p is the generalized travel cost of the traveler, and the generalized travel cost p of the traveler is as follows:

[0163] p=F+e K K+e A A+e R R, formula (2)

[0164] Among them, F, K, A, and R are the average fare, average walking time, average waiting time, and average time in the car of travelers respectively; K ,e A ,e R They are the value of unit walking time, unit waiting time and unit time in the car respectively.

[0165] The average fare F, average walking time K, average waiting time A and average in-vehicle time R of travelers are as shown in formulas (3) to (6);

[0166]

[0167] Among them, Pi ; i=1,2,3,4 are the probabilities that travelers choose four travel strategies a, b, c, d, F i ; i=1,2,3,4 are the expected fares of travelers who choose four travel strategies a, b, c, d, K i ; i = 1, 2, 3, 4 is the expected walking time for travelers to choose four travel strategies a, b, c, d, A i ; i=1,2,3,4 are the expected waiting times for travelers to choose four travel strategies a, b, c, d, R i ; i=1,2,3,4 is the expected in-vehicle time for travelers who choose four travel strategies a, b, c, d;

[0168] S 3.2 : The probability of travelers choosing four travel strategies, P1, P2, P3, P4, is calculated through the probability selection function, as shown in formula (7);

[0169]

[0170] Among them, X1, X2, X3, and X4 are the travel disutilities of travelers who choose four travel strategies: a, b, c, and d;

[0171] The travel disutility X1, X2, X3, X4 of the four travel strategies chosen by travelers a, b, c, d is as shown in formula (8);

[0172] X i =β0+β F F i +β K K i +β A A i +β R R i ,i=1,2,3,4, formula (8)

[0173] Among them, β0,β F ,β K ,β A ,β R is the travel disutility coefficient;

[0174] S 3.3 : Calculate the negative utility term of travel, as shown in formula (9)-formula (10);

[0175] For type I travelers who choose travel strategy a, their starting point falls evenly within the service area with a probability of 1, while the probability of their destination is approximately Therefore, the probability of a type I traveler choosing travel strategy a is As formula (9);

[0176]

[0177] The sum of the probabilities of type I and type II travelers choosing travel strategy a is 1, so the probability of type II travelers choosing travel strategy a is As formula (10);

[0178]

[0179] S 3.4 : Calculate the fare. Type I travelers who choose travel strategy a walk directly from the starting point to the end point without taking a car, that is, they do not need to buy a ticket. Travelers who choose travel strategies b and c receive one demand response service, while travelers who choose travel strategy d receive two demand response services. The expected fares F1, F2, F3, and F4 of travelers who choose travel strategies a, b, c, and d are as shown in formulas (11) to (13);

[0180]

[0181] F2=F3=(α+1)f, formula (12)

[0182] F4=(2α+1)f, formula (13)

[0183] Among them, f is the basic fare of the variable route bus, α is the fare adjustment coefficient for the traveler to receive an additional demand response service under the variable route bus service;

[0184] S 3.5 : Calculate the traveler's walking time; as shown in formula (14)-formula (17);

[0185] The expected longitudinal distance d1 between the fixed station in the service area and the adjacent demand response station is as follows:

[0186]

[0187] The expected longitudinal distance d2 between any two adjacent demand response sites in the service area is as follows:

[0188]

[0189] Travelers who choose travel strategy d do not need to walk (K4 = 0). The expected walking time K1, K2, K3 of travelers who choose travel strategies a, b, and c are as follows:

[0190]

[0191] Among them, V wk is the average walking speed of travelers;

[0192] S3.6 : Calculate the waiting time of travelers;

[0193] The one-way travel distance D of a bus is as follows:

[0194]

[0195] Where δ is the number of passengers served by a bus in a single trip in the variable route bus service, and the single trip time T of the bus service is as follows:

[0196]

[0197] Among them, V rd is the average speed of the service vehicles, T p is the fixed station dwell time, T q Dwell time at a demand response site;

[0198] Since all ride demands are accepted, the relationship between T and δ can be expressed as follows:

[0199]

[0200] Where M is the fleet size of the bus service,

[0201] Combining equations (14) to (20), we can obtain T and δ as shown in equations (21) to (22):

[0202]

[0203] The departure interval h is equal to the quotient of the single-trip time and the fleet size, as shown in formula (23):

[0204]

[0205] Passengers who walk to a fixed bus stop can usually be divided into two groups: passengers who are unfamiliar with the bus schedule and arrive at the station randomly (proportion μ), and passengers who tend to follow the bus schedule to eliminate their waiting time (proportion 1-μ). According to the research of Qiu et al. [4], the waiting time w of passengers who arrive at the station randomly is as follows:

[0206]

[0207] Among them, E(·) is the expectation, Var(·) is the variance;

[0208] In particular, type I travelers who choose travel strategy a walk directly to the destination and can choose to depart at any time without waiting. The expected waiting time A1 and A2 of travelers who choose travel strategies a and b are as follows: (25)-(26);

[0209]

[0210] A2=μw. Formula (26)

[0211] Travelers who choose travel strategies c and d depart from the demand response station. The system will provide them with the estimated arrival time of the vehicle in advance. The waiting time is defined as the difference between the actual arrival time and the estimated arrival time of the vehicle. For any traveler who chooses travel strategies c and d, waiting will only occur when the travelers who successfully make reservations after them are arranged to board the vehicle before them. Referring to the research of Zheng et al.

[10] , the expected waiting time A3 and A4 of travelers who choose travel strategies c and d are as follows:

[0212]

[0213] Among them, j is the number of people who need to wait.

[0214] S 3.7 : Calculate the time in the car; among the travelers who choose travel strategy a, type I travelers walk directly from the starting point to the end point, and they do not need to take a car; type II travelers will rely on variable route buses in the middle of the journey; the essence of the problem is to randomly select two points from N fixed stations and calculate the expected distance; referring to the research of Zheng et al.

[10] , the expected time in the car R1 of travelers who choose travel strategy a is as follows:

[0215]

[0216] The boarding (disembarking) point of travelers who choose travel strategies b and c is located at a fixed station, and the disembarking (disembarking) point is located within the fixed station interval. The essence of the problem is to randomly select two points, one at a fixed station and the other at any position within the fixed station interval, and calculate the expected distance. Referring to the research of Zheng et al.

[10] , the expected in-vehicle time R2 and R3 of travelers who choose travel strategies b and c are as follows:

[0217]

[0218] The boarding and alighting times of travelers who choose d travel strategies are randomly distributed in [0, T]; the essence of the problem is to randomly select two points on the line segment and calculate the expected distance. Referring to the research of Zheng et al.

[10] , the expected in-vehicle time R4 of travelers who choose d travel strategies is as follows:

[0219]

[0220] S 3.8: To calculate the maximum social welfare without fiscal constraints, first calculate the unit time

[0221] Operator profits;

[0222] The operating income per unit time V is as shown in formula (31);

[0223] V=QF. Formula (31)

[0224] The linear relationship between vehicle hourly cost and capacity has been widely used in the literature, such as Jara-Díaz et al.

[26] of research

[0225] The unit time operating cost C is as shown in formula (32):

[0226] C=(c0+c θ θ)M, formula (32)

[0227] Among them, c0 is the fixed cost coefficient, c θ is the variable cost coefficient, θ is the capacity of the vehicle;

[0228] The operator profit per unit time Z is the sum of the operating income per unit time V and the operating

[0229] The difference between the cost C, as shown in formula (33);

[0230] Z=VC. Formula (33)

[0231] S 3.8 : Calculate the consumer surplus per unit time by the following method

[0232] Consumer surplus U, as shown in formula (34);

[0233]

[0234] The calculation principle of this formula is as follows Figure 5 As shown, the integral upper limit in formula (34) is the generalized travel cost corresponding to Q = 0, that is Figure 4 Point A in the equation, the lower limit of the integral is any generalized travel cost value p, that is, Figure 5 Point B in the formula, where the integral element Q(p') corresponds to Figure 5 The horizontal length of the area element in the equation, dp' is the vertical length of the area element, so in The definite integral on the above gives the area of ​​the shaded region, which is the consumer surplus.

[0235] S 3.9 : Calculate the maximum social welfare per unit time; as shown in formula (35)-formula (41);

[0236] S=Z+U. Formula (35)

[0237] Max S(N,h,f), formula (36)

[0238]

[0239] N≥4, formula (39)

[0240] M≥1, formula (40)

[0241] h,f≥0, formula (41)

[0242] Wherein, N and M are integers;

[0243] Formula (37) ensures that the actual travel demand is less than or equal to the potential travel demand. Formula (38) defines the vehicle capacity constraint, ensuring that the vehicle capacity is sufficient to meet the accumulated passenger demand in a departure interval. Formula (39) indicates that there are at least two intermediate fixed stations in the setting scenario, Formula (40) indicates that there are vehicle pick-ups, Formula (41) is the non-negative constraint of the departure interval and the basic fare, Formula (42) is the integer constraint of the number of fixed stations and the fleet size, and Formulas (39)-(42) ensure feasibility;

[0244] S 3.10 : Calculate the fiscal constraint welfare maximization, as shown in formula (42);

[0245] Z+Z f ≥0,

[0246] Z f is the fiscal subsidy per unit time, when the subsidy limit Z f The fiscal constraint (43) may or may not be binding for different values ​​of Z. f When Z is less than the deficit generated when social welfare per unit time is maximized without fiscal constraints, the fiscal constraint (43) becomes binding, forcing the bus company to ensure that it does not lose money in operation under the condition of fiscal subsidies, which will directly affect the pricing of variable route buses. f When the deficit is greater than the deficit that would result from maximizing social welfare per unit time without fiscal constraints, fiscal feasibility is guaranteed and fiscal constraints (43) are redundant.

[0247] S4: Solve the model numerically, specifically using the BONMIN solver to solve the model numerically.

[0248] The BONMIN solver is used to solve the model. The specific steps are as follows:

[0249] S 4.1: Parameter setting, specifically setting the values ​​of unit walking time value, unit waiting time value, unit in-car time value, travel negative utility coefficient, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly, specifically set to 25$ / h, 15$ / h, 20$ / h, and then set the travel negative utility coefficient to 1.24, -0.1, -0.01, -0.13, -0.07, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly is set to 0.46; as shown in Table 2;

[0250] Table 2 Model parameter values

[0251]

[0252]

[0253] S 4.2 :Based on the parameter values ​​of the model, the optimal values ​​of the decision variables are solved by the BONMIN solver. The number of fixed stations is 9, the departure interval is 0.18h, and the basic fare is 2.26$. In addition, the operator profit per unit time, consumer surplus per unit time, and social welfare per unit time under the optimal conditions are also solved; as shown in Table 3; the operator profit per unit time is negative, and to achieve operator profitability, it is necessary to rely on government financial support and subsidies.

[0254] Table 3 Optimization results based on model parameter values

[0255] variable N h f Z U S result 9 0.18 2.26 -44.52 1816.98 1772.46

[0256] The parameter results of the BONMIN solver are shown in Table 4. The branch and bound method was used to solve the problem, combining the IPOPT and CBC solvers. After 25 iterations and 2 node branches, the optimal solution was successfully obtained. The absolute gap is 0.0460, which is much smaller than the objective function value of 1772.46, and the relative gap is 2.5956e-05, which is very different from the objective function value. Therefore, the convergence is very good and the solution is reliable. The solver takes 4.3106 seconds to solve, which is fast.

[0257] Table 4 Parameter results obtained by BONMIN solver

[0258] parameter result Iterations 25 node 2 Absolute clearance 0.0460 Relative clearance 2.5956e-05 time 4.3106 seconds algorithm BONMIN: Branch and Bound (using IPOPT and CBC) state success

[0259] S 4.3 : Analyze potential travel demand and parameter sensitivity;

[0260] In this embodiment, it is assumed that the potential travel demand increases from 30 people / h to 300 people / h, with an increase of 30 people / h each time. Figure 6The optimal number of fixed stops and optimal departure intervals at different potential travel demand levels are shown. When the potential travel demand is 30 people / h, the optimal number of fixed stops is 8; when the potential travel demand is between 60 and 300 people / h, the optimal number of fixed stops remains unchanged at 9. This is because as travel demand increases, adjusting the departure interval can meet more travel needs without adding more stops. As potential travel demand increases, the optimal departure interval gradually decreases from 0.87h to 0.09h. This is because in order to meet the increase in travel demand, the entire route requires more variable route bus services.

[0261] exist Figure 7 In the figure, the price of four travel strategies for variable route buses increases as the potential travel demand increases. A moderate increase in fares helps to increase operating revenue per unit time. It is worth noting that when the potential travel demand increases from 30 people / h to 60 people / h, the fares of the four travel strategies increase significantly. This is because the increase in the number of optimal fixed stations shortens the expected distance for travelers to reach the fixed stations, thereby reducing travel costs, which in turn leads to a significant increase in actual travel demand. When the potential travel demand increases from 60 people / h to 300 people / h, the fare of the four travel strategies increases more slowly. This is because the number of optimal fixed stations remains unchanged, while the actual travel demand increases slowly. The increase in potential travel demand will inevitably lead to an increase in consumer surplus per unit time, and the linear increase in potential travel demand corresponds to a linear increase in social welfare per unit time.

[0262] Assuming the area of ​​the rectangular service area remains constant (10miles2), the aspect ratio increases from 9.7 to 10.3, increasing by 0.1 each time. Figure 8 The results show that as the length-width ratio of the rectangular service area increases, the departure interval of the variable route bus service gradually increases. This is because the one-way travel time of the service vehicle will become longer in a narrower service area. The larger departure interval significantly increases the time travelers spend in the car, causing more potential travelers to no longer choose variable route buses, but choose other travel modes. In other words, the ratio of actual travel demand to potential travel demand gradually decreases, which is consistent with the Figure 8 The actual travel probability shown is consistent with the decreasing trend.

[0263] Fig. 9It shows that as the length-width ratio of the rectangular service area increases, the ticket prices of the four travel strategies gradually increase. This is because although the operating cost per unit time of the operator remains unchanged, the actual travel probability decreases, which increases the per capita operating cost per unit time. In order to make up for the loss of income and maintain operations, the operator will transfer the cost to the passengers by appropriately raising the ticket price. As the actual travel probability decreases, the consumer surplus per unit time decreases significantly, and the decrease is greater than the increase in the operator's profit per unit time. Therefore, the social welfare per unit time also gradually decreases, which is consistent with Fig. 9 The trend of decreasing social welfare per unit time is consistent. Figure 8 and Fig. 9 The results show that the length-to-width ratio of the rectangular service area has an important impact on the actual travel probability and social welfare per unit time of travelers in the service area.

[0264] S 4.4 : The fare adjustment coefficient is obtained, and it is better to implement differentiated fares than to implement a unified fare.

[0265] Assume that the fare adjustment factor increases from 0 to 1, with each increase of 0.1. Fig.11 It shows that as the fare adjustment coefficient increases, the fares of the three travel strategies gradually decrease, while the fare of the first travel strategy gradually increases. This is because when implementing differentiated fares, low fares will attract price-sensitive passengers and reduce their travel costs, while high fares provide options for passengers who are willing to pay more for better services. By lowering fares for low-fare passengers and raising fares for high-fare passengers, we can better serve different passenger groups, meet their different needs, and improve passengers' travel experience and welfare levels. Since the increase in fares suppresses travelers' travel needs, the probability of travelers choosing the three travel strategies gradually increases, while the probability of choosing the first travel strategy gradually decreases, which is consistent with Fig.10 The increasing and decreasing trends of the probabilities of travelers choosing the four travel strategies are consistent.

[0266] Fig.11 The social welfare per unit time under different fare adjustment coefficients is shown. Fig.11 It can be concluded that when the fare adjustment coefficient is between 0 and 0.3, as the fare adjustment coefficient increases, the social welfare per unit time increases; when the fare adjustment coefficient is 0.3, the social welfare per unit time reaches the maximum value, which is 1773.09$ / h; when the fare adjustment coefficient is between 0.3 and 1, as the fare adjustment coefficient decreases, the social welfare per unit time decreases. When the fare adjustment coefficient is 0, that is, a uniform fare is implemented, the social welfare per unit time is 1772.53$ / h. Comparing the social welfare per unit time when the fare adjustment coefficient is 0 and 0.3 respectively, we can conclude that when there is a specific potential travel demand, using a fare adjustment coefficient within a certain range and implementing differentiated fares is better than implementing a uniform fare.

[0267] In this embodiment, it is assumed that the vehicle capacity increases from 10 seats / vehicle to 30 seats / vehicle, with one seat / vehicle added each time. Fig.12 The social welfare per unit time under different vehicle capacities is shown. Observing the results in the figure, it is found that when the vehicle capacity is 17 seats / vehicle, the social welfare per unit time reaches the maximum value (1772.46$ / h), which indicates that the optimal vehicle capacity is 17 seats / vehicle. It is worth noting that as the vehicle capacity increases, the increase or decrease of social welfare per unit time does not remain constant. When the increase or decrease trend of social welfare per unit time changes significantly, the first threshold of vehicle capacity is 14 seats / vehicle, and the second threshold is 17 seats / vehicle.

[0268] When the vehicle capacity is less than 14 seats / vehicle, the vehicle capacity is small and a larger fleet size is required, which makes the unit time operation cost higher, resulting in lower social welfare per unit time. As the vehicle capacity increases, the fleet size decreases, the vehicles are fully utilized, and the social welfare per unit time increases significantly. When the vehicle capacity is between 14 seats / vehicle and 17 seats / vehicle, the vehicle capacity is large, a smaller fleet size is required, and the social welfare per unit time is higher. However, as the vehicle capacity increases, the reduction in fleet size decreases, resulting in a smaller increase in social welfare per unit time. When the vehicle capacity is greater than 17 seats / vehicle, the vehicle capacity is sufficient to meet the demand, but at this time there are empty seats in the service vehicle, resulting in a sharp decrease in social welfare per unit time.

[0269] Assume that the unit time financial subsidy increases from 0 to 60$ / h, with each increase of 5$ / h. Fig.13 , Fig.14 and Fig.15 As shown in the figure, there are two thresholds for fiscal subsidies per unit time (the first threshold is about 5$ / h, and the second threshold is 44.52$ / h). When the fiscal subsidies per unit time are less than the first threshold, between the first and second thresholds, and greater than the second threshold, the increase and decrease trends of variables such as the optimal fare, financial results, and actual travel demand are inconsistent.

[0270] Fig.13It shows that when the fiscal subsidy per unit time is between 0 and 5$ / h, the fares of the four travel strategies gradually decrease, and the decrease is relatively fast. This is because a smaller fiscal subsidy per unit time can quickly stimulate consumer demand and prompt operators to lower prices to attract more consumers. When the fiscal subsidy per unit time is between 5$ / h and 44.52$ / h, the fares of the four travel strategies gradually decrease, but the decrease is relatively slow. This is because as the fiscal subsidy per unit time increases, the marginal effect decreases, and the additional travel demand growth brought about by the fiscal subsidy per unit time gradually decreases. When the fiscal subsidy per unit time is greater than 44.52$ / h, the fares of the four travel strategies no longer change. This is because even if the fiscal subsidy per unit time is very large, the travel demand is not infinite. As the fare decreases, the travel demand will be saturated. When the demand reaches saturation, there is no need for operators to further reduce prices, because even lower prices will not bring more travel demand.

[0271] Fig.13 It shows that as the fiscal subsidy per unit time increases, the fares of the four travel strategies gradually decrease until they reach the optimal fiscal subsidy per unit time (44.52$ / h). When the available fiscal subsidy per unit time exceeds the optimal fiscal subsidy per unit time, the social welfare per unit time remains unchanged (1772.46$ / h), which means that no additional social benefits will be generated as the available fiscal subsidy per unit time increases. At this point, fiscal constraints become redundant, as described in Section 2.7. It is worth noting further that the optimal fiscal subsidy per unit time is actually the deficit generated by maximizing social welfare per unit time in the absence of fiscal constraints. When the available fiscal subsidy per unit time is greater than this deficit and continues to grow, the fares of the four travel strategies no longer decrease, but remain unchanged and greater than zero. This shows that even when the possible fiscal subsidy per unit time is infinite, the fares of the four travel strategies should not drop to zero.

[0272] Fig.14The financial results of variable route bus services with different levels of unit time subsidies are shown. As the unit time subsidy increases, the unit time operating cost does not change (468$ / h), because the optimal fleet size of the variable route bus service does not change. As the unit time subsidy increases, the unit time operating revenue gradually decreases until it reaches 423.48$ / h and then stabilizes, because the fares of the four travel strategies gradually decrease and eventually stop decreasing. Before the unit time subsidy reaches the optimal unit time subsidy, the unit time operator profit decreases as the unit time subsidy increases. When the unit time subsidy exceeds the optimal unit time subsidy, the unit time operator profit remains unchanged, because both the unit time operating revenue and the unit time operating cost remain unchanged at this time. The financial results of the variable route bus service also include "profit + subsidy". When the unit time subsidy exceeds the optimal unit time subsidy, the sum of profit and subsidy is positive, which means that there is some unused budget available. Before the unit time subsidy reaches the optimal unit time subsidy, the higher the unit time subsidy, the more actual trips, and the unit time consumer surplus increases as the unit time subsidy increases.

[0273] Fig.15 It shows that when the fiscal subsidy per unit time is 0, 33.55% of travelers choose the 100 travel strategies, 18.36% of travelers choose the 200 travel strategies, and the probability of travelers choosing the two travel strategies is very close, about 24%. With the increase of the fiscal subsidy per unit time, the proportion of actual travel demand to potential travel demand gradually increases, because the higher the fiscal subsidy per unit time, the lower the fare, and more potential travelers will choose variable route buses for travel. After the fiscal subsidy per unit time exceeds the optimal fiscal subsidy per unit time, the proportion of actual travel demand to potential travel demand remains unchanged, because the fare of the four travel strategies remains unchanged at this time. Before the fiscal subsidy per unit time reaches the optimal fiscal subsidy per unit time, with the increase of the fiscal subsidy per unit time, in the actual travel demand, the probability of travelers choosing the 100 travel strategy relative to the three travel strategies continues to increase, because the fare of the 100 travel strategy is reduced to a greater extent than the three travel strategies. Similarly, the probability of travelers choosing the 100 travel strategies relative to the 100 travel strategies also continues to increase, because the fare of the 100 travel strategies is reduced to a greater extent than the 100 travel strategies.

[0274] In this embodiment, Fig.16Taking MTA Route 646 Night Bus as an example, it is a typical variable route bus system and is widely used as a case study. Therefore, in this embodiment, MTA Route 646 Night Bus is used as the basis for case study. When the demand is large during the day, the line adopts the operation mode of conventional bus; at night, it implements the operation mode of variable route bus. MTA Route 646 Night Bus includes two terminal stations and one intermediate fixed station.

[0275] The rectangular service area for bus operation is 10 miles long and 1 mile wide. The potential travel demand is from 30 people / h to 300 people / h, increasing by 30 people / h each time. The fixed cost coefficient is 120, the variable cost coefficient is 1.2, and other parameter settings are shown in Section 3.2. In order to study the advantages and disadvantages of uniform fares and differentiated fares, we enter the model with a fare adjustment coefficient of 0 to calculate the social welfare per unit time of implementing uniform fares; we enter the model with fare adjustment coefficients of 0.2, 0.4, 0.6, 0.8 and 1 to calculate the social welfare per unit time of implementing differentiated fares. We define the optimization degree of social welfare per unit time of implementing differentiated fares compared to implementing uniform fares to compare the advantages and disadvantages of implementing the two fares. The formula is as follows:

[0276]

[0277] Among them, Op is the optimization degree of social welfare per unit time by implementing differentiated fares compared with implementing uniform fares, and S d S is the social welfare per unit time of implementing differentiated fares. u To implement social welfare per unit time with uniform fares.

[0278] The optimization degree of social welfare per unit time by implementing differentiated fares compared with implementing uniform fares is as follows: Fig.16. As can be seen from the figure, there is a critical value for potential travel demand (between 90 people / h and 120 people / h). When the potential travel demand is equal to the critical value, the optimization degree is 0, that is, the social welfare per unit time of implementing differentiated fares is the same as that of implementing uniform fares. When the potential travel demand is low, that is, when the potential travel demand is from 0 to the critical value, the social welfare per unit time of implementing uniform fares is higher than that of implementing differentiated fares. This is because the lower travel demand leads to less demand response, and the uniform fare can better meet the fixed station needs of most travelers. When the potential travel demand is high, that is, when the potential travel demand is from the critical value to 300 people / h, the social welfare per unit time of implementing uniform fares is lower than that of implementing differentiated fares. This is because the higher travel demand means a large number of demand responses. Differentiated fares can reasonably distinguish travelers who receive different times of demand response services, which can greatly reduce the cross-subsidy between travelers with different travel strategies and significantly improve the travel selection efficiency of travelers. When the potential travel demand exceeds 120 people / h, the optimization degree of social welfare per unit time when the fare adjustment coefficient is 1 is higher than that when the fare adjustment coefficient is 0.2, 0.4, 0.6 and 0.8, which is about 1%. This helps relevant departments to formulate fare policies based on actual conditions.

[0279] In this embodiment, the present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a main controller, the method described in any one of the above is implemented.

[0280] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention has various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for optimizing bus services with variable routes based on passenger starting and ending point selection, characterized in that: Please follow the steps below: S1: Model the service area, where the service area is a rectangular area with a length of L and a width of W; S2: Choose a travel strategy. Travelers on the variable route bus can get on and off at fixed stops or at demand response stops. S3: Build a model to maximize the unit time and social welfare of a bus system with variable routes; S4: Solve the model numerically, specifically using the BONMIN solver to solve the model numerically.

2. The variable route bus service optimization method based on passenger starting and ending point selection according to claim 1 is characterized in that: In step S1, N fixed stations are set in the service area, and the fixed stations are evenly distributed along the reference route according to the labels 1, 2, ..., N, and the maximum allowable width W / 2 of the reference route and the two sides from which the service vehicle can deviate is set.

3. The variable route bus service optimization method based on passenger starting and ending point selection according to claim 2 is characterized in that: In step S2, the travel strategy includes four travel strategies: I: No demand response service is accepted, that is, the traveler first walks from the starting point to the fixed station closest to the starting point, then takes a bus from the fixed station to the fixed station closest to the destination, and finally walks from the fixed station to the destination; II: Only receiving demand response services at the destination, that is, travelers first walk from the starting point to the fixed station closest to the starting point, and then take a bus to the destination; III: Only receiving demand response services at the starting point, that is, the traveler first takes a bus from the starting point to the fixed station closest to the destination, and then walks from the fixed station to the destination; IV: Receive demand response service at the starting and ending points, that is, travelers take the bus directly from the starting point to the destination.

4. The variable route bus service optimization method based on passenger starting and ending point selection according to claim 1 is characterized in that: In step S3, the model includes travel demand uniformly distributed in space in the service area and within the service period; the scheduled departure time of fixed stations is predetermined according to the expected demand level, the distance between two points is represented by the Manhattan distance, and the linear motion calculation is first carried out in the horizontal direction of the X axis and then in the vertical direction of the Y axis when the traveler walks and the variable route bus travels.

5. The variable route bus service optimization method based on passenger starting and ending point selection according to claim 4 is characterized in that: The model building is carried out according to the following steps: S 3.1 : The actual travel demand Q of the service area is characterized by the elastic demand function, as shown in formula (1); Q(p)=q(1-e p p), formula (1) Where q is the potential travel demand in the service area, e p is a coefficient to measure the sensitivity of demand to travel impedance, p is the generalized travel cost of the traveler, and the generalized travel cost p of the traveler is as follows: p = F + e K K + e A A + e R R, Equation (2) Among them, F, K, A, and R are the average fare, average walking time, average waiting time, and average time in the car of travelers respectively; K ,e A ,e R They are the value of unit walking time, the value of unit waiting time and the value of unit time in the car; The average fare F, average walking time K, average waiting time A and average in-vehicle time R of travelers are as shown in formulas (3) to (6); Among them, P i ; i=1,2,3,4 are the probabilities that travelers choose four travel strategies a, b, c, d, F i ; i=1,2,3,4 are the expected fares of travelers who choose four travel strategies a, b, c, d, K i ; i = 1, 2, 3, 4 is the expected walking time for travelers to choose four travel strategies a, b, c, d, A i ; i=1,2,3,4 are the expected waiting times for travelers to choose four travel strategies a, b, c, d, R i ; i=1,2,3,4 is the expected in-vehicle time for travelers who choose four travel strategies a, b, c, d; S 3.2 : The probability of travelers choosing four travel strategies, P1, P2, P3, P4, is calculated through the probability selection function, as shown in formula (7); Among them, X1, X2, X3, and X4 are the travel disutilities of travelers choosing four travel strategies: a, b, c, and d; The travel disutility X1, X2, X3, X4 of the four travel strategies chosen by travelers a, b, c, d is as shown in formula (8); X i = β0 + β F F i + β K K i + β A A i + β R R i , i = 1, 2, 3, 4, Equation (8) Among them, β0,β F ,β K ,β A ,β R is the travel disutility coefficient; S 3.3 : Calculate the negative utility term of travel, as shown in formula (9)-formula (10); For type I travelers who choose travel strategy a, their starting point is evenly located in the service area with a probability of 1, while the probability of their destination is approximately Therefore, the probability of a type I traveler choosing travel strategy a is As formula (9); The sum of the probabilities of type I and type II travelers choosing travel strategy a is 1, so the probability of type II travelers choosing travel strategy a is As formula (10); S 3.4 : Calculate the fare. Type I travelers who choose travel strategy a walk directly from the starting point to the end point without taking a car, that is, they do not need to buy a ticket. Travelers who choose travel strategies b and c receive one demand response service, while travelers who choose travel strategy d receive two demand response services. The expected fares F1, F2, F3, and F4 of travelers who choose travel strategies a, b, c, and d are as shown in formulas (11) to (13); F2=F3=(α+1)f, formula (12) F4=(2α+1)f, formula (13) Among them, f is the basic fare of the variable route bus, α is the fare adjustment coefficient for the traveler to receive an additional demand response service under the variable route bus service; S 3.5 : Calculate the traveler's walking time; as shown in formula (14)-formula (17); The expected longitudinal distance d1 between the fixed station in the service area and the adjacent demand response station is as follows: The expected longitudinal distance d2 between any two adjacent demand response stations in the service area is as follows: Travelers who choose travel strategy d do not need to walk (K4 = 0). The expected walking time K1, K2, and K3 of travelers who choose travel strategies a, b, and c are as follows: Among them, V wk is the average walking speed of travelers; S 3.6 : Calculate the waiting time of travelers; The one-way travel distance D of a bus is as follows: where δ is the number of passengers served by a bus in a single trip in a variable route bus service, The one-way travel time T of bus service is as shown in formula (19): Among them, V rd is the average speed of the service vehicles, T p is the fixed station dwell time, T q Dwell time at a demand response site; Since all ride demands are accepted, the relationship between T and δ can be expressed as follows: Where M is the fleet size of the bus service, Combining equations (14) to (20), we can obtain T and δ as shown in equations (21) to (22): The departure interval h is equal to the quotient of the single-trip time and the fleet size, as shown in formula (23): Passengers who walk to a fixed bus stop can usually be divided into two groups: passengers who are not familiar with the bus schedule and arrive at the station randomly (proportion μ), and passengers who tend to follow the bus schedule to eliminate their waiting time (proportion 1-μ). The waiting time w of passengers who arrive at the station randomly is as shown in formula (24): Among them, E(·) is the expectation, Var(·) is the variance; In particular, type I travelers who choose travel strategy a walk directly to the destination. The expected waiting time A1 and A2 of travelers who can choose to depart at any time without waiting and choose travel strategies a and b are as follows: (25)-(26); A2=μw. Formula (26) Travelers who choose travel strategies c and d start from the demand response station. The estimated arrival time of the vehicle will be fed back to them in advance. The waiting time is defined as the difference between the actual arrival time and the estimated arrival time of the vehicle. Travelers with a travel strategy are arranged only when the travelers who successfully make reservations after them are Waiting will only occur when getting on the bus before that. The expected waiting time A3 and A4 of travelers who choose the two travel strategies c and d are as follows: Among them, j is the number of people who need to wait; S 3.7 : Calculate the time in the car; among the travelers who choose a travel strategy, type I Walkers walk directly from the starting point to the end point, they do not need to take a car; Type II travelers will Rely on variable bus routes in the middle; the essence of the problem is to go from N fixed stations Randomly select two points and calculate the expected distance; the expected in-vehicle time R1 of the traveler who chooses a travel strategy is as follows: The boarding (disembarking) points of travelers who choose travel strategies b and c are located at fixed stations, and the disembarking (disembarking) points are located within the fixed station interval. The expected in-vehicle time R2 and R3 of travelers who choose travel strategies b and c are as follows: The boarding time and alighting time of travelers who choose d travel strategies are randomly distributed in [0, T]. The expected in-car time R4 of travelers who choose d travel strategies is as follows: S 3.8 : To calculate the maximum social welfare without fiscal constraints, first calculate the operator profit per unit time; The operating income per unit time V is as shown in formula (31); V=QF. Formula (31) The unit time operating cost C is as shown in formula (32): C = (c0 + c θ θ)M, Equation (32) Among them, c0 is the fixed cost coefficient, c θ is the variable cost coefficient, θ is the capacity of the vehicle; The operator profit per unit time Z is the sum of the operating income per unit time V and the operating The difference between the cost C, as shown in formula (33); Z=VC. Formula (33) S 3.8 : Calculate the consumer surplus per unit time by the following method Consumer surplus U, as shown in formula (34); The integral upper limit in formula (34) is the generalized travel cost corresponding to Q = 0, The lower limit of the integral is any generalized travel cost value p, and dp' is the longitudinal length of the area element. The definite integral on the above gives the area of ​​the shaded region, which is the consumer surplus. S 3.9 : Calculate the maximum social welfare per unit time; as shown in formula (35)-formula (41); S=Z+U. Formula (35) Max S(N,h,f), formula (36) N≥4, formula (39) M≥1, formula (40) h,f≥0, formula (41) Wherein, N and M are integers; S 3.10 : Calculate the fiscal constraint welfare maximization, as shown in formula (42); Z+Z f ≥0, Z f It is a financial subsidy per unit time.

6. The variable route bus service optimization method based on passenger starting and ending point selection according to claim 1 is characterized in that: In step S4, the BONMIN solver is used to solve the model, which is specifically performed in the following steps: S 4.1 :Set parameters, specifically set the values ​​of unit walking time value, unit waiting time value, unit in-car time value, travel negative utility coefficient, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly, specifically set to 25$ / h, 15$ / h, 20$ / h, and then set the travel negative utility coefficient to 1.24, -0.1, -0.01, -0.13, -0.07, and the proportion of passengers who are unfamiliar with the bus schedule and arrive at the station randomly is set to 0.46; S 4.2 :Based on the parameter values ​​of the model, the optimal values ​​of the decision variables are obtained through the BONMIN solver; S 4.3 : Analyze potential travel demand and parameter sensitivity; S 4.4 : The fare adjustment coefficient is obtained, and it is better to implement differentiated fares than to implement a unified fare.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the main controller, the method according to any one of claims 1 to 6 is implemented.

Citation Information

Cited By

  • Dynamic cruise bus dynamic scheduling and path planning method, electronic equipment and storage medium

    CN120220452A